BO - Bar's direction Signal - BacktestingBO - Bar's direction Signal - Backtesting Options:
A. Factors Calculate probability of x bars same direction
1. Periods Counting: Data to count From day/month/year To day/month/year
2. Trading Time: only cases occurred in trading time were counted.
B. Timezone
1. Trading time depend on Time zone and specified chart.
2. Enable Highlight Trading Time to check your period time is correct
C. Date Backtesting
* Only cases occurred in Date Backtesting were reported.
D. Setup Options & Rule
1. Reversal after 2 bars same direction
* Probability of 3 bars same direction < 50
* 2 bars same direction is start of series
2. Reversal after 3 bars same direction
* Probability of 4 bars same direction < 50
* 3 bars same direction is start of series
3. Reversal after 4 bars same direction
* Probability of 4 bars same direction < 50
* 3 bars same direction is start of series
4. Reversal after 5 bars same direction
* Probability of 5 bars same direction < 50
* 4 bars same direction is start of series
5. Reversal after 6 bars same direction
* Probability of 6 bars same direction < 50
* 5 bars same direction is start of series
Pesquisar nos scripts por "binary"
Chatterjee CorrelationThis is my first attempt on implementing a statistical method. This problem was given to me by @lejmer (who also helped me later on building more efficient code to achieve this) when we were debating on the need for higher resource allocation to run scripts so it can run longer and faster. The major problem faced by those who want to implement statistics based methods is that they run out of processing time or need to limit the data samples. My point was that such things need be implemented with an algorithm which suits pine instead of trying to port a python code directly. And yes, I am able to demonstrate that by using this implementation of Chatterjee Correlation.
🎲 What is Chatterjee Correlation?
The Chatterjee rank Correlation Coefficient (CCC) is a method developed by Sourav Chatterjee which can be used to study non linear correlation between two series.
Full documentation on the method can be found here:
arxiv.org
In short, the formula which we are implementing here is:
Algorithm can be simplified as follows:
1. Get the ranks of X
2. Get the ranks of Y
3. Sort ranks of Y in the order of X (Lets call this SortedYIndices)
4. Calculate the sum of adjacent Y ranks in SortedYIndices (Lets call it as SumOfAdjacentSortedIndices)
5. And finally the correlation coefficient can be calculated by using simple formula
CCC = 1 - (3*SumOfAdjacentSortedIndices)/(n^2 - 1)
🎲 Looks simple? What is the catch?
Mistake many people do here is that they think in Python/Java/C etc while coding in Pine. This makes code less efficient if it involves arrays and loops. And the simple code may look something like this.
var xArray = array.new()
var yArray = array.new()
array.push(xArray, x)
array.push(yArray, y)
sortX = array.sort_indices(xArray)
sortY = array.sort_indices(yArray)
SumOfAdjacentSortedIndices = 0.0
index = array.get(xSortIndices, 0)
for i=1 to n > 1? n -1 : na
indexNext = array.get(sortX, i)
SumOfAdjacentSortedIndices += math.abs(array.get(sortY, indexNext)-array.get(sortY, index))
index := indexNext
correlation := 1 - 3*SumOfAdjacentSortedIndices/(math.pow(n,2)-1)
But, problem here is the number of loops run. Remember pine executes the code on every bar. There are loops run in array.sort_indices and another loop we are running to calculate SumOfAdjacentSortedIndices. Due to this, chances of program throwing runtime errors due to script running for too long are pretty high. This limits greatly the number of samples against which we can run the study. The options to overcome are
Limit the sample size and calculate only between certain bars - this is not ideal as smaller sets are more likely to yield false or inconsistent results.
Start thinking in pine instead of python and code in such a way that it is optimised for pine. - This is exactly what we have done in the published code.
🎲 How to think in Pine?
In order to think in pine, you should try to eliminate the loops as much as possible. Specially on the data which is continuously growing.
My first thought was that sorting takes lots of time and need to find a better way to sort series - specially when it is a growing data set. Hence, I came up with this library which implements Binary Insertion Sort.
Replacing array.sort_indices with binary insertion sort will greatly reduce the number of loops run on each bar. In binary insertion sort, the array will remain sorted and any item we add, it will keep adding it in the existing sort order so that there is no need to run separate sort. This allows us to work with bigger data sets and can utilise full 20,000 bars for calculation instead of few 100s.
However, last loop where we calculate SumOfAdjacentSortedIndices is not replaceable easily. Hence, we only limit these iterations to certain bars (Even though we use complete sample size). Plots are made for only those bars where the results need to be printed.
🎲 Implementation
Current implementation is limited to few combinations of x and fixed y. But, will be converting this into library soon - which means, programmers can plug any x and y and get the correlation.
Our X here can be
Average volume
ATR
And our Y is distance of price from moving average - which identifies trend.
Thus, the indicator here helps to understand the correlation coefficient between volume and trend OR volatility and trend for given ticker and timeframe. Value closer to 1 means highly correlated and value closer to 0 means least correlated. Please note that this method will not tell how these values are correlated. That is, we will not be able to know if higher volume leads to higher trend or lower trend. But, we can say whether volume impacts trend or not.
Please note that values can differ by great extent for different timeframes. For example, if you look at 1D timeframe, you may get higher value of correlation coefficient whereas lower value for 1m timeframe. This means, volume to trend correlation is higher in 1D timeframe and lower in lower timeframes.
InsertionSortLibrary "InsertionSort"
Library of sorting algorithm for binary insertion sort and related methods
method binary_insertion_sort(sortedArray, item, order)
binary insertion sort - inserts item into sorted array while maintaining sort order
Namespace types: array
Parameters:
sortedArray (array) : array which is assumed to be sorted in the requested order
item (float) : float|int item which needs to be inserted into sorted array
order (series ORDER) : Sort order - positive number means ascending order whereas negative number represents descending order
Returns: int index at which the item is inserted into sorted array
method binary_insertion_sort(sortedArray, item, order)
Namespace types: array
Parameters:
sortedArray (array)
item (int)
order (series ORDER)
Trading IQ - ICT LibraryLibrary "ICTlibrary"
Used to calculate various ICT related price levels and strategies. An ongoing project.
Hello Coders!
This library is meant for sourcing ICT related concepts. While some functions might generate more output than you require, you can specify "Lite Mode" as "true" in applicable functions to slim down necessary inputs.
isLastBar(userTF)
Identifies the last bar on the chart before a timeframe change
Parameters:
userTF (simple int) : the timeframe you wish to calculate the last bar for, must be converted to integer using 'timeframe.in_seconds()'
Returns: bool true if bar on chart is last bar of higher TF, dalse if bar on chart is not last bar of higher TF
necessaryData(atrTF)
returns necessaryData UDT for historical data access
Parameters:
atrTF (float) : user-selected timeframe ATR value.
Returns: logZ. log return Z score, used for calculating order blocks.
method gradBoxes(gradientBoxes, idColor, timeStart, bottom, top, rightCoordinate)
creates neon like effect for box drawings
Namespace types: array
Parameters:
gradientBoxes (array) : an array.new() to store the gradient boxes
idColor (color)
timeStart (int) : left point of box
bottom (float) : bottom of box price point
top (float) : top of box price point
rightCoordinate (int) : right point of box
Returns: void
checkIfTraded(tradeName)
checks if recent trade is of specific name
Parameters:
tradeName (string)
Returns: bool true if recent trade id matches target name, false otherwise
checkIfClosed(tradeName)
checks if recent closed trade is of specific name
Parameters:
tradeName (string)
Returns: bool true if recent closed trade id matches target name, false otherwise
IQZZ(atrMult, finalTF)
custom ZZ to quickly determine market direction.
Parameters:
atrMult (float) : an atr multiplier used to determine the required price move for a ZZ direction change
finalTF (string) : the timeframe used for the atr calcuation
Returns: dir market direction. Up => 1, down => -1
method drawBos(id, startPoint, getKeyPointTime, getKeyPointPrice, col, showBOS, isUp)
calculates and draws Break Of Structure
Namespace types: array
Parameters:
id (array)
startPoint (chart.point)
getKeyPointTime (int) : the actual time of startPoint, simplystartPoint.time
getKeyPointPrice (float) : the actual time of startPoint, simplystartPoint.price
col (color) : color of the BoS line / label
showBOS (bool) : whether to show label/line. This function still calculates internally for other ICT related concepts even if not drawn.
isUp (bool) : whether BoS happened during price increase or price decrease.
Returns: void
method drawMSS(id, startPoint, getKeyPointTime, getKeyPointPrice, col, showMSS, isUp, upRejections, dnRejections, highArr, lowArr, timeArr, closeArr, openArr, atrTFarr, upRejectionsPrices, dnRejectionsPrices)
calculates and draws Market Structure Shift. This data is also used to calculate Rejection Blocks.
Namespace types: array
Parameters:
id (array)
startPoint (chart.point)
getKeyPointTime (int) : the actual time of startPoint, simplystartPoint.time
getKeyPointPrice (float) : the actual time of startPoint, simplystartPoint.price
col (color) : color of the MSS line / label
showMSS (bool) : whether to show label/line. This function still calculates internally for other ICT related concepts even if not drawn.
isUp (bool) : whether MSS happened during price increase or price decrease.
upRejections (array)
dnRejections (array)
highArr (array) : array containing historical highs, should be taken from the UDT "necessaryData" defined above
lowArr (array) : array containing historical lows, should be taken from the UDT "necessaryData" defined above
timeArr (array) : array containing historical times, should be taken from the UDT "necessaryData" defined above
closeArr (array) : array containing historical closes, should be taken from the UDT "necessaryData" defined above
openArr (array) : array containing historical opens, should be taken from the UDT "necessaryData" defined above
atrTFarr (array) : array containing historical atr values (of user-selected TF), should be taken from the UDT "necessaryData" defined above
upRejectionsPrices (array) : array containing up rejections prices. Is sorted and used to determine selective looping for invalidations.
dnRejectionsPrices (array) : array containing down rejections prices. Is sorted and used to determine selective looping for invalidations.
Returns: void
method getTime(id, compare, timeArr)
gets time of inputted price (compare) in an array of data
this is useful when the user-selected timeframe for ICT concepts is greater than the chart's timeframe
Namespace types: array
Parameters:
id (array) : the array of data to search through, to find which index has the same value as "compare"
compare (float) : the target data point to find in the array
timeArr (array) : array of historical times
Returns: the time that the data point in the array was recorded
method OB(id, highArr, signArr, lowArr, timeArr, sign)
store bullish orderblock data
Namespace types: array
Parameters:
id (array)
highArr (array) : array of historical highs
signArr (array) : array of historical price direction "math.sign(close - open)"
lowArr (array) : array of historical lows
timeArr (array) : array of historical times
sign (int) : orderblock direction, -1 => bullish, 1 => bearish
Returns: void
OTEstrat(OTEstart, future, closeArr, highArr, lowArr, timeArr, longOTEPT, longOTESL, longOTElevel, shortOTEPT, shortOTESL, shortOTElevel, structureDirection, oteLongs, atrTF, oteShorts)
executes the OTE strategy
Parameters:
OTEstart (chart.point)
future (int) : future time point for drawings
closeArr (array) : array of historical closes
highArr (array) : array of historical highs
lowArr (array) : array of historical lows
timeArr (array) : array of historical times
longOTEPT (string) : user-selected long OTE profit target, please create an input.string() for this using the example below
longOTESL (int) : user-selected long OTE stop loss, please create an input.string() for this using the example below
longOTElevel (float) : long entry price of selected retracement ratio for OTE
shortOTEPT (string) : user-selected short OTE profit target, please create an input.string() for this using the example below
shortOTESL (int) : user-selected short OTE stop loss, please create an input.string() for this using the example below
shortOTElevel (float) : short entry price of selected retracement ratio for OTE
structureDirection (string) : current market structure direction, this should be "Up" or "Down". This is used to cancel pending orders if market structure changes
oteLongs (bool) : input.bool() for whether OTE longs can be executed
atrTF (float) : atr of the user-seleceted TF
oteShorts (bool) : input.bool() for whether OTE shorts can be executed
@exampleInputs
oteLongs = input.bool(defval = false, title = "OTE Longs", group = "Optimal Trade Entry")
longOTElevel = input.float(defval = 0.79, title = "Long Entry Retracement Level", options = , group = "Optimal Trade Entry")
longOTEPT = input.string(defval = "-0.5", title = "Long TP", options = , group = "Optimal Trade Entry")
longOTESL = input.int(defval = 0, title = "How Many Ticks Below Swing Low For Stop Loss", group = "Optimal Trade Entry")
oteShorts = input.bool(defval = false, title = "OTE Shorts", group = "Optimal Trade Entry")
shortOTElevel = input.float(defval = 0.79, title = "Short Entry Retracement Level", options = , group = "Optimal Trade Entry")
shortOTEPT = input.string(defval = "-0.5", title = "Short TP", options = , group = "Optimal Trade Entry")
shortOTESL = input.int(defval = 0, title = "How Many Ticks Above Swing Low For Stop Loss", group = "Optimal Trade Entry")
Returns: void (0)
displacement(logZ, atrTFreg, highArr, timeArr, lowArr, upDispShow, dnDispShow, masterCoords, labelLevels, dispUpcol, rightCoordinate, dispDncol, noBorders)
calculates and draws dispacements
Parameters:
logZ (float) : log return of current price, used to determine a "significant price move" for a displacement
atrTFreg (float) : atr of user-seleceted timeframe
highArr (array) : array of historical highs
timeArr (array) : array of historical times
lowArr (array) : array of historical lows
upDispShow (int) : amount of historical upside displacements to show
dnDispShow (int) : amount of historical downside displacements to show
masterCoords (map) : a map to push the most recent displacement prices into, useful for having key levels in one data structure
labelLevels (string) : used to determine label placement for the displacement, can be inside box, outside box, or none, example below
dispUpcol (color) : upside displacement color
rightCoordinate (int) : future time for displacement drawing, best is "last_bar_time"
dispDncol (color) : downside displacement color
noBorders (bool) : input.bool() to remove box borders, example below
@exampleInputs
labelLevels = input.string(defval = "Inside" , title = "Box Label Placement", options = )
noBorders = input.bool(defval = false, title = "No Borders On Levels")
Returns: void
method getStrongLow(id, startIndex, timeArr, lowArr, strongLowPoints)
unshift strong low data to array id
Namespace types: array
Parameters:
id (array)
startIndex (int) : the starting index for the timeArr array of the UDT "necessaryData".
this point should start from at least 1 pivot prior to find the low before an upside BoS
timeArr (array) : array of historical times
lowArr (array) : array of historical lows
strongLowPoints (array) : array of strong low prices. Used to retrieve highest strong low price and see if need for
removal of invalidated strong lows
Returns: void
method getStrongHigh(id, startIndex, timeArr, highArr, strongHighPoints)
unshift strong high data to array id
Namespace types: array
Parameters:
id (array)
startIndex (int) : the starting index for the timeArr array of the UDT "necessaryData".
this point should start from at least 1 pivot prior to find the high before a downside BoS
timeArr (array) : array of historical times
highArr (array) : array of historical highs
strongHighPoints (array)
Returns: void
equalLevels(highArr, lowArr, timeArr, rightCoordinate, equalHighsCol, equalLowsCol, liteMode)
used to calculate recent equal highs or equal lows
Parameters:
highArr (array) : array of historical highs
lowArr (array) : array of historical lows
timeArr (array) : array of historical times
rightCoordinate (int) : a future time (right for boxes, x2 for lines)
equalHighsCol (color) : user-selected color for equal highs drawings
equalLowsCol (color) : user-selected color for equal lows drawings
liteMode (bool) : optional for a lite mode version of an ICT strategy. For more control over drawings leave as "True", "False" will apply neon effects
Returns: void
quickTime(timeString)
used to quickly determine if a user-inputted time range is currently active in NYT time
Parameters:
timeString (string) : a time range
Returns: true if session is active, false if session is inactive
macros(showMacros, noBorders)
used to calculate and draw session macros
Parameters:
showMacros (bool) : an input.bool() or simple bool to determine whether to activate the function
noBorders (bool) : an input.bool() to determine whether the box anchored to the session should have borders
Returns: void
po3(tf, left, right, show)
use to calculate HTF po3 candle
@tip only call this function on "barstate.islast"
Parameters:
tf (simple string)
left (int) : the left point of the candle, calculated as bar_index + left,
right (int) : :the right point of the candle, calculated as bar_index + right,
show (bool) : input.bool() whether to show the po3 candle or not
Returns: void
silverBullet(silverBulletStratLong, silverBulletStratShort, future, userTF, H, L, H2, L2, noBorders, silverBulletLongTP, historicalPoints, historicalData, silverBulletLongSL, silverBulletShortTP, silverBulletShortSL)
used to execute the Silver Bullet Strategy
Parameters:
silverBulletStratLong (simple bool)
silverBulletStratShort (simple bool)
future (int) : a future time, used for drawings, example "last_bar_time"
userTF (simple int)
H (float) : the high price of the user-selected TF
L (float) : the low price of the user-selected TF
H2 (float) : the high price of the user-selected TF
L2 (float) : the low price of the user-selected TF
noBorders (bool) : an input.bool() used to remove the borders from box drawings
silverBulletLongTP (series silverBulletLevels)
historicalPoints (array)
historicalData (necessaryData)
silverBulletLongSL (series silverBulletLevels)
silverBulletShortTP (series silverBulletLevels)
silverBulletShortSL (series silverBulletLevels)
Returns: void
method invalidFVGcheck(FVGarr, upFVGpricesSorted, dnFVGpricesSorted)
check if existing FVGs are still valid
Namespace types: array
Parameters:
FVGarr (array)
upFVGpricesSorted (array) : an array of bullish FVG prices, used to selective search through FVG array to remove invalidated levels
dnFVGpricesSorted (array) : an array of bearish FVG prices, used to selective search through FVG array to remove invalidated levels
Returns: void (0)
method drawFVG(counter, FVGshow, FVGname, FVGcol, data, masterCoords, labelLevels, borderTransp, liteMode, rightCoordinate)
draws FVGs on last bar
Namespace types: map
Parameters:
counter (map) : a counter, as map, keeping count of the number of FVGs drawn, makes sure that there aren't more FVGs drawn
than int FVGshow
FVGshow (int) : the number of FVGs to show. There should be a bullish FVG show and bearish FVG show. This function "drawFVG" is used separately
for bearish FVG and bullish FVG.
FVGname (string) : the name of the FVG, "FVG Up" or "FVG Down"
FVGcol (color) : desired FVG color
data (FVG)
masterCoords (map) : a map containing the names and price points of key levels. Used to define price ranges.
labelLevels (string) : an input.string with options "Inside", "Outside", "Remove". Determines whether FVG labels should be inside box, outside,
or na.
borderTransp (int)
liteMode (bool)
rightCoordinate (int) : the right coordinate of any drawings. Must be a time point.
Returns: void
invalidBlockCheck(bullishOBbox, bearishOBbox, userTF)
check if existing order blocks are still valid
Parameters:
bullishOBbox (array) : an array declared using the UDT orderBlock that contains bullish order block related data
bearishOBbox (array) : an array declared using the UDT orderBlock that contains bearish order block related data
userTF (simple int)
Returns: void (0)
method lastBarRejections(id, rejectionColor, idShow, rejectionString, labelLevels, borderTransp, liteMode, rightCoordinate, masterCoords)
draws rejectionBlocks on last bar
Namespace types: array
Parameters:
id (array) : the array, an array of rejection block data declared using the UDT rejection block
rejectionColor (color) : the desired color of the rejection box
idShow (int)
rejectionString (string) : the desired name of the rejection blocks
labelLevels (string) : an input.string() to determine if labels for the block should be inside the box, outside, or none.
borderTransp (int)
liteMode (bool) : an input.bool(). True = neon effect, false = no neon.
rightCoordinate (int) : atime for the right coordinate of the box
masterCoords (map) : a map that stores the price of key levels and assigns them a name, used to determine price ranges
Returns: void
method OBdraw(id, OBshow, BBshow, OBcol, BBcol, bullishString, bearishString, isBullish, labelLevels, borderTransp, liteMode, rightCoordinate, masterCoords)
draws orderblocks and breaker blocks for data stored in UDT array()
Namespace types: array
Parameters:
id (array) : the array, an array of order block data declared using the UDT orderblock
OBshow (int) : the number of order blocks to show
BBshow (int) : the number of breaker blocks to show
OBcol (color) : color of order blocks
BBcol (color) : color of breaker blocks
bullishString (string) : the title of bullish blocks, which is a regular bullish orderblock or a bearish orderblock that's converted to breakerblock
bearishString (string) : the title of bearish blocks, which is a regular bearish orderblock or a bullish orderblock that's converted to breakerblock
isBullish (bool) : whether the array contains bullish orderblocks or bearish orderblocks. If bullish orderblocks,
the array will naturally contain bearish BB, and if bearish OB, the array will naturally contain bullish BB
labelLevels (string) : an input.string() to determine if labels for the block should be inside the box, outside, or none.
borderTransp (int)
liteMode (bool) : an input.bool(). True = neon effect, false = no neon.
rightCoordinate (int) : atime for the right coordinate of the box
masterCoords (map) : a map that stores the price of key levels and assigns them a name, used to determine price ranges
Returns: void
FVG
UDT for FVG calcualtions
Fields:
H (series float) : high price of user-selected timeframe
L (series float) : low price of user-selected timeframe
direction (series string) : FVG direction => "Up" or "Down"
T (series int) : => time of bar on user-selected timeframe where FVG was created
fvgLabel (series label) : optional label for FVG
fvgLineTop (series line) : optional line for top of FVG
fvgLineBot (series line) : optional line for bottom of FVG
fvgBox (series box) : optional box for FVG
labelLine
quickly pair a line and label together as UDT
Fields:
lin (series line) : Line you wish to pair with label
lab (series label) : Label you wish to pair with line
orderBlock
UDT for order block calculations
Fields:
orderBlockData (array) : array containing order block x and y points
orderBlockBox (series box) : optional order block box
vioCount (series int) : = 0 violation count of the order block. 0 = Order Block, 1 = Breaker Block
traded (series bool)
status (series string) : = "OB" status == "OB" => Level is order block. status == "BB" => Level is breaker block.
orderBlockLab (series label) : options label for the order block / breaker block.
strongPoints
UDT for strong highs and strong lows
Fields:
price (series float) : price of the strong high or strong low
timeAtprice (series int) : time of the strong high or strong low
strongPointLabel (series label) : optional label for strong point
strongPointLine (series line) : optional line for strong point
overlayLine (series line) : optional lines for strong point to enhance visibility
overlayLine2 (series line) : optional lines for strong point to enhance visibility
displacement
UDT for dispacements
Fields:
highPrice (series float) : high price of displacement
lowPrice (series float) : low price of displacement
timeAtPrice (series int) : time of bar where displacement occurred
displacementBox (series box) : optional box to draw displacement
displacementLab (series label) : optional label for displacement
po3data
UDT for po3 calculations
Fields:
dHigh (series float) : higher timeframe high price
dLow (series float) : higher timeframe low price
dOpen (series float) : higher timeframe open price
dClose (series float) : higher timeframe close price
po3box (series box) : box to draw po3 candle body
po3line (array) : line array to draw po3 wicks
po3Labels (array) : label array to label price points of po3 candle
macros
UDT for session macros
Fields:
sessions (array) : Array of sessions, you can populate this array using the "quickTime" function located above "export macros".
prices (matrix) : Matrix of session data -> open, high, low, close, time
sessionTimes (array) : Array of session names. Pairs with array sessions.
sessionLines (matrix) : Optional array for sesion drawings.
OTEtimes
UDT for data storage and drawings associated with OTE strategy
Fields:
upTimes (array) : time of highest point before trade is taken
dnTimes (array) : time of lowest point before trade is taken
tpLineLong (series line) : line to mark tp level long
tpLabelLong (series label) : label to mark tp level long
slLineLong (series line) : line to mark sl level long
slLabelLong (series label) : label to mark sl level long
tpLineShort (series line) : line to mark tp level short
tpLabelShort (series label) : label to mark tp level short
slLineShort (series line) : line to mark sl level short
slLabelShort (series label) : label to mark sl level short
sweeps
UDT for data storage and drawings associated with liquidity sweeps
Fields:
upSweeps (matrix) : matrix containing liquidity sweep price points and time points for up sweeps
dnSweeps (matrix) : matrix containing liquidity sweep price points and time points for down sweeps
upSweepDrawings (array) : optional up sweep box array. Pair the size of this array with the rows or columns,
dnSweepDrawings (array) : optional up sweep box array. Pair the size of this array with the rows or columns,
raidExitDrawings
UDT for drawings associated with the Liquidity Raid Strategy
Fields:
tpLine (series line) : tp line for the liquidity raid entry
tpLabel (series label) : tp label for the liquidity raid entry
slLine (series line) : sl line for the liquidity raid entry
slLabel (series label) : sl label for the liquidity raid entry
m2022
UDT for data storage and drawings associated with the Model 2022 Strategy
Fields:
mTime (series int) : time of the FVG where entry limit order is placed
mIndex (series int) : array index of FVG where entry limit order is placed. This requires an array of FVG data, which is defined above.
mEntryDistance (series float) : the distance of the FVG to the 50% range. M2022 looks for the fvg closest to 50% mark of range.
mEntry (series float) : the entry price for the most eligible fvg
fvgHigh (series float) : the high point of the eligible fvg
fvgLow (series float) : the low point of the eligible fvg
longFVGentryBox (series box) : long FVG box, used to draw the eligible FVG
shortFVGentryBox (series box) : short FVG box, used to draw the eligible FVG
line50P (series line) : line used to mark 50% of the range
line100P (series line) : line used to mark 100% (top) of the range
line0P (series line) : line used to mark 0% (bottom) of the range
label50P (series label) : label used to mark 50% of the range
label100P (series label) : label used to mark 100% (top) of the range
label0P (series label) : label used to mark 0% (bottom) of the range
sweepData (array)
silverBullet
UDT for data storage and drawings associated with the Silver Bullet Strategy
Fields:
session (series bool)
sessionStr (series string) : name of the session for silver bullet
sessionBias (series string)
sessionHigh (series float) : = high high of session // use math.max(silverBullet.sessionHigh, high)
sessionLow (series float) : = low low of session // use math.min(silverBullet.sessionLow, low)
sessionFVG (series float) : if applicable, the FVG created during the session
sessionFVGdraw (series box) : if applicable, draw the FVG created during the session
traded (series bool)
tp (series float) : tp of trade entered at the session FVG
sl (series float) : sl of trade entered at the session FVG
sessionDraw (series box) : optional draw session with box
sessionDrawLabel (series label) : optional label session with label
silverBulletDrawings
UDT for trade exit drawings associated with the Silver Bullet Strategy
Fields:
tpLine (series line) : tp line drawing for strategy
tpLabel (series label) : tp label drawing for strategy
slLine (series line) : sl line drawing for strategy
slLabel (series label) : sl label drawing for strategy
unicornModel
UDT for data storage and drawings associated with the Unicorn Model Strategy
Fields:
hPoint (chart.point)
hPoint2 (chart.point)
hPoint3 (chart.point)
breakerBlock (series box) : used to draw the breaker block required for the Unicorn Model
FVG (series box) : used to draw the FVG required for the Unicorn model
topBlock (series float) : price of top of breaker block, can be used to detail trade entry
botBlock (series float) : price of bottom of breaker block, can be used to detail trade entry
startBlock (series int) : start time of the breaker block, used to set the "left = " param for the box
includes (array) : used to store the time of the breaker block, or FVG, or the chart point sequence that setup the Unicorn Model.
entry (series float) : // eligible entry price, for longs"math.max(topBlock, FVG.get_top())",
tpLine (series line) : optional line to mark PT
tpLabel (series label) : optional label to mark PT
slLine (series line) : optional line to mark SL
slLabel (series label) : optional label to mark SL
rejectionBlocks
UDT for data storage and drawings associated with rejection blocks
Fields:
rejectionPoint (chart.point)
bodyPrice (series float) : candle body price closest to the rejection point, for "Up" rejections => math.max(open, close),
rejectionBox (series box) : optional box drawing of the rejection block
rejectionLabel (series label) : optional label for the rejection block
equalLevelsDraw
UDT for data storage and drawings associated with equal highs / equal lows
Fields:
connector (series line) : single line placed at the first high or low, y = avgerage of distinguished equal highs/lows
connectorLab (series label) : optional label to be placed at the highs or lows
levels (array) : array containing the equal highs or lows prices
times (array) : array containing the equal highs or lows individual times
startTime (series int) : the time of the first high or low that forms a sequence of equal highs or lows
radiate (array) : options label to "radiate" the label in connector lab. Can be used for anything
necessaryData
UDT for data storage of historical price points.
Fields:
highArr (array) : array containing historical high points
lowArr (array) : array containing historical low points
timeArr (array) : array containing historical time points
logArr (array) : array containing historical log returns
signArr (array) : array containing historical price directions
closeArr (array) : array containing historical close points
binaryTimeArr (array) : array containing historical time points, uses "push" instead of "unshift" to allow for binary search
binaryCloseArr (array) : array containing historical close points, uses "push" instead of "unshift" to allow the correct
binaryOpenArr (array) : array containing historical optn points, uses "push" instead of "unshift" to allow the correct
atrTFarr (array) : array containing historical user-selected TF atr points
openArr (array) : array containing historical open points
kNNLibrary "kNN"
Collection of experimental kNN functions. This is a work in progress, an improvement upon my original kNN script:
The script can be recreated with this library. Unlike the original script, that used multiple arrays, this has been reworked with the new Pine Script matrix features.
To make a kNN prediction, the following data should be supplied to the wrapper:
kNN : filter type. Right now either Binary or Percent . Binary works like in the original script: the system stores whether the price has increased (+1) or decreased (-1) since the previous knnStore event (called when either long or short condition is supplied). Percent works the same, but the values stored are the difference of prices in percents. That way larger differences in prices would give higher scores.
k : number k. This is how many nearest neighbors are to be selected (and summed up to get the result).
skew : kNN minimum difference. Normally, the prediction is done with a simple majority of the neighbor votes. If skew is given, then more than a simple majority is needed for a prediction. This also means that there are inputs for which no prediction would be given (if the majority votes are between -skew and +skew). Note that in Percent mode more profitable trades will have higher voting power.
depth : kNN matrix size limit. Originally, the whole available history of trades was used to make a prediction. This not only requires more computational power, but also neglects the fact that the market conditions are changing. This setting restricts the memory matrix to a finite number of past trades.
price : price series
long : long condition. True if the long conditions are met, but filters are not yet applied. For example, in my original script, trades are only made on crossings of fast and slow MAs. So, whenever it is possible to go long, this value is set true. False otherwise.
short : short condition. Same as long , but for short condition.
store : whether the inputs should be stored. Additional filters may be applied to prevent bad trades (for example, trend-based filters), so if you only need to consult kNN without storing the trade, this should be set to false.
feature1 : current value of feature 1. A feature in this case is some kind of data derived from the price. Different features may be used to analyse the price series. For example, oscillator values. Not all of them may be used for kNN prediction. As the current kNN implementation is 2-dimensional, only two features can be used.
feature2 : current value of feature 2.
The wrapper returns a tuple: [ longOK, shortOK ]. This is a pair of filters. When longOK is true, then kNN predicts a long trade may be taken. When shortOK is true, then kNN predicts a short trade may be taken. The kNN filters are returned whenever long or short conditions are met. The trade is supposed to happen when long or short conditions are met and when the kNN filter for the desired direction is true.
Exported functions :
knnStore(knn, p1, p2, src, maxrows)
Store the previous trade; buffer the current one until results are in. Results are binary: up/down
Parameters:
knn : knn matrix
p1 : feature 1 value
p2 : feature 2 value
src : current price
maxrows : limit the matrix size to this number of rows (0 of no limit)
Returns: modified knn matrix
knnStorePercent(knn, p1, p2, src, maxrows)
Store the previous trade; buffer the current one until results are in. Results are in percents
Parameters:
knn : knn matrix
p1 : feature 1 value
p2 : feature 2 value
src : current price
maxrows : limit the matrix size to this number of rows (0 of no limit)
Returns: modified knn matrix
knnGet(distance, result)
Get neighbours by getting k results with the smallest distances
Parameters:
distance : distance array
result : result array
Returns: array slice of k results
knnDistance(knn, p1, p2)
Create a distance array from the two given parameters
Parameters:
knn : knn matrix
p1 : feature 1 value
p2 : feature 2 value
Returns: distance array
knnSum(knn, p1, p2, k)
Make a prediction, finding k nearest neighbours and summing them up
Parameters:
knn : knn matrix
p1 : feature 1 value
p2 : feature 2 value
k : sum k nearest neighbors
Returns: sum of k nearest neighbors
doKNN(kNN, k, skew, depth, price, long, short, store, feature1, feature2)
execute kNN filter
Parameters:
kNN : filter type
k : number k
skew : kNN minimum difference
depth : kNN matrix size limit
price : series
long : long condition
short : short condition
store : store the supplied features (if false, only checks the results without storage)
feature1 : feature 1 value
feature2 : feature 2 value
Returns: filter output
Bifurcation Zone - CAEBifurcation Zone — Cognitive Adversarial Engine (BZ-CAE)
Bifurcation Zone — CAE (BZ-CAE) is a next-generation divergence detection system enhanced by a Cognitive Adversarial Engine that evaluates both sides of every potential trade before presenting signals. Unlike traditional divergence indicators that show every price-oscillator disagreement regardless of context, BZ-CAE applies comprehensive market-state intelligence to identify only the divergences that occur in favorable conditions with genuine probability edges.
The system identifies structural bifurcation points — critical junctures where price and momentum disagree, signaling potential reversals or continuations — then validates these opportunities through five interconnected intelligence layers: Trend Conviction Scoring , Directional Momentum Alignment , Multi-Factor Exhaustion Modeling , Adversarial Validation , and Confidence Scoring . The result is a selective, context-aware signal system that filters noise and highlights high-probability setups.
This is not a "buy the arrow" indicator. It's a decision support framework that teaches you how to read market state, evaluate divergence quality, and make informed trading decisions based on quantified intelligence rather than hope.
What Sets BZ-CAE Apart: Technical Architecture
The Problem With Traditional Divergence Indicators
Most divergence indicators operate on a simple rule: if price makes a higher high and RSI makes a lower high, show a bearish signal. If price makes a lower low and RSI makes a higher low, show a bullish signal. This creates several critical problems:
Context Blindness : They show counter-trend signals in powerful trends that rarely reverse, leading to repeated losses as you fade momentum.
Signal Spam : Every minor price-oscillator disagreement generates an alert, overwhelming you with low-quality setups and creating analysis paralysis.
No Quality Ranking : All signals are treated identically. A marginal divergence in choppy conditions receives the same visual treatment as a high-conviction setup at a major exhaustion point.
Single-Sided Evaluation : They ask "Is this a good long?" without checking if the short case is overwhelmingly stronger, leading you into obvious bad trades.
Static Configuration : You manually choose RSI 14 or Stochastic 14 and hope it works, with no systematic way to validate if that's optimal for your instrument.
BZ-CAE's Solution: Cognitive Adversarial Intelligence
BZ-CAE solves these problems through an integrated five-layer intelligence architecture:
1. Trend Conviction Score (TCS) — 0 to 1 Scale
Most indicators check if ADX is above 25 to determine "trending" conditions. This binary approach misses nuance. TCS is a weighted composite metric:
Formula : 0.35 × normalize(ADX, 10, 35) + 0.35 × structural_strength + 0.30 × htf_alignment
Structural Strength : 10-bar SMA of consecutive directional bars. Captures persistence — are bulls or bears consistently winning?
HTF Alignment : Multi-timeframe EMA stacking (20/50/100/200). When all EMAs align in the same direction, you're in institutional trend territory.
Purpose : Quantifies how "locked in" the trend is. When TCS exceeds your threshold (default 0.80), the system knows to avoid counter-trend trades unless other factors override.
Interpretation :
TCS > 0.85: Very strong trend — counter-trading is extremely high risk
TCS 0.70-0.85: Strong trend — favor continuation, require exhaustion for reversals
TCS 0.50-0.70: Moderate trend — context matters, both directions viable
TCS < 0.50: Weak/choppy — reversals more viable, range-bound conditions
2. Directional Momentum Alignment (DMA) — ATR-Normalized
Formula : (EMA21 - EMA55) / ATR14
This isn't just "price above EMA" — it's a regime-aware momentum gauge. The same $100 price movement reads completely differently in high-volatility crypto versus low-volatility forex. By normalizing with ATR, DMA adapts its interpretation to current market conditions.
Purpose : Quantifies the directional "force" behind current price action. Positive = bullish push, negative = bearish push. Magnitude = strength.
Interpretation :
DMA > 0.7: Strong bullish momentum — bearish divergences risky
DMA 0.3 to 0.7: Moderate bullish bias
DMA -0.3 to 0.3: Balanced/choppy conditions
DMA -0.7 to -0.3: Moderate bearish bias
DMA < -0.7: Strong bearish momentum — bullish divergences risky
3. Multi-Factor Exhaustion Modeling — 0 to 1 Probability
Single-metric exhaustion detection (like "RSI > 80") misses complex market states. BZ-CAE aggregates five independent exhaustion signals:
Volume Spikes : Current volume versus 50-bar average
2.5x average: 0.25 weight
2.0x average: 0.15 weight
1.5x average: 0.10 weight
Divergence Present : The fact that a divergence exists contributes 0.30 weight — structural momentum disagreement is itself an exhaustion signal.
RSI Extremes : Captures oscillator climax zones
RSI > 80 or < 20: 0.25 weight
RSI > 75 or < 25: 0.15 weight
Pin Bar Detection : Identifies rejection candles (2:1 wick-to-body ratio, indicating failed breakout attempts): 0.15 weight
Extended Runs : Consecutive bars above/below EMA20 without pullback
30+ bars: 0.15 weight (market hasn't paused to consolidate)
Total exhaustion score is the sum of all applicable weights, capped at 1.0.
Purpose : Detects when strong trends become vulnerable to reversal. High exhaustion can override trend filters, allowing counter-trend trades at genuine turning points that basic indicators would miss.
Interpretation :
Exhaustion > 0.75: High probability of climax — yellow background shading alerts you visually
Exhaustion 0.50-0.75: Moderate overextension — watch for confirmation
Exhaustion < 0.50: Fresh move — trend can continue, counter-trend trades higher risk
4. Adversarial Validation — Game Theory Applied to Trading
This is BZ-CAE's signature innovation. Before approving any signal, the engine quantifies BOTH sides of the trade simultaneously:
For Bullish Divergences , it calculates:
Bull Case Score (0-1+) :
Distance below EMA20 (pullback quality): up to 0.25
Bullish EMA alignment (close > EMA20 > EMA50): 0.25
Oversold RSI (< 40): 0.25
Volume confirmation (> 1.2x average): 0.25
Bear Case Score (0-1+) :
Price below EMA50 (structural weakness): 0.30
Very oversold RSI (< 30, indicating knife-catching): 0.20
Differential = Bull Case - Bear Case
If differential < -0.10 (default threshold), the bear case is dominating — signal is BLOCKED or ANNOTATED.
For Bearish Divergences , the logic inverts (Bear Case vs Bull Case).
Purpose : Prevents trades where you're fighting obvious strength in the opposite direction. This is institutional-grade risk management — don't just evaluate your trade, evaluate the counter-trade simultaneously.
Why This Matters : You might see a bullish divergence at a local low, but if price is deeply below major support EMAs with strong bearish momentum, you're catching a falling knife. The adversarial check catches this and blocks the signal.
5. Confidence Scoring — 0 to 1 Quality Assessment
Every signal that passes initial filters receives a comprehensive quality score:
Formula :
0.30 × normalize(TCS) // Trend context
+ 0.25 × normalize(|DMA|) // Momentum magnitude
+ 0.20 × pullback_quality // Entry distance from EMA20
+ 0.15 × state_quality // ADX + alignment + structure
+ 0.10 × divergence_strength // Slope separation magnitude
+ adversarial_bonus (0-0.30) // Your side's advantage
Purpose : Ranks setup quality for filtering and position sizing decisions. You can set a minimum confidence threshold (default 0.35) to ensure only quality setups reach your chart.
Interpretation :
Confidence > 0.70: Premium setup — consider increased position size
Confidence 0.50-0.70: Good quality — standard size
Confidence 0.35-0.50: Acceptable — reduced size or skip if conservative
Confidence < 0.35: Marginal — blocked in Filtering mode, annotated in Advisory mode
CAE Operating Modes: Learning vs Enforcement
Off : Disables all CAE logic. Raw divergence pipeline only. Use for baseline comparison.
Advisory : Shows ALL signals regardless of CAE evaluation, but annotates signals that WOULD be blocked with specific warnings (e.g., "Bull: strong downtrend (TCS=0.87)" or "Adversarial bearish"). This is your learning mode — see CAE's decision logic in action without missing educational opportunities.
Filtering : Actively blocks low-quality signals. Only setups that pass all enabled gates (Trend Filter, Adversarial Validation, Confidence Gating) reach your chart. This is your live trading mode — trust the system to enforce discipline.
CAE Filter Gates: Three-Layer Protection
When CAE is enabled, signals must pass through three independent gates (each can be toggled on/off):
Gate 1: Strong Trend Filter
If TCS ≥ tcs_threshold (default 0.80)
And signal is counter-trend (bullish in downtrend or bearish in uptrend)
And exhaustion < exhaustion_required (default 0.50)
Then: BLOCK signal
Logic: Don't fade strong trends unless the move is clearly overextended
Gate 2: Adversarial Validation
Calculate both bull case and bear case scores
If opposing case dominates by more than adv_threshold (default 0.10)
Then: BLOCK signal
Logic: Avoid trades where you're fighting obvious strength in the opposite direction
Gate 3: Confidence Gating
Calculate composite confidence score (0-1)
If confidence < min_confidence (default 0.35)
Then: In Filtering mode, BLOCK signal; in Advisory mode, ANNOTATE with warning
Logic: Only take setups with minimum quality threshold
All three gates work together. A signal must pass ALL enabled gates to fire.
Visual Intelligence System
Bifurcation Zones (Supply/Demand Blocks)
When a divergence signal fires, BZ-CAE draws a semi-transparent box extending 15 bars forward from the signal pivot:
Demand Zones (Bullish) : Theme-colored box (cyan in Cyberpunk, blue in Professional, etc.) labeled "Demand" — marks where smart money likely placed buy orders as price diverged at the low.
Supply Zones (Bearish) : Theme-colored box (magenta in Cyberpunk, orange in Professional) labeled "Supply" — marks where smart money likely placed sell orders as price diverged at the high.
Theory : Divergences represent institutional disagreement with the crowd. The crowd pushed price to an extreme (new high or low), but momentum (oscillator) is waning, indicating smart money is taking the opposite side. These zones mark order placement areas that become future support/resistance.
Use Cases :
Exit targets: Take profit when price returns to opposite-side zone
Re-entry levels: If price returns to your entry zone, consider adding
Stop placement: Place stops just beyond your zone (below demand, above supply)
Auto-Cleanup : System keeps the last 20 zones to prevent chart clutter.
Adversarial Bar Coloring — Real-Time Market Debate Heatmap
Each bar is colored based on the Bull Case vs Bear Case differential:
Strong Bull Advantage (diff > 0.3): Full theme bull color (e.g., cyan)
Moderate Bull Advantage (diff > 0.1): 50% transparency bull
Neutral (diff -0.1 to 0.1): Gray/neutral theme
Moderate Bear Advantage (diff < -0.1): 50% transparency bear
Strong Bear Advantage (diff < -0.3): Full theme bear color (e.g., magenta)
This creates a real-time visual heatmap showing which side is "winning" the market debate. When bars flip from cyan to magenta (or vice versa), you're witnessing a shift in adversarial advantage — a leading indicator of potential momentum changes.
Exhaustion Shading
When exhaustion score exceeds 0.75, the chart background displays a semi-transparent yellow highlight. This immediate visual warning alerts you that the current move is at high risk of reversal, even if trend indicators remain strong.
Visual Themes — Six Aesthetic Options
Cyberpunk : Cyan/Magenta/Yellow — High contrast, neon aesthetic, excellent for dark-themed trading environments
Professional : Blue/Orange/Green — Corporate color palette, suitable for presentations and professional documentation
Ocean : Teal/Red/Cyan — Aquatic palette, calming for extended monitoring sessions
Fire : Orange/Red/Coral — Warm aggressive colors, high energy
Matrix : Green/Red/Lime — Code aesthetic, homage to classic hacker visuals
Monochrome : White/Gray — Minimal distraction, maximum focus on price action
All visual elements (signal markers, zones, bar colors, dashboard) adapt to your selected theme.
Divergence Engine — Core Detection System
What Are Divergences?
Divergences occur when price action and momentum indicators disagree, creating structural tension that often resolves in a change of direction:
Regular Divergence (Reversal Signal) :
Bearish Regular : Price makes higher high, oscillator makes lower high → Potential trend reversal down
Bullish Regular : Price makes lower low, oscillator makes higher low → Potential trend reversal up
Hidden Divergence (Continuation Signal) :
Bearish Hidden : Price makes lower high, oscillator makes higher high → Downtrend continuation
Bullish Hidden : Price makes higher low, oscillator makes lower low → Uptrend continuation
Both types can be enabled/disabled independently in settings.
Pivot Detection Methods
BZ-CAE uses symmetric pivot detection with separate lookback and lookforward periods (default 5/5):
Pivot High : Bar where high > all highs within lookback range AND high > all highs within lookforward range
Pivot Low : Bar where low < all lows within lookback range AND low < all lows within lookforward range
This ensures structural validity — the pivot must be a clear local extreme, not just a minor wiggle.
Divergence Validation Requirements
For a divergence to be confirmed, it must satisfy:
Slope Disagreement : Price slope and oscillator slope must move in opposite directions (for regular divs) or same direction with inverted highs/lows (for hidden divs)
Minimum Slope Change : |osc_slope| > min_slope_change / 100 (default 1.0) — filters weak, marginal divergences
Maximum Lookback Range : Pivots must be within max_lookback bars (default 60) — prevents ancient, irrelevant divergences
ATR-Normalized Strength : Divergence strength = min(|price_slope| × |osc_slope| × 10, 1.0) — quantifies the magnitude of disagreement in volatility context
Regular divergences receive 1.0× weight; hidden divergences receive 0.8× weight (slightly less reliable historically).
Oscillator Options — Five Professional Indicators
RSI (Relative Strength Index) : Classic overbought/oversold momentum indicator. Best for: General purpose divergence detection across all instruments.
Stochastic : Range-bound %K momentum comparing close to high-low range. Best for: Mean reversion strategies and range-bound markets.
CCI (Commodity Channel Index) : Measures deviation from statistical mean, auto-normalized to 0-100 scale. Best for: Cyclical instruments and commodities.
MFI (Money Flow Index) : Volume-weighted RSI incorporating money flow. Best for: Volume-driven markets like stocks and crypto.
Williams %R : Inverse stochastic looking back over period, auto-adjusted to 0-100. Best for: Reversal detection at extremes.
Each oscillator has adjustable length (2-200, default 14) and smoothing (1-20, default 1). You also set overbought (50-100, default 70) and oversold (0-50, default 30) thresholds.
Signal Timing Modes — Understanding Repainting
BZ-CAE offers two timing policies with complete transparency about repainting behavior:
Realtime (1-bar, peak-anchored)
How It Works :
Detects peaks 1 bar ago using pattern: high > high AND high > high
Signal prints on the NEXT bar after peak detection (bar_index)
Visual marker anchors to the actual PEAK bar (bar_index - 1, offset -1)
Signal locks in when bar CONFIRMS (closes)
Repainting Behavior :
On the FORMING bar (before close), the peak condition may change as new prices arrive
Once bar CLOSES (barstate.isconfirmed), signal is locked permanently
This is preview/early warning behavior by design
Best For :
Active monitoring and immediate alerts
Learning the system (seeing signals develop in real-time)
Responsive entry if you're watching the chart live
Confirmed (lookforward)
How It Works :
Uses Pine Script's built-in ta.pivothigh() and ta.pivotlow() functions
Requires full pivot validation period (lookback + lookforward bars)
Signal prints pivot_lookforward bars after the actual peak (default 5-bar delay)
Visual marker anchors to the actual peak bar (offset -pivot_lookforward)
No Repainting Behavior
Best For :
Backtesting and historical analysis
Conservative entries requiring full confirmation
Automated trading systems
Swing trading with larger timeframes
Tradeoff :
Delayed entry by pivot_lookforward bars (typically 5 bars)
On a 5-minute chart, this is a 25-minute delay
On a 4-hour chart, this is a 20-hour delay
Recommendation : Use Confirmed for backtesting to verify system performance honestly. Use Realtime for live monitoring only if you're actively watching the chart and understand pre-confirmation repainting behavior.
Signal Spacing System — Anti-Spam Architecture
Even after CAE filtering, raw divergences can cluster. The spacing system enforces separation:
Three Independent Filters
1. Min Bars Between ANY Signals (default 12):
Prevents rapid-fire clustering across both directions
If last signal (bull or bear) was within N bars, block new signal
Ensures breathing room between all setups
2. Min Bars Between SAME-SIDE Signals (default 24, optional enforcement):
Prevents bull-bull or bear-bear spam
Separate tracking for bullish and bearish signal timelines
Toggle enforcement on/off
3. Min ATR Distance From Last Signal (default 0, optional):
Requires price to move N × ATR from last signal location
Ensures meaningful price movement between setups
0 = disabled, 0.5-2.0 = typical range for enabled
All three filters work independently. A signal must pass ALL enabled filters to proceed.
Practical Guidance :
Scalping (1-5m) : Any 6-10, Same-side 12-20, ATR 0-0.5
Day Trading (15m-1H) : Any 12, Same-side 24, ATR 0-1.0
Swing Trading (4H-D) : Any 20-30, Same-side 40-60, ATR 1.0-2.0
Dashboard — Real-Time Control Center
The dashboard (toggleable, four corner positions, three sizes) provides comprehensive system intelligence:
Oscillator Section
Current oscillator type and value
State: OVERBOUGHT / OVERSOLD / NEUTRAL (color-coded)
Length parameter
Cognitive Engine Section
TCS (Trend Conviction Score) :
Current value with emoji state indicator
🔥 = Strong trend (>0.75)
📊 = Moderate trend (0.50-0.75)
〰️ = Weak/choppy (<0.50)
Color: Red if above threshold (trend filter active), yellow if moderate, green if weak
DMA (Directional Momentum Alignment) :
Current value with emoji direction indicator
🐂 = Bullish momentum (>0.5)
⚖️ = Balanced (-0.5 to 0.5)
🐻 = Bearish momentum (<-0.5)
Color: Green if bullish, red if bearish
Exhaustion :
Current value with emoji warning indicator
⚠️ = High exhaustion (>0.75)
🟡 = Moderate (0.50-0.75)
✓ = Low (<0.50)
Color: Red if high, yellow if moderate, green if low
Pullback :
Quality of current distance from EMA20
Values >0.6 are ideal entry zones (not too close, not too far)
Bull Case / Bear Case (if Adversarial enabled):
Current scores for both sides of the market debate
Differential with emoji indicator:
📈 = Bull advantage (>0.2)
➡️ = Balanced (-0.2 to 0.2)
📉 = Bear advantage (<-0.2)
Last Signal Metrics Section (New Feature)
When a signal fires, this section captures and displays:
Signal type (BULL or BEAR)
Bars elapsed since signal
Confidence % at time of signal
TCS value at signal time
DMA value at signal time
Purpose : Provides a historical reference for learning. You can see what the market state looked like when the last signal fired, helping you correlate outcomes with conditions.
Statistics Section
Total Signals : Lifetime count across session
Blocked Signals : Count and percentage (filter effectiveness metric)
Bull Signals : Total bullish divergences
Bear Signals : Total bearish divergences
Purpose : System health monitoring. If blocked % is very high (>60%), filters may be too strict. If very low (<10%), filters may be too loose.
Advisory Annotations
When CAE Mode = Advisory, this section displays warnings for signals that would be blocked in Filtering mode:
Examples:
"Bull spacing: wait 8 bars"
"Bear: strong uptrend (TCS=0.87)"
"Adversarial bearish"
"Low confidence 32%"
Multiple warnings can stack, separated by " | ". This teaches you CAE's decision logic transparently.
How to Use BZ-CAE — Complete Workflow
Phase 1: Initial Setup (First Session)
Apply BZ-CAE to your chart
Select your preferred Visual Theme (Cyberpunk recommended for visibility)
Set Signal Timing to "Confirmed (lookforward)" for learning
Choose your Oscillator Type (RSI recommended for general use, length 14)
Set Overbought/Oversold to 70/30 (standard)
Enable both Regular Divergence and Hidden Divergence
Set Pivot Lookback/Lookforward to 5/5 (balanced structure)
Enable CAE Intelligence
Set CAE Mode to "Advisory" (learning mode)
Enable all three CAE filters: Strong Trend Filter , Adversarial Validation , Confidence Gating
Enable Show Dashboard , position Top Right, size Normal
Enable Draw Bifurcation Zones and Adversarial Bar Coloring
Phase 2: Learning Period (Weeks 1-2)
Goal : Understand how CAE evaluates market state and filters signals.
Activities :
Watch the dashboard during signals :
Note TCS values when counter-trend signals fail — this teaches you the trend strength threshold for your instrument
Observe exhaustion patterns at actual turning points — learn when overextension truly matters
Study adversarial differential at signal times — see when opposing cases dominate
Review blocked signals (orange X-crosses):
In Advisory mode, you see everything — signals that would pass AND signals that would be blocked
Check the advisory annotations to understand why CAE would block
Track outcomes: Were the blocks correct? Did those signals fail?
Use Last Signal Metrics :
After each signal, check the dashboard capture of confidence, TCS, and DMA
Journal these values alongside trade outcomes
Identify patterns: Do confidence >0.70 signals work better? Does your instrument respect TCS >0.85?
Understand your instrument's "personality" :
Trending instruments (indices, major forex) may need TCS threshold 0.85-0.90
Choppy instruments (low-cap stocks, exotic pairs) may work best with TCS 0.70-0.75
High-volatility instruments (crypto) may need wider spacing
Low-volatility instruments may need tighter spacing
Phase 3: Calibration (Weeks 3-4)
Goal : Optimize settings for your specific instrument, timeframe, and style.
Calibration Checklist :
Min Confidence Threshold :
Review confidence distribution in your signal journal
Identify the confidence level below which signals consistently fail
Set min_confidence slightly above that level
Day trading : 0.35-0.45
Swing trading : 0.40-0.55
Scalping : 0.30-0.40
TCS Threshold :
Find the TCS level where counter-trend signals consistently get stopped out
Set tcs_threshold at or slightly below that level
Trending instruments : 0.85-0.90
Mixed instruments : 0.80-0.85
Choppy instruments : 0.75-0.80
Exhaustion Override Level :
Identify exhaustion readings that marked genuine reversals
Set exhaustion_required just below the average
Typical range : 0.45-0.55
Adversarial Threshold :
Default 0.10 works for most instruments
If you find CAE is too conservative (blocking good trades), raise to 0.15-0.20
If signals are still getting caught in opposing momentum, lower to 0.07-0.09
Spacing Parameters :
Count bars between quality signals in your journal
Set min bars ANY to ~60% of that average
Set min bars SAME-SIDE to ~120% of that average
Scalping : Any 6-10, Same 12-20
Day trading : Any 12, Same 24
Swing : Any 20-30, Same 40-60
Oscillator Selection :
Try different oscillators for 1-2 weeks each
Track win rate and average winner/loser by oscillator type
RSI : Best for general use, clear OB/OS
Stochastic : Best for range-bound, mean reversion
MFI : Best for volume-driven markets
CCI : Best for cyclical instruments
Williams %R : Best for reversal detection
Phase 4: Live Deployment
Goal : Disciplined execution with proven, calibrated system.
Settings Changes :
Switch CAE Mode from Advisory to Filtering
System now actively blocks low-quality signals
Only setups passing all gates reach your chart
Keep Signal Timing on Confirmed for conservative entries
OR switch to Realtime if you're actively monitoring and want faster entries (accept pre-confirmation repaint risk)
Use your calibrated thresholds from Phase 3
Enable high-confidence alerts: "⭐ High Confidence Bullish/Bearish" (>0.70)
Trading Discipline Rules :
Respect Blocked Signals :
If CAE blocks a trade you wanted to take, TRUST THE SYSTEM
Don't manually override — if you consistently disagree, return to Phase 2/3 calibration
The block exists because market state failed intelligence checks
Confidence-Based Position Sizing :
Confidence >0.70: Standard or increased size (e.g., 1.5-2.0% risk)
Confidence 0.50-0.70: Standard size (e.g., 1.0% risk)
Confidence 0.35-0.50: Reduced size (e.g., 0.5% risk) or skip if conservative
TCS-Based Management :
High TCS + counter-trend signal: Use tight stops, quick exits (you're fading momentum)
Low TCS + reversal signal: Use wider stops, trail aggressively (genuine reversal potential)
Exhaustion Awareness :
Exhaustion >0.75 (yellow shading): Market is overextended, reversal risk is elevated — consider early exit or tighter trailing stops even on winning trades
Exhaustion <0.30: Continuation bias — hold for larger move, wide trailing stops
Adversarial Context :
Strong differential against you (e.g., bullish signal with bear diff <-0.2): Use very tight stops, consider skipping
Strong differential with you (e.g., bullish signal with bull diff >0.2): Trail aggressively, this is your tailwind
Practical Settings by Timeframe & Style
Scalping (1-5 Minute Charts)
Objective : High frequency, tight stops, quick reversals in fast-moving markets.
Oscillator :
Type: RSI or Stochastic (fast response to quick moves)
Length: 9-11 (more responsive than standard 14)
Smoothing: 1 (no lag)
OB/OS: 65/35 (looser thresholds ensure frequent crossings in fast conditions)
Divergence :
Pivot Lookback/Lookforward: 3/3 (tight structure, catch small swings)
Max Lookback: 40-50 bars (recent structure only)
Min Slope Change: 0.8-1.0 (don't be overly strict)
CAE :
Mode: Advisory first (learn), then Filtering
Min Confidence: 0.30-0.35 (lower bar for speed, accept more signals)
TCS Threshold: 0.70-0.75 (allow more counter-trend opportunities)
Exhaustion Required: 0.45-0.50 (moderate override)
Strong Trend Filter: ON (still respect major intraday trends)
Adversarial: ON (critical for scalping protection — catches bad entries quickly)
Spacing :
Min Bars ANY: 6-10 (fast pace, many setups)
Min Bars SAME-SIDE: 12-20 (prevent clustering)
Min ATR Distance: 0 or 0.5 (loose)
Timing : Realtime (speed over precision, but understand repaint risk)
Visuals :
Signal Size: Tiny (chart clarity in busy conditions)
Show Zones: Optional (can clutter on low timeframes)
Bar Coloring: ON (helps read momentum shifts quickly)
Dashboard: Small size (corner reference, not main focus)
Key Consideration : Scalping generates noise. Even with CAE, expect lower win rate (45-55%) but aim for favorable R:R (2:1 or better). Size conservatively.
Day Trading (15-Minute to 1-Hour Charts)
Objective : Balance quality and frequency. Standard divergence trading approach.
Oscillator :
Type: RSI or MFI (proven reliability, volume confirmation with MFI)
Length: 14 (industry standard, well-studied)
Smoothing: 1-2
OB/OS: 70/30 (classic levels)
Divergence :
Pivot Lookback/Lookforward: 5/5 (balanced structure)
Max Lookback: 60 bars
Min Slope Change: 1.0 (standard strictness)
CAE :
Mode: Filtering (enforce discipline from the start after brief Advisory learning)
Min Confidence: 0.35-0.45 (quality filter without being too restrictive)
TCS Threshold: 0.80-0.85 (respect strong trends)
Exhaustion Required: 0.50 (balanced override threshold)
Strong Trend Filter: ON
Adversarial: ON
Confidence Gating: ON (all three filters active)
Spacing :
Min Bars ANY: 12 (breathing room between all setups)
Min Bars SAME-SIDE: 24 (prevent bull/bear clusters)
Min ATR Distance: 0-1.0 (optional refinement, typically 0.5-1.0)
Timing : Confirmed (1-bar delay for reliability, no repainting)
Visuals :
Signal Size: Tiny or Small
Show Zones: ON (useful reference for exits/re-entries)
Bar Coloring: ON (context awareness)
Dashboard: Normal size (full visibility)
Key Consideration : This is the "sweet spot" timeframe for BZ-CAE. Market structure is clear, CAE has sufficient data, and signal frequency is manageable. Expect 55-65% win rate with proper execution.
Swing Trading (4-Hour to Daily Charts)
Objective : Quality over quantity. High conviction only. Larger stops and targets.
Oscillator :
Type: RSI or CCI (robust on higher timeframes, smooth longer waves)
Length: 14-21 (capture larger momentum swings)
Smoothing: 1-3
OB/OS: 70/30 or 75/25 (strict extremes)
Divergence :
Pivot Lookback/Lookforward: 5/5 or 7/7 (structural purity, major swings only)
Max Lookback: 80-100 bars (broader historical context)
Min Slope Change: 1.2-1.5 (require strong, undeniable divergence)
CAE :
Mode: Filtering (strict enforcement, premium setups only)
Min Confidence: 0.40-0.55 (high bar for entry)
TCS Threshold: 0.85-0.95 (very strong trend protection — don't fade established HTF trends)
Exhaustion Required: 0.50-0.60 (higher bar for override — only extreme exhaustion justifies counter-trend)
Strong Trend Filter: ON (critical on HTF)
Adversarial: ON (avoid obvious bad trades)
Confidence Gating: ON (quality gate essential)
Spacing :
Min Bars ANY: 20-30 (substantial separation)
Min Bars SAME-SIDE: 40-60 (significant breathing room)
Min ATR Distance: 1.0-2.0 (require meaningful price movement)
Timing : Confirmed (purity over speed, zero repaint for swing accuracy)
Visuals :
Signal Size: Small or Normal (clear markers on zoomed-out view)
Show Zones: ON (important HTF levels)
Bar Coloring: ON (long-term trend awareness)
Dashboard: Normal or Large (comprehensive analysis)
Key Consideration : Swing signals are rare but powerful. Expect 2-5 signals per month per instrument. Win rate should be 60-70%+ due to stringent filtering. Position size can be larger given confidence.
Dashboard Interpretation Reference
TCS (Trend Conviction Score) States
0.00-0.50: Weak/Choppy
Emoji: 〰️
Color: Green/cyan
Meaning: No established trend. Range-bound or consolidating. Both reversal and continuation signals viable.
Action: Reversals (regular divs) are safer. Use wider profit targets (market has room to move). Consider mean reversion strategies.
0.50-0.75: Moderate Trend
Emoji: 📊
Color: Yellow/neutral
Meaning: Developing trend but not locked in. Context matters significantly.
Action: Check DMA and exhaustion. If DMA confirms trend and exhaustion is low, favor continuation (hidden divs). If exhaustion is high, reversals are viable.
0.75-0.85: Strong Trend
Emoji: 🔥
Color: Orange/warning
Meaning: Well-established trend with persistence. Counter-trend is high risk.
Action: Require exhaustion >0.50 for counter-trend entries. Favor continuation signals. Use tight stops on counter-trend attempts.
0.85-1.00: Very Strong Trend
Emoji: 🔥🔥
Color: Red/danger (if counter-trading)
Meaning: Locked-in institutional trend. Extremely high risk to fade.
Action: Avoid counter-trend unless exhaustion >0.75 (yellow shading). Focus exclusively on continuation opportunities. Momentum is king here.
DMA (Directional Momentum Alignment) Zones
-2.0 to -1.0: Strong Bearish Momentum
Emoji: 🐻🐻
Color: Dark red
Meaning: Powerful downside force. Sellers are in control.
Action: Bullish divergences are counter-momentum (high risk). Bearish divergences are with-momentum (lower risk). Size down on longs.
-0.5 to 0.5: Neutral/Balanced
Emoji: ⚖️
Color: Gray/neutral
Meaning: No strong directional bias. Choppy or consolidating.
Action: Both directions have similar probability. Focus on confidence score and adversarial differential for edge.
1.0 to 2.0: Strong Bullish Momentum
Emoji: 🐂🐂
Color: Bright green/cyan
Meaning: Powerful upside force. Buyers are in control.
Action: Bearish divergences are counter-momentum (high risk). Bullish divergences are with-momentum (lower risk). Size down on shorts.
Exhaustion States
0.00-0.50: Fresh Move
Emoji: ✓
Color: Green
Meaning: Trend is healthy, not overextended. Room to run.
Action: Counter-trend trades are premature. Favor continuation. Hold winners for larger moves. Avoid early exits.
0.50-0.75: Mature Move
Emoji: 🟡
Color: Yellow
Meaning: Move is aging. Watch for signs of climax.
Action: Tighten trailing stops on winning trades. Be ready for reversals. Don't add to positions aggressively.
0.75-0.85: High Exhaustion
Emoji: ⚠️
Color: Orange
Background: Yellow shading appears
Meaning: Move is overextended. Reversal risk elevated significantly.
Action: Counter-trend reversals are higher probability. Consider early exits on with-trend positions. Size up on reversal divergences (if CAE allows).
0.85-1.00: Critical Exhaustion
Emoji: ⚠️⚠️
Color: Red
Background: Yellow shading intensifies
Meaning: Climax conditions. Reversal imminent or underway.
Action: Aggressive reversal trades justified. Exit all with-trend positions. This is where major turns occur.
Confidence Score Tiers
0.00-0.30: Low Quality
Color: Red
Status: Blocked in Filtering mode
Action: Skip entirely. Setup lacks fundamental quality across multiple factors.
0.30-0.50: Moderate Quality
Color: Yellow/orange
Status: Marginal — passes in Filtering only if >min_confidence
Action: Reduced position size (0.5-0.75% risk). Tight stops. Conservative profit targets. Skip if you're selective.
0.50-0.70: High Quality
Color: Green/cyan
Status: Good setup across most quality factors
Action: Standard position size (1.0-1.5% risk). Normal stops and targets. This is your bread-and-butter trade.
0.70-1.00: Premium Quality
Color: Bright green/gold
Status: Exceptional setup — all factors aligned
Visual: Double confidence ring appears
Action: Consider increased position size (1.5-2.0% risk, maximum). Wider stops. Larger targets. High probability of success. These are rare — capitalize when they appear.
Adversarial Differential Interpretation
Bull Differential > 0.3 :
Visual: Strong cyan/green bar colors
Meaning: Bull case strongly dominates. Buyers have clear advantage.
Action: Bullish divergences favored (with-advantage). Bearish divergences face headwind (reduce size or skip). Momentum is bullish.
Bull Differential 0.1 to 0.3 :
Visual: Moderate cyan/green transparency
Meaning: Moderate bull advantage. Buyers have edge but not overwhelming.
Action: Both directions viable. Slight bias toward longs.
Differential -0.1 to 0.1 :
Visual: Gray/neutral bars
Meaning: Balanced debate. No clear advantage either side.
Action: Rely on other factors (confidence, TCS, exhaustion) for direction. Adversarial is neutral.
Bear Differential -0.3 to -0.1 :
Visual: Moderate red/magenta transparency
Meaning: Moderate bear advantage. Sellers have edge but not overwhelming.
Action: Both directions viable. Slight bias toward shorts.
Bear Differential < -0.3 :
Visual: Strong red/magenta bar colors
Meaning: Bear case strongly dominates. Sellers have clear advantage.
Action: Bearish divergences favored (with-advantage). Bullish divergences face headwind (reduce size or skip). Momentum is bearish.
Last Signal Metrics — Post-Trade Analysis
After a signal fires, dashboard captures:
Type : BULL or BEAR
Bars Ago : How long since signal (updates every bar)
Confidence : What was the quality score at signal time
TCS : What was trend conviction at signal time
DMA : What was momentum alignment at signal time
Use Case : Post-trade journaling and learning.
Example: "BULL signal 12 bars ago. Confidence: 68%, TCS: 0.42, DMA: -0.85"
Analysis : This was a bullish reversal (regular div) with good confidence, weak trend (TCS), but strong bearish momentum (DMA). The bet was that momentum would reverse — a counter-momentum play requiring exhaustion confirmation. Check if exhaustion was high at that time to justify the entry.
Track patterns:
Do your best trades have confidence >0.65?
Do low-TCS signals (<0.50) work better for you?
Are you more successful with-momentum (DMA aligned with signal) or counter-momentum?
Troubleshooting Guide
Problem: No Signals Appearing
Symptoms : Chart loads, dashboard shows metrics, but no divergence signals fire.
Diagnosis Checklist :
Check dashboard oscillator value : Is it crossing OB/OS levels (70/30)? If oscillator stays in 40-60 range constantly, it can't reach extremes needed for divergence detection.
Are pivots forming? : Look for local swing highs/lows on your chart. If price is in tight consolidation, pivots may not meet lookback/lookforward requirements.
Is spacing too tight? : Check "Last Signal" metrics — how many bars since last signal? If <12 and your min_bars_ANY is 12, spacing filter is blocking.
Is CAE blocking everything? : Check dashboard Statistics section — what's the blocked signal count? High blocks indicate overly strict filters.
Solutions :
Loosen OB/OS Temporarily :
Try 65/35 to verify divergence detection works
If signals appear, the issue was threshold strictness
Gradually tighten back to 67/33, then 70/30 as appropriate
Lower Min Confidence :
Try 0.25-0.30 (diagnostic level)
If signals appear, filter was too strict
Raise gradually to find sweet spot (0.35-0.45 typical)
Disable Strong Trend Filter Temporarily :
Turn off in CAE settings
If signals appear, TCS threshold was blocking everything
Re-enable and lower TCS_threshold to 0.70-0.75
Reduce Min Slope Change :
Try 0.7-0.8 (from default 1.0)
Allows weaker divergences through
Helpful on low-volatility instruments
Widen Spacing :
Set min_bars_ANY to 6-8
Set min_bars_SAME_SIDE to 12-16
Reduces time between allowed signals
Check Timing Mode :
If using Confirmed, remember there's a pivot_lookforward delay (5+ bars)
Switch to Realtime temporarily to verify system is working
Realtime has no delay but repaints
Verify Oscillator Settings :
Length 14 is standard but might not fit all instruments
Try length 9-11 for faster response
Try length 18-21 for slower, smoother response
Problem: Too Many Signals (Signal Spam)
Symptoms : Dashboard shows 50+ signals in Statistics, confidence scores mostly <0.40, signals clustering close together.
Solutions :
Raise Min Confidence :
Try 0.40-0.50 (quality filter)
Blocks bottom-tier setups
Targets top 50-60% of divergences only
Tighten OB/OS :
Use 70/30 or 75/25
Requires more extreme oscillator readings
Reduces false divergences in mid-range
Increase Min Slope Change :
Try 1.2-1.5 (from default 1.0)
Requires stronger, more obvious divergences
Filters marginal slope disagreements
Raise TCS Threshold :
Try 0.85-0.90 (from default 0.80)
Stricter trend filter blocks more counter-trend attempts
Favors only strongest trend alignment
Enable ALL CAE Gates :
Turn on Trend Filter + Adversarial + Confidence
Triple-layer protection
Blocks aggressively — expect 20-40% reduction in signals
Widen Spacing :
min_bars_ANY: 15-20 (from 12)
min_bars_SAME_SIDE: 30-40 (from 24)
Creates substantial breathing room
Switch to Confirmed Timing :
Removes realtime preview noise
Ensures full pivot validation
5-bar delay filters many false starts
Problem: Signals in Strong Trends Get Stopped Out
Symptoms : You take a bullish divergence in a downtrend (or bearish in uptrend), and it immediately fails. Dashboard showed high TCS at the time.
Analysis : This is INTENDED behavior — CAE is protecting you from low-probability counter-trend trades.
Understanding :
Check Last Signal Metrics in dashboard — what was TCS when signal fired?
If TCS was >0.85 and signal was counter-trend, CAE correctly identified it as high risk
Strong trends rarely reverse cleanly without major exhaustion
Your losses here are the system working as designed (blocking bad odds)
If You Want to Override (Not Recommended) :
Lower TCS_threshold to 0.70-0.75 (allows more counter-trend)
Lower exhaustion_required to 0.40 (easier override)
Disable Strong Trend Filter entirely (very risky)
Better Approach :
TRUST THE FILTER — it's preventing costly mistakes
Wait for exhaustion >0.75 (yellow shading) before counter-trending strong TCS
Focus on continuation signals (hidden divs) in high-TCS environments
Use Advisory mode to see what CAE is blocking and learn from outcomes
Problem: Adversarial Blocking Seems Wrong
Symptoms : You see a divergence that "looks good" visually, but CAE blocks with "Adversarial bearish/bullish" warning.
Diagnosis :
Check dashboard Bull Case and Bear Case scores at that moment
Look at Differential value
Check adversarial bar colors — was there strong coloring against your intended direction?
Understanding :
Adversarial catches "obvious" opposing momentum that's easy to miss
Example: Bullish divergence at a local low, BUT price is deeply below EMA50, bearish momentum is strong, and RSI shows knife-catching conditions
Bull Case might be 0.20 while Bear Case is 0.55
Differential = -0.35, far beyond threshold
Block is CORRECT — you'd be fighting overwhelming opposing flow
If You Disagree Consistently
Review blocked signals on chart — scroll back and check outcomes
Did those blocked signals actually work, or did they fail as adversarial predicted?
Raise adv_threshold to 0.15-0.20 (more permissive, allows closer battles)
Disable Adversarial Validation temporarily (diagnostic) to isolate its effect
Use Advisory mode to learn adversarial patterns over 50-100 signals
Remember : Adversarial is conservative BY DESIGN. It prevents "obvious" bad trades where you're fighting strong strength the other way.
Problem: Dashboard Not Showing or Incomplete
Solutions :
Toggle "Show Dashboard" to ON in settings
Try different dashboard sizes (Small/Normal/Large)
Try different positions (Top Left/Right, Bottom Left/Right) — might be off-screen
Some sections require CAE Enable = ON (Cognitive Engine section won't appear if CAE is disabled)
Statistics section requires at least 1 lifetime signal to populate
Check that visual theme is set (dashboard colors adapt to theme)
Problem: Performance Lag, Chart Freezing
Symptoms : Chart loading is slow, indicator calculations cause delays, pinch-to-zoom lags.
Diagnosis : Visual features are computationally expensive, especially adversarial bar coloring (recalculates every bar).
Solutions (In Order of Impact) :
Disable Adversarial Bar Coloring (MOST EXPENSIVE):
Turn OFF "Adversarial Bar Coloring" in settings
This is the single biggest performance drain
Immediate improvement
Reduce Vertical Lines :
Lower "Keep last N vertical lines" to 20-30
Or set to 0 to disable entirely
Moderate improvement
Disable Bifurcation Zones :
Turn OFF "Draw Bifurcation Zones"
Reduces box drawing calculations
Moderate improvement
Set Dashboard Size to Small :
Smaller dashboard = fewer cells = less rendering
Minor improvement
Use Shorter Max Lookback :
Reduce max_lookback to 40-50 (from 60+)
Fewer bars to scan for divergences
Minor improvement
Disable Exhaustion Shading :
Turn OFF "Show Market State"
Removes background coloring calculations
Minor improvement
Extreme Performance Mode :
Disable ALL visual enhancements
Keep only triangle markers
Dashboard Small or OFF
Use Minimal theme if available
Problem: Realtime Signals Repainting
Symptoms : You see a signal appear, but on next bar it disappears or moves.
Explanation :
Realtime mode detects peaks 1 bar ago: high > high AND high > high
On the FORMING bar (before close), this condition can change as new prices arrive
Example: At 10:05, high (10:04 bar) was 100, current high is 99 → peak detected
At 10:05:30, new high of 101 arrives → peak condition breaks → signal disappears
At 10:06 (bar close), final high is 101 → no peak at 10:04 anymore → signal gone permanently
This is expected behavior for realtime responsiveness. You get preview/early warning, but it's not locked until bar confirms.
Solutions :
Use Confirmed Timing :
Switch to "Confirmed (lookforward)" mode
ZERO repainting — pivot must be fully validated
5-bar delay (pivot_lookforward)
What you see in history is exactly what would have appeared live
Accept Realtime Repaint as Tradeoff :
Keep Realtime mode for speed and alerts
Understand that pre-confirmation signals may vanish
Only trade signals that CONFIRM at bar close (check barstate.isconfirmed)
Use for live monitoring, NOT for backtesting
Trade Only After Confirmation :
In Realtime mode, wait 1 full bar after signal appears before entering
If signal survives that bar close, it's locked
This adds 1-bar delay but removes repaint risk
Recommendation : Use Confirmed for backtesting and conservative trading. Use Realtime only for active monitoring with full understanding of preview behavior.
Risk Management Integration
BZ-CAE is a signal generation system, not a complete trading strategy. You must integrate proper risk management:
Position Sizing by Confidence
Confidence 0.70-1.00 (Premium) :
Risk: 1.5-2.0% of account (MAXIMUM)
Reasoning: High-quality setup across all factors
Still cap at 2% — even premium setups can fail
Confidence 0.50-0.70 (High Quality) :
Risk: 1.0-1.5% of account
Reasoning: Standard good setup
Your bread-and-butter risk level
Confidence 0.35-0.50 (Moderate Quality) :
Risk: 0.5-1.0% of account
Reasoning: Marginal setup, passes minimum threshold
Reduce size or skip if you're selective
Confidence <0.35 (Low Quality) :
Risk: 0% (blocked in Filtering mode)
Reasoning: Insufficient quality factors
System protects you by not showing these
Stop Placement Strategies
For Reversal Signals (Regular Divergences) :
Place stop beyond the divergence pivot plus buffer
Bullish : Stop below the divergence low - 1.0-1.5 × ATR
Bearish : Stop above the divergence high + 1.0-1.5 × ATR
Reasoning: If price breaks the pivot, divergence structure is invalidated
For Continuation Signals (Hidden Divergences) :
Place stop beyond recent swing in opposite direction
Bullish continuation : Stop below recent swing low (not the divergence pivot itself)
Bearish continuation : Stop above recent swing high
Reasoning: You're trading with trend, allow more breathing room
ATR-Based Stops :
1.5-2.0 × ATR is standard
Scale by timeframe:
Scalping (1-5m): 1.0-1.5 × ATR (tight)
Day trading (15m-1H): 1.5-2.0 × ATR (balanced)
Swing (4H-D): 2.0-3.0 × ATR (wide)
Never Use Fixed Dollar/Pip Stops :
Markets have different volatility
50-pip stop on EUR/USD ≠ 50-pip stop on GBP/JPY
Always normalize by ATR or pivot structure
Profit Targets and Scaling
Primary Target :
2-3 × ATR from entry (minimum 2:1 reward-risk)
Example : Entry at 100, ATR = 2, stop at 97 (1.5 × ATR) → target at 106 (3 × ATR) = 2:1 R:R
Scaling Out Strategy :
Take 50% off at 1.5 × ATR (secure partial profit)
Move stop to breakeven
Trail remaining 50% with 1.0 × ATR trailing stop
Let winners run if trend persists
Targets by Confidence :
High Confidence (>0.70) : Aggressive targets (3-4 × ATR), trail wider (1.5 × ATR)
Standard Confidence (0.50-0.70) : Normal targets (2-3 × ATR), standard trail (1.0 × ATR)
Low Confidence (0.35-0.50) : Conservative targets (1.5-2 × ATR), tight trail (0.75 × ATR)
Use Bifurcation Zones :
If opposite-side zone is visible on chart (from previous signal), use it as target
Example : Bullish signal at 100, prior supply zone at 110 → use 110 as target
Zones mark institutional resistance/support
Exhaustion-Based Exits :
If you're in a trade and exhaustion >0.75 develops (yellow shading), consider early exit
Market is overextended — reversal risk is high
Take profit even if target not reached
Trade Management by TCS
High TCS + Counter-Trend Trade (Risky) :
Use very tight stops (1.0-1.5 × ATR)
Conservative targets (1.5-2 × ATR)
Quick exit if trade doesn't work immediately
You're fading momentum — respect it
Low TCS + Reversal Trade (Safer) :
Use wider stops (2.0-2.5 × ATR)
Aggressive targets (3-4 × ATR)
Trail with patience
Genuine reversal potential in weak trend
High TCS + Continuation Trade (Safest) :
Standard stops (1.5-2.0 × ATR)
Very aggressive targets (4-5 × ATR)
Trail wide (1.5-2.0 × ATR)
You're with institutional momentum — let it run
Educational Value — Learning Machine Intelligence
BZ-CAE is designed as a learning platform, not just a tool:
Advisory Mode as Teacher
Most indicators are binary: signal or no signal. You don't learn WHY certain setups are better.
BZ-CAE's Advisory mode shows you EVERY potential divergence, then annotates the ones that would be blocked in Filtering mode with specific reasons:
"Bull: strong downtrend (TCS=0.87)" teaches you that TCS >0.85 makes counter-trend very risky
"Adversarial bearish" teaches you that the opposing case was dominating
"Low confidence 32%" teaches you that the setup lacked quality across multiple factors
"Bull spacing: wait 8 bars" teaches you that signals need breathing room
After 50-100 signals in Advisory mode, you internalize the CAE's decision logic. You start seeing these factors yourself BEFORE the indicator does.
Dashboard Transparency
Most "intelligent" indicators are black boxes — you don't know how they make decisions.
BZ-CAE shows you ALL metrics in real-time:
TCS tells you trend strength
DMA tells you momentum alignment
Exhaustion tells you overextension
Adversarial shows both sides of the debate
Confidence shows composite quality
You learn to interpret market state holistically, a skill applicable to ANY trading system beyond this indicator.
Divergence Quality Education
Not all divergences are equal. BZ-CAE teaches you which conditions produce high-probability setups:
Quality divergence : Regular bullish div at a low, TCS <0.50 (weak trend), exhaustion >0.75 (overextended), positive adversarial differential, confidence >0.70
Low-quality divergence : Regular bearish div at a high, TCS >0.85 (strong uptrend), exhaustion <0.30 (not overextended), negative adversarial differential, confidence <0.40
After using the system, you can evaluate divergences manually with similar intelligence.
Risk Management Discipline
Confidence-based position sizing teaches you to adjust risk based on setup quality, not emotions:
Beginners often size all trades identically
Or worse, size UP on marginal setups to "make up" for losses
BZ-CAE forces systematic sizing: premium setups get larger size, marginal setups get smaller size
This creates a probabilistic approach where your edge compounds over time.
What This Indicator Is NOT
Complete transparency about limitations and positioning:
Not a Prediction System
BZ-CAE does not predict future prices. It identifies structural divergences (price-momentum disagreements) and assesses current market state (trend, exhaustion, adversarial conditions). It tells you WHEN conditions favor a potential reversal or continuation, not WHAT WILL HAPPEN.
Markets are probabilistic. Even premium-confidence setups fail ~30-40% of the time. The system improves your probability distribution over many trades — it doesn't eliminate risk.
Not Fully Automated
This is a decision support tool, not a trading robot. You must:
Execute trades manually based on signals
Manage positions (stops, targets, trailing)
Apply discretionary judgment (news events, liquidity, context)
Integrate with your broader strategy and risk rules
The confidence scores guide position sizing, but YOU determine final risk allocation based on your account size, risk tolerance, and portfolio context.
Not Beginner-Friendly
BZ-CAE requires understanding of:
Divergence trading concepts (regular vs hidden, reversal vs continuation)
Market state interpretation (trend vs range, momentum, exhaustion)
Basic technical analysis (pivots, support/resistance, EMAs)
Risk management fundamentals (position sizing, stops, R:R)
This is designed for intermediate to advanced traders willing to invest time learning the system. If you want "buy the arrow" simplicity, this isn't the tool.
Not a Holy Grail
There is no perfect indicator. BZ-CAE filters noise and improves signal quality significantly, but:
Losing trades are inevitable (even at 70% win rate, 30% still fail)
Market conditions change rapidly (yesterday's strong trend becomes today's chop)
Black swan events occur (fundamentals override technicals)
Execution matters (slippage, fees, emotional discipline)
The system provides an EDGE, not a guarantee. Your job is to execute that edge consistently with proper risk management over hundreds of trades.
Not Financial Advice
BZ-CAE is an educational and analytical tool. All trading decisions are your responsibility. Past performance (backtested or live) does not guarantee future results. Only risk capital you can afford to lose. Consult a licensed financial advisor for investment advice specific to your situation.
Ideal Market Conditions
Best Performance Characteristics
Liquid Instruments :
Major forex pairs (EUR/USD, GBP/USD, USD/JPY)
Large-cap stocks and index ETFs (SPY, QQQ, AAPL, MSFT)
High-volume crypto (BTC, ETH)
Major commodities (Gold, Oil, Natural Gas)
Reasoning: Clean price structure, clear pivots, meaningful oscillator behavior
Trending with Consolidations :
Markets that trend for 20-40 bars, then consolidate 10-20 bars, repeat
Creates divergences at consolidation boundaries (reversals) and within trends (continuations)
Both regular and hidden divs find opportunities
5-Minute to Daily Timeframes :
Below 5m: too much noise, false pivots, CAE metrics unstable
Above daily: too few signals, edge diminishes (fundamentals dominate)
Sweet spot: 15m to 4H for most traders
Consistent Volume and Participation :
Regular trading sessions (not holidays or thin markets)
Predictable volatility patterns
Avoid instruments with sudden gaps or circuit breakers
Challenging Conditions
Extremely Low Liquidity :
Penny stocks, exotic forex pairs, low-volume crypto
Erratic pivots, unreliable oscillator readings
CAE metrics can't assess market state properly
Very Low Timeframes (1-Minute or Below) :
Dominated by market microstructure noise
Divergences are everywhere but meaningless
CAE filtering helps but still unreliable
Extended Sideways Consolidation :
100+ bars of tight range with no clear pivots
Oscillator hugs midpoint (45-55 range)
No divergences to detect
Fundamentally-Driven Gap Markets :
Earnings releases, economic data, geopolitical events
Price gaps over stops and targets
Technical structure breaks down
Recommendation: Disable trading around known events
Calculation Methodology — Technical Depth
For users who want to understand the math:
Oscillator Computation
Each oscillator type calculates differently, but all normalize to 0-100:
RSI : ta.rsi(close, length) — Standard Relative Strength Index
Stochastic : ta.stoch(high, low, close, length) — %K calculation
CCI : (ta.cci(hlc3, length) + 100) / 2 — Normalized from -100/+100 to 0-100
MFI : ta.mfi(hlc3, length) — Volume-weighted RSI equivalent
Williams %R : ta.wpr(length) + 100 — Inverted stochastic adjusted to 0-100
Smoothing: If smoothing > 1, apply ta.sma(oscillator, smoothing)
Divergence Detection Algorithm
Identify Pivots :
Price high pivot: ta.pivothigh(high, lookback, lookforward)
Price low pivot: ta.pivotlow(low, lookback, lookforward)
Oscillator high pivot: ta.pivothigh(osc, lookback, lookforward)
Oscillator low pivot: ta.pivotlow(osc, lookback, lookforward)
Store Recent Pivots :
Maintain arrays of last 10 pivots with bar indices
When new pivot confirmed, unshift to array, pop oldest if >10
Scan for Slope Disagreements :
Loop through last 5 pivots
For each pair (current pivot, historical pivot):
Check if within max_lookback bars
Calculate slopes: (current - historical) / bars_between
Regular bearish: price_slope > 0, osc_slope < 0, |osc_slope| > min_threshold
Regular bullish: price_slope < 0, osc_slope > 0, |osc_slope| > min_threshold
Hidden bearish: price_slope < 0, osc_slope > 0, osc_slope > min_threshold
Hidden bullish: price_slope > 0, osc_slope < 0, |osc_slope| > min_threshold
Important Disclaimers and Terms
Performance Disclosure
Past performance, whether backtested or live-traded, does not guarantee future results. Markets change. What works today may not work tomorrow. Hypothetical or simulated performance results have inherent limitations and do not represent actual trading.
Risk of Loss
Trading involves substantial risk of loss. Only trade with risk capital you can afford to lose entirely. The high degree of leverage often available in trading can work against you as well as for you. Leveraged trading may result in losses exceeding your initial deposit.
Not Financial Advice
BZ-CAE is an educational and analytical tool for technical analysis. It is not financial advice, investment advice, or a recommendation to buy or sell any security or instrument. All trading decisions are your sole responsibility. Consult a licensed financial advisor for advice specific to your circumstances.
Technical Indicator Limitations
BZ-CAE is a technical analysis tool based on price and volume data. It does not account for:
Fundamental analysis (earnings, economic data, financial health)
Market sentiment and positioning
Geopolitical events and news
Liquidity conditions and market microstructure changes
Regulatory changes or exchange rules
Integrate with broader analysis and strategy. Do not rely solely on technical indicators for trading decisions.
Repainting Acknowledgment
As disclosed throughout this documentation:
Realtime mode may repaint on forming bars before confirmation (by design for preview functionality)
Confirmed mode has zero repainting (fully validated pivots only)
Choose timing mode appropriate for your use case. Understand the tradeoffs.
Testing Recommendation
ALWAYS test on demo/paper accounts before committing real capital. Validate the indicator's behavior on your specific instruments and timeframes. Learn the system thoroughly in Advisory mode before using Filtering mode.
Learning Resources :
In-indicator tooltips (hover over setting names for detailed explanations)
This comprehensive publishing statement (save for reference)
User guide in script comments (top of code)
Final Word — Philosophy of BZ-CAE
BZ-CAE is not designed to replace your judgment — it's designed to enhance it.
The indicator identifies structural inflection points (bifurcations) where price and momentum disagree. The Cognitive Engine evaluates market state to determine if this disagreement is meaningful or noise. The Adversarial model debates both sides of the trade to catch obvious bad setups. The Confidence system ranks quality so you can choose your risk appetite.
But YOU still execute. YOU still manage risk. YOU still learn from outcomes.
This is intelligence amplification, not intelligence replacement.
Use Advisory mode to learn how expert traders evaluate market state. Use Filtering mode to enforce discipline when emotions run high. Use the dashboard to develop a systematic approach to reading markets. Use confidence scores to size positions probabilistically.
The system provides an edge. Your job is to execute that edge with discipline, patience, and proper risk management over hundreds of trades.
Markets are probabilistic. No system wins every trade. But a systematic edge + disciplined execution + proper risk management compounds over time. That's the path to consistent profitability. BZ-CAE gives you the edge. The discipline and risk management are on you.
Taking you to school. — Dskyz, Trade with insight. Trade with anticipation.
BO - RSI - M5 BacktestingBO - RSI - M5 Backtesting -Rule of Strategy
A. Data
1. Chart M5 IDC
2. Symbol: EURJPY
B. Indicator
1. RSI
2. Length: 12 (adjustable)
3. Extreme Top: 75 (adjustable)
4. Extreme Bottom: 25 (adjustable)
C. Rule of Signal
1. Put Signal
* Rsi create a temporary peak over Extreme Top
row61: peak_rsi= rsi >rsi and rsi >rsi and rsi rsi_top
2. Call Signal
* Rsi create a temporary bottom under Extreme Bottom
row62: bott_rsi= rsi rsi and rsi
Inflection PointInflection Point - The Adaptive Confluence Reversal Engine
This is not just another peak and valley indicator; it is a complete and total reimagining of how market turning points are detected, qualified, and acted upon. Born from the foundational concepts explored in systems like my earlier creation, DAFE - Turning Point, Inflection Point is a ground-up engineering feat designed for the modern trader. It moves beyond static rules and simple pattern recognition into the realm of dynamic, multi-factor confluence analysis and adaptive machine learning.
Where other indicators provide a guess, Inflection Point provides a probability. It meticulously analyzes the market's deepest currents—momentum, exhaustion, and reversal velocity—and fuses them into a single, unified "Confluence Score." This is not a simple combination of indicators; it is an intelligent, weighted system where each component works in concert, creating an analytical engine that is orders of magnitude more sophisticated and reliable than any standard reversal tool.
Furthermore, Inflection Point learns. Through its advanced Adaptive Learning Engine, it constantly monitors its own performance, adjusting its confidence and selectivity in real-time based on its recent success rate. This allows it to adapt its behavior to any security, on any timeframe, with remarkable success.
Theoretical Foundation - Confluence Core
Inflection Point's predictive power does not come from a single, magical formula. It comes from the intelligent synthesis of three critical market phenomena, weighted and scored in real-time to generate a single, high-conviction probability rating.
1. Factor One: Pre-Reversal Momentum State (RSI Analysis)
Instead of reacting to a simple RSI cross, Inflection Point proactively scans for the build-up of momentum that precedes a reversal.
• Formulaic Concept: It measures the highest RSI value over a lookback period for peaks and the lowest RSI for valleys. A signal is only considered valid if significant momentum has been established before the turn, indicating a stretched market condition ripe for reversal.
• Asymmetric Sophistication: The engine uses different, optimized thresholds for bull and bear momentum, recognizing that markets often fall faster than they rise.
2. Factor Two: Volatility Exhaustion (Bollinger Band Analysis)
A true reversal often occurs when price makes a final, exhaustive push into unsustainable territory.
• Formulaic Concept: The engine detects when price has significantly pierced the outer Bollinger Bands. This is not just a touch, but a statistical deviation from the mean that signals volatility exhaustion, where the energy for the current move is likely depleted.
3. Factor Three: Reversal Strength (Rate of Change Analysis)
The character of a reversal matters. A sharp, decisive turn is more significant than a slow, meandering one.
• Formulaic Concept: Using a short-term Rate of Change (ROC), the engine measures the velocity of the reversal itself. A higher ROC score adds significant weight to the final probability, confirming that the new direction has conviction.
4. The Final Calculation: The Adaptive Learning Engine
This is the system's "brain." It maintains a history of its past signals and calculates its real-time win rate. This hitRate is then used to generate an adaptiveMultiplier.
• Self-Correction: In "Quality Control" mode, a high win rate makes the indicator more selective, demanding a higher probability score to issue a signal, thereby protecting streaks. A lower win rate makes it slightly less selective to ensure it continues learning from new market conditions.
• The result is a system that is not static, but a living, breathing tool that adapts its personality to the unique rhythm of any chart.
Why Inflection Point is a Paradigm Shift
Inflection Point is fundamentally different from other reversal indicators for three key reasons:
Confluence Over Isolation: Standard indicators look at one thing (e.g., RSI > 70). Inflection Point simultaneously analyzes momentum, volatility, and velocity, understanding that true reversals are a product of multiple converging factors. It answers not just "if," but "why" a reversal is likely.
Probabilistic Over Binary: Other tools give you a simple "yes" or "no." Inflection Point provides a probability score from 0-100, allowing you to gauge the conviction of every potential signal. This empowers you to differentiate between a weak setup and an A+ opportunity.
Adaptive Over Static: Every other indicator uses the same rules forever. Inflection Point's Adaptive Engine means it is constantly refining its own logic based on what is actually working in the current market, on the specific asset you are trading. It is tailored to the now.
The Inputs Menu - Your Command Center
Every setting is a lever of control, allowing you to tune the engine to your precise trading style and market focus.
🧠 Neural Core Engine
Analysis Depth: This is the primary lookback for the Bollinger Band and other core calculations. A shorter depth makes the indicator faster and more sensitive, ideal for scalping. A longer depth makes it slower and more stable, ideal for swing trading.
Minimum Probability %: This is your master signal filter. It sets the minimum Confluence Score required to plot a signal. Higher values (85-95) will give you only the highest-conviction A+ setups. Lower values (70-80) will show more potential opportunities.
🤖 Adaptive Neural Learning
Enable Adaptive Learning Engine: Toggles the entire learning system. Disabling it will make the indicator's logic static.
Peak/Valley Success Threshold (ATR): This defines what constitutes a "successful" trade for the learning engine. A value of 1.5 means price must move 1.5x the ATR in your favor for the signal to be marked as a win. Adjust this to match your personal take-profit strategy.
Adaptive Mode: This dictates how the engine uses its hitRate. "Quality Control" is recommended for its intelligent filtering. "Aggressive" will always boost signal scores, useful for finding more setups in a known, trending environment.
Asymmetric Balance: Allows you to apply a "boost" to either peak (short) or valley (long) signals. If you find the market you're trading has stronger long reversals, you can increase the "Valley Signal Boost" to catch them more effectively.
🛡️ Elite Filters
Market Noise Filter: An exceptional tool for avoiding choppy markets. It counts the number of directional changes in the last 5 bars. If the market is whipping back and forth too much, it will block the signal. Lower the "Max Direction Changes" to be extremely selective.
Volume Filter: Requires signal confirmation from a significant volume spike. The "Volume Multiplier" dictates how large this spike must be (e.g., 1.2 = 20% above average volume). This is invaluable for filtering out low-conviction moves in stocks and crypto.
The Dashboard - Your Analytical Co-Pilot
The dashboard is not just a set of numbers; it is a holistic overview of the market's health and the engine's current state.
Unified AI Score: This section provides the most critical, at-a-glance information. "Total Score" is the current probability reading, while "Quality" gives you a human-readable interpretation. "Win Rate" shows the real-time performance of the Adaptive Engine.
Order Flow (OFPI): This measures the "weight" of money behind recent price moves by analyzing price change relative to volume. A high positive OFPI suggests strong buying pressure, while a high negative value suggests strong selling pressure. It gives you a peek into the market's underlying flow.
Component Analysis: This allows you to see the individual "Peak" and "Valley" confidence scores before they are filtered, giving you insight into building momentum before a signal forms.
Market Structure: This panel assesses the broader environment. "HTF Trend" tells you the direction of the larger trend (based on EMAs), while "Vol Regime" tells you if the market is in a high, medium, or low volatility state. Use this to align your signals with the broader market context.
Filter & Engine Statistics: Available on the "Large" dashboard, this provides deep insight into how many signals are being blocked by your filters and the current status of the Adaptive Engine's multiplier.
The Visual Interface - A Symphony of Data
Every visual element on the chart is designed for instant interpretation and insight.
Signal Markers: Simple, clean triangles mark the exact bar of a valid signal. A box is drawn around the high/low of the signal bar to highlight the precise point of inflection.
Dynamic Support/Resistance Zones: These are the glowing lines on your chart. They are not static lines; they are dynamic levels that represent the current battlefield between buyers and sellers.
Cyber Cyan (Valley Blue): This is the current Support Zone. This is the price level the market is currently trying to defend.
Neural Pink (Peak Red): This is the current Resistance Zone. This is the price level the market is currently trying to break through.
Grey (Next Level): This line is a projection, based on the current momentum and the size of the S/R range, of where the next major level of conflict will likely be. It acts as a potential price target.
Development & Philosophy
Inflection Point was not assembled; it was engineered. It represents hundreds of hours of research into market dynamics, statistical analysis, and machine learning principles. The goal was to create a tool that moves beyond the limitations of traditional technical analysis, which often fails in modern, algorithm-driven markets. By building a system based on multi-factor confluence and self-adaptive logic, Inflection Point provides a quantifiable, statistical edge that is simply unattainable with simpler tools. This is the result of a relentless pursuit of a better, more intelligent way to trade.
Universal Applicability
The principles of momentum, exhaustion, and velocity are universal to all freely traded markets. Because of its adaptive core and robust filtering options, Inflection Point has proven to be exceptionally effective on any security (stocks, crypto, forex, indices, futures) and on any timeframe (from 1-minute scalping charts to daily swing trading charts).
" Markets are constantly in a state of uncertainty and flux and money is made by discounting the obvious and betting on the unexpected. "
— George Soros
Trade with insight. Trade with anticipation.
— Dskyz, for DAFE Trading Systems
MLLossFunctionsLibrary "MLLossFunctions"
Methods for Loss functions.
mse(expects, predicts) Mean Squared Error (MSE) " MSE = 1/N * sum ((y - y')^2) ".
Parameters:
expects : float array, expected values.
predicts : float array, prediction values.
Returns: float
binary_cross_entropy(expects, predicts) Binary Cross-Entropy Loss (log).
Parameters:
expects : float array, expected values.
predicts : float array, prediction values.
Returns: float
B3 HL2 Method Candle PainterThis script is similar to the "Hi-Lo" or "Clear" methods of painting bars. Instead of using the tips/edges of the candles like those two, the "(H+L)/2" method uses the change in (high+low)/2 to paint the bars. This gives you some similar results if you were to be binary with the candle coloring. However, my coloring scheme is not entirely binary. There are 5 possible colors:
HL2>LastHigh = Bright Green
HL2LastHL2 = Dull Green
HL2
DataCorrelationLibrary "DataCorrelation"
Implementation of functions related to data correlation calculations. Formulas have been transformed in such a way that we avoid running loops and instead make use of time series to gradually build the data we need to perform calculation. This allows the calculations to run on unbound series, and/or higher number of samples
🎲 Simplifying Covariance
Original Formula
//For Sample
Covₓᵧ = ∑ ((xᵢ-x̄)(yᵢ-ȳ)) / (n-1)
//For Population
Covₓᵧ = ∑ ((xᵢ-x̄)(yᵢ-ȳ)) / n
Now, if we look at numerator, this can be simplified as follows
∑ ((xᵢ-x̄)(yᵢ-ȳ))
=> (x₁-x̄)(y₁-ȳ) + (x₂-x̄)(y₂-ȳ) + (x₃-x̄)(y₃-ȳ) ... + (xₙ-x̄)(yₙ-ȳ)
=> (x₁y₁ + x̄ȳ - x₁ȳ - y₁x̄) + (x₂y₂ + x̄ȳ - x₂ȳ - y₂x̄) + (x₃y₃ + x̄ȳ - x₃ȳ - y₃x̄) ... + (xₙyₙ + x̄ȳ - xₙȳ - yₙx̄)
=> (x₁y₁ + x₂y₂ + x₃y₃ ... + xₙyₙ) + (x̄ȳ + x̄ȳ + x̄ȳ ... + x̄ȳ) - (x₁ȳ + x₂ȳ + x₃ȳ ... xₙȳ) - (y₁x̄ + y₂x̄ + y₃x̄ + yₙx̄)
=> ∑xᵢyᵢ + n(x̄ȳ) - ȳ∑xᵢ - x̄∑yᵢ
So, overall formula can be simplified to be used in pine as
//For Sample
Covₓᵧ = (∑xᵢyᵢ + n(x̄ȳ) - ȳ∑xᵢ - x̄∑yᵢ) / (n-1)
//For Population
Covₓᵧ = (∑xᵢyᵢ + n(x̄ȳ) - ȳ∑xᵢ - x̄∑yᵢ) / n
🎲 Simplifying Standard Deviation
Original Formula
//For Sample
σ = √(∑(xᵢ-x̄)² / (n-1))
//For Population
σ = √(∑(xᵢ-x̄)² / n)
Now, if we look at numerator within square root
∑(xᵢ-x̄)²
=> (x₁² + x̄² - 2x₁x̄) + (x₂² + x̄² - 2x₂x̄) + (x₃² + x̄² - 2x₃x̄) ... + (xₙ² + x̄² - 2xₙx̄)
=> (x₁² + x₂² + x₃² ... + xₙ²) + (x̄² + x̄² + x̄² ... + x̄²) - (2x₁x̄ + 2x₂x̄ + 2x₃x̄ ... + 2xₙx̄)
=> ∑xᵢ² + nx̄² - 2x̄∑xᵢ
=> ∑xᵢ² + x̄(nx̄ - 2∑xᵢ)
So, overall formula can be simplified to be used in pine as
//For Sample
σ = √(∑xᵢ² + x̄(nx̄ - 2∑xᵢ) / (n-1))
//For Population
σ = √(∑xᵢ² + x̄(nx̄ - 2∑xᵢ) / n)
🎲 Using BinaryInsertionSort library
Chatterjee Correlation and Spearman Correlation functions make use of BinaryInsertionSort library to speed up sorting. The library in turn implements mechanism to insert values into sorted order so that load on sorting is reduced by higher extent allowing the functions to work on higher sample size.
🎲 Function Documentation
chatterjeeCorrelation(x, y, sampleSize, plotSize)
Calculates chatterjee correlation between two series. Formula is - ξnₓᵧ = 1 - (3 * ∑ |rᵢ₊₁ - rᵢ|)/ (n²-1)
Parameters:
x : First series for which correlation need to be calculated
y : Second series for which correlation need to be calculated
sampleSize : number of samples to be considered for calculattion of correlation. Default is 20000
plotSize : How many historical values need to be plotted on chart.
Returns: float correlation - Chatterjee correlation value if falls within plotSize, else returns na
spearmanCorrelation(x, y, sampleSize, plotSize)
Calculates spearman correlation between two series. Formula is - ρ = 1 - (6∑dᵢ²/n(n²-1))
Parameters:
x : First series for which correlation need to be calculated
y : Second series for which correlation need to be calculated
sampleSize : number of samples to be considered for calculattion of correlation. Default is 20000
plotSize : How many historical values need to be plotted on chart.
Returns: float correlation - Spearman correlation value if falls within plotSize, else returns na
covariance(x, y, include, biased)
Calculates covariance between two series of unbound length. Formula is Covₓᵧ = ∑ ((xᵢ-x̄)(yᵢ-ȳ)) / (n-1) for sample and Covₓᵧ = ∑ ((xᵢ-x̄)(yᵢ-ȳ)) / n for population
Parameters:
x : First series for which covariance need to be calculated
y : Second series for which covariance need to be calculated
include : boolean flag used for selectively including sample
biased : boolean flag representing population covariance instead of sample covariance
Returns: float covariance - covariance of selective samples of two series x, y
stddev(x, include, biased)
Calculates Standard Deviation of a series. Formula is σ = √( ∑(xᵢ-x̄)² / n ) for sample and σ = √( ∑(xᵢ-x̄)² / (n-1) ) for population
Parameters:
x : Series for which Standard Deviation need to be calculated
include : boolean flag used for selectively including sample
biased : boolean flag representing population covariance instead of sample covariance
Returns: float stddev - standard deviation of selective samples of series x
correlation(x, y, include)
Calculates pearson correlation between two series of unbound length. Formula is r = Covₓᵧ / σₓσᵧ
Parameters:
x : First series for which correlation need to be calculated
y : Second series for which correlation need to be calculated
include : boolean flag used for selectively including sample
Returns: float correlation - correlation between selective samples of two series x, y
Wick Detection (1 and 0) - AYNETDetailed Scientific Explanation
1. Wick Detection Logic
Definition of a Wick:
A wick, also known as a shadow, represents the price action outside the range of a candlestick's body (the region between open and close).
Upper Wick: Occurs when the high value exceeds the greater of open and close.
Lower Wick: Occurs when the low value is lower than the smaller of open and close.
Upper Wick Detection:
pinescript
Kodu kopyala
bool has_upper_wick = high > math.max(open, close)
This checks if the high price of the candle is greater than the maximum of the open and close prices. If true, an upper wick exists.
Lower Wick Detection:
pinescript
Kodu kopyala
bool has_lower_wick = low < math.min(open, close)
This checks if the low price of the candle is less than the minimum of the open and close prices. If true, a lower wick exists.
2. Binary Representation
The presence of a wick is encoded as a binary value for simplicity and computational analysis:
Upper Wick: Represented as 1 if present, otherwise 0.
pinescript
Kodu kopyala
float upper_wick_binary = has_upper_wick ? 1 : 0
Lower Wick: Represented as 1 if present, otherwise 0. This value is inverted (-1) for visualization purposes.
pinescript
Kodu kopyala
float lower_wick_binary = has_lower_wick ? 1 : 0
3. Visualization with Histograms
The plot function is used to create histograms for visualizing the binary wick data:
Upper Wicks: Plotted as positive values with green columns:
pinescript
Kodu kopyala
plot(upper_wick_binary, title="Upper Wick", color=color.new(color.green, 0), style=plot.style_columns, linewidth=2)
Lower Wicks: Plotted as negative values with red columns:
pinescript
Kodu kopyala
plot(lower_wick_binary * -1, title="Lower Wick", color=color.new(color.red, 0), style=plot.style_columns, linewidth=2)
Features and Applications
1. Wick Visualization:
Upper wicks are displayed as positive green columns.
Lower wicks are displayed as negative red columns.
This provides a clear visual representation of wick presence in historical data.
2. Technical Analysis:
Wick formations often indicate market sentiment:
Upper Wicks: Sellers pushed the price lower after buyers drove it higher, signaling rejection at the top.
Lower Wicks: Buyers pushed the price higher after sellers drove it lower, signaling rejection at the bottom.
3. Signal Generation:
Traders can use wick detection to build strategies, such as identifying key price levels or market reversals.
Enhancements and Future Improvements
1. Wick Length Measurement
Instead of binary detection, measure the actual length of the wick:
pinescript
Kodu kopyala
float upper_wick_length = high - math.max(open, close)
float lower_wick_length = math.min(open, close) - low
This approach allows for thresholds to identify significant wicks:
pinescript
Kodu kopyala
bool significant_upper_wick = upper_wick_length > 10 // For wicks longer than 10 units.
bool significant_lower_wick = lower_wick_length > 10
2. Alerts for Long Wicks
Trigger alerts when significant wicks are detected:
pinescript
Kodu kopyala
alertcondition(significant_upper_wick, title="Long Upper Wick", message="A significant upper wick has been detected.")
alertcondition(significant_lower_wick, title="Long Lower Wick", message="A significant lower wick has been detected.")
3. Combined Wick Analysis
Analyze both upper and lower wicks to assess volatility:
pinescript
Kodu kopyala
float total_wick_length = upper_wick_length + lower_wick_length
bool high_volatility = total_wick_length > 20 // Combined wick length exceeds 20 units.
Conclusion
This script provides a compact and computationally efficient way to detect candlestick wicks and represent them as binary data. By visualizing the data with histograms, traders can easily identify wick formations and use them for technical analysis, signal generation, and volatility assessment. The approach can be extended further to measure wick length, detect significant wicks, and integrate these insights into automated trading systems.
Zero-lag Volatility-Breakout EMA Trend StrategyThis is a simple volatility-breakout strategy which uses the difference in two different zero-lag* EMAs (explained below on what exactly I mean by this) to track the upwards or downwards strength of an instrument. When the difference breaks above a Bollinger Band of a configurable standard deviation multiple, the strategy enters based off the direction of the base EMA used (i.e. if the difference breaks above and the current EMA is rising, a long entry is produced. If the difference breaks above and the current EMA is falling, a short entry is produced).
The two EMA-type metrics used to calculate the volatility difference are calculated by the following formula:
top_ema = math.max(src, ta.ema(src, length))
bottom_ema = math.min(src, ta.ema(src, length))
ema_difference = (top_ema - bottom_ema) - 1
This produces a difference which responds immediately to large price movements, instead of lagging if it used strictly the EMA itself.
SETTINGS
Source : The source of the strategy - close, hlc3, another indicator plot, etc.
EMA Difference Length : The length of both the EMA difference statistics and the base EMA used to calculate the entry side.
Standard Deviation Multiple : The Bollinger Bands multiple used when the difference is breaking out.
Use Binary Strategy : The strategy has two configurations: Binary and Rapid-Exit. 'Binary' means that it will not close a long position until a short position is generated, and vice-versa. 'Rapid-Exit' will close a long or short position once the difference reaches the middle Bollinger Band MA. This means that turning on 'Binary' will expose you to more market risk, but potentially greater market return. Turning off 'Binary' will exit quickly and reduce drawdown.
The strategy results below use 10% equity and 0.1% fees per trade.
Qualitative Smoothed Strength Index***RSI CHART BELOW IS FOR COMPARSION TO SHOW HOE THEY MAKE SIMILIAR PATTERNS*** IT IS NOT PART OF THE INDICATOR***
The Qualitative Smoothed Strength Index (QSSI) is a simplified momentum oscillator whose values will oscillate between 0 and 1 . By converting price differences into binary values and smoothing them with a moving average, it identifies qualitative strength of price movements. This simplification allows traders to easily interpret trends and reversals. The QSSI offers advantages such as noise reduction, clear trend identification, and early signal detection, resulting in less lag compared to traditional oscillators. Traders can customize the indicator based on their preferences and use it across various markets.
QSSI Indicator uses the input function is used to define the input parameters of the indicator. In this case, there are two inputs:
length: The number of periods used for calculating the differences (a, b, c) and their assigned values. Default value is 5.
MAL: The length of the moving average used for smoothing the assigned values. Default value is 14.
The next few lines calculate 'a', 'b', and 'c', which represent the differences between the high, low, and close prices, respectively, and their corresponding previous simple moving averages (SMAs) of specified length. These differences are used to identify price movements.
The code assigns binary values (0 or 1) to a_assigned, b_assigned, and c_assigned, depending on whether the corresponding differences (a, b, c) are greater than 0. This step converts the differences into a binary representation, indicating upward or downward price movements.
Average_assigned calculates the average of the assigned binary values of a, b, and c. This average value represents the overall strength of the price movement.ma_assigned calculates the 14-day moving average of average_assigned, which smoothens the indicator and helps traders identify trends more easily.
The code plots the 14-day moving average (ma_assigned) on the chart as a blue line. It also plots the individual assigned values of a, b, and c as dots on the chart. a_assigned is shown in green, b_assigned in red, and c_assigned in black. These dots indicate the presence of upward or downward movements in the respective price components. By visualizing these dots on the chart, the trader can quickly identify the presence and direction of price movements for each of the price components. This information can be valuable for understanding how the different price elements (high, low, and close) are contributing to the overall trend and strength of the market. Traders can use this data to make more informed decisions, such as confirming the presence of trends, identifying potential reversals, or gauging the overall market sentiment based on the distribution of upward and downward movements across the price components.
Finally, the code draws horizontal dotted lines at levels 0.70 (0.8)and 0.30 (0.2). These levels are typically used to identify overbought (above 0.70 or 0.8) and oversold (below 0.30 or 0.2) conditions in the market.
The Qualitative Smoothed Strength Index (QSSI) provides traders with information about the strength and direction of price movements. By using assigned binary values, the indicator simplifies the interpretation of price data, making it easier to identify trends and potential reversals.
Historical Matrix Analyzer [PhenLabs]📊Historical Matrix Analyzer
Version: PineScriptv6
📌Description
The Historical Matrix Analyzer is an advanced probabilistic trading tool that transforms technical analysis into a data-driven decision support system. By creating a comprehensive 56-cell matrix that tracks every combination of RSI states and multi-indicator conditions, this indicator reveals which market patterns have historically led to profitable outcomes and which have not.
At its core, the indicator continuously monitors seven distinct RSI states (ranging from Extreme Oversold to Extreme Overbought) and eight unique indicator combinations (MACD direction, volume levels, and price momentum). For each of these 56 possible market states, the system calculates average forward returns, win rates, and occurrence counts based on your configurable lookback period. The result is a color-coded probability matrix that shows you exactly where you stand in the historical performance landscape.
The standout feature is the Current State Panel, which provides instant clarity on your active market conditions. This panel displays signal strength classifications (from Strong Bullish to Strong Bearish), the average return percentage for similar past occurrences, an estimated win rate using Bayesian smoothing to prevent small-sample distortions, and a confidence level indicator that warns you when insufficient data exists for reliable conclusions.
🚀Points of Innovation
Multi-dimensional state classification combining 7 RSI levels with 8 indicator combinations for 56 unique trackable market conditions
Bayesian win rate estimation with adjustable smoothing strength to provide stable probability estimates even with limited historical samples
Real-time active cell highlighting with “NOW” marker that visually connects current market conditions to their historical performance data
Configurable color intensity sensitivity allowing traders to adjust heat-map responsiveness from conservative to aggressive visual feedback
Dual-panel display system separating the comprehensive statistics matrix from an easy-to-read current state summary panel
Intelligent confidence scoring that automatically warns traders when occurrence counts fall below reliable thresholds
🔧Core Components
RSI State Classification: Segments RSI readings into 7 distinct zones (Extreme Oversold <20, Oversold 20-30, Weak 30-40, Neutral 40-60, Strong 60-70, Overbought 70-80, Extreme Overbought >80) to capture momentum extremes and transitions
Multi-Indicator Condition Tracking: Simultaneously monitors MACD crossover status (bullish/bearish), volume relative to moving average (high/low), and price direction (rising/falling) creating 8 binary-encoded combinations
Historical Data Storage Arrays: Maintains rolling lookback windows storing RSI states, indicator states, prices, and bar indices for precise forward-return calculations
Forward Performance Calculator: Measures price changes over configurable forward bar periods (1-20 bars) from each historical state, accumulating total returns and win counts per matrix cell
Bayesian Smoothing Engine: Applies statistical prior assumptions (default 50% win rate) weighted by user-defined strength parameter to stabilize estimated win rates when sample sizes are small
Dynamic Color Mapping System: Converts average returns into color-coded heat map with intensity adjusted by sensitivity parameter and transparency modified by confidence levels
🔥Key Features
56-Cell Probability Matrix: Comprehensive grid displaying every possible combination of RSI state and indicator condition, with each cell showing average return percentage, estimated win rate, and occurrence count for complete statistical visibility
Current State Info Panel: Dedicated display showing your exact position in the matrix with signal strength emoji indicators, numerical statistics, and color-coded confidence warnings for immediate situational awareness
Customizable Lookback Period: Adjustable historical window from 50 to 500 bars allowing traders to focus on recent market behavior or capture longer-term pattern stability across different market cycles
Configurable Forward Performance Window: Select target holding periods from 1 to 20 bars ahead to align probability calculations with your trading timeframe, whether day trading or swing trading
Visual Heat Mapping: Color-coded cells transition from red (bearish historical performance) through gray (neutral) to green (bullish performance) with intensity reflecting statistical significance and occurrence frequency
Intelligent Data Filtering: Minimum occurrence threshold (1-10) removes unreliable patterns with insufficient historical samples, displaying gray warning colors for low-confidence cells
Flexible Layout Options: Independent positioning of statistics matrix and info panel to any screen corner, accommodating different chart layouts and personal preferences
Tooltip Details: Hover over any matrix cell to see full RSI label, complete indicator status description, precise average return, estimated win rate, and total occurrence count
🎨Visualization
Statistics Matrix Table: A 9-column by 8-row grid with RSI states labeling vertical axis and indicator combinations on horizontal axis, using compact abbreviations (XOverS, OverB, MACD↑, Vol↓, P↑) for space efficiency
Active Cell Indicator: The current market state cell displays “⦿ NOW ⦿” in yellow text with enhanced color saturation to immediately draw attention to relevant historical performance
Signal Strength Visualization: Info panel uses emoji indicators (🔥 Strong Bullish, ✅ Bullish, ↗️ Weak Bullish, ➖ Neutral, ↘️ Weak Bearish, ⛔ Bearish, ❄️ Strong Bearish, ⚠️ Insufficient Data) for rapid interpretation
Histogram Plot: Below the price chart, a green/red histogram displays the current cell’s average return percentage, providing a time-series view of how historical performance changes as market conditions evolve
Color Intensity Scaling: Cell background transparency and saturation dynamically adjust based on both the magnitude of average returns and the occurrence count, ensuring visual emphasis on reliable patterns
Confidence Level Display: Info panel bottom row shows “High Confidence” (green), “Medium Confidence” (orange), or “Low Confidence” (red) based on occurrence counts relative to minimum threshold multipliers
📖Usage Guidelines
RSI Period
Default: 14
Range: 1 to unlimited
Description: Controls the lookback period for RSI momentum calculation. Standard 14-period provides widely-recognized overbought/oversold levels. Decrease for faster, more sensitive RSI reactions suitable for scalping. Increase (21, 28) for smoother, longer-term momentum assessment in swing trading. Changes affect how quickly the indicator moves between the 7 RSI state classifications.
MACD Fast Length
Default: 12
Range: 1 to unlimited
Description: Sets the faster exponential moving average for MACD calculation. Standard 12-period setting works well for daily charts and captures short-term momentum shifts. Decreasing creates more responsive MACD crossovers but increases false signals. Increasing smooths out noise but delays signal generation, affecting the bullish/bearish indicator state classification.
MACD Slow Length
Default: 26
Range: 1 to unlimited
Description: Defines the slower exponential moving average for MACD calculation. Traditional 26-period setting balances trend identification with responsiveness. Must be greater than Fast Length. Wider spread between fast and slow increases MACD sensitivity to trend changes, impacting the frequency of indicator state transitions in the matrix.
MACD Signal Length
Default: 9
Range: 1 to unlimited
Description: Smoothing period for the MACD signal line that triggers bullish/bearish state changes. Standard 9-period provides reliable crossover signals. Shorter values create more frequent state changes and earlier signals but with more whipsaws. Longer values produce more confirmed, stable signals but with increased lag in detecting momentum shifts.
Volume MA Period
Default: 20
Range: 1 to unlimited
Description: Lookback period for volume moving average used to classify volume as “high” or “low” in indicator state combinations. 20-period default captures typical monthly trading patterns. Shorter periods (10-15) make volume classification more reactive to recent spikes. Longer periods (30-50) require more sustained volume changes to trigger state classification shifts.
Statistics Lookback Period
Default: 200
Range: 50 to 500
Description: Number of historical bars used to calculate matrix statistics. 200 bars provides substantial data for reliable patterns while remaining responsive to regime changes. Lower values (50-100) emphasize recent market behavior and adapt quickly but may produce volatile statistics. Higher values (300-500) capture long-term patterns with stable statistics but slower adaptation to changing market dynamics.
Forward Performance Bars
Default: 5
Range: 1 to 20
Description: Number of bars ahead used to calculate forward returns from each historical state occurrence. 5-bar default suits intraday to short-term swing trading (5 hours on hourly charts, 1 week on daily charts). Lower values (1-3) target short-term momentum trades. Higher values (10-20) align with position trading and longer-term pattern exploitation.
Color Intensity Sensitivity
Default: 2.0
Range: 0.5 to 5.0, step 0.5
Description: Amplifies or dampens the color intensity response to average return magnitudes in the matrix heat map. 2.0 default provides balanced visual emphasis. Lower values (0.5-1.0) create subtle coloring requiring larger returns for full saturation, useful for volatile instruments. Higher values (3.0-5.0) produce vivid colors from smaller returns, highlighting subtle edges in range-bound markets.
Minimum Occurrences for Coloring
Default: 3
Range: 1 to 10
Description: Required minimum sample size before applying color-coded performance to matrix cells. Cells with fewer occurrences display gray “insufficient data” warning. 3-occurrence default filters out rare patterns. Lower threshold (1-2) shows more data but includes unreliable single-event statistics. Higher thresholds (5-10) ensure only well-established patterns receive visual emphasis.
Table Position
Default: top_right
Options: top_left, top_right, bottom_left, bottom_right
Description: Screen location for the 56-cell statistics matrix table. Position to avoid overlapping critical price action or other indicators on your chart. Consider chart orientation and candlestick density when selecting optimal placement.
Show Current State Panel
Default: true
Options: true, false
Description: Toggle visibility of the dedicated current state information panel. When enabled, displays signal strength, RSI value, indicator status, average return, estimated win rate, and confidence level for active market conditions. Disable to declutter charts when only the matrix table is needed.
Info Panel Position
Default: bottom_left
Options: top_left, top_right, bottom_left, bottom_right
Description: Screen location for the current state information panel (when enabled). Position independently from statistics matrix to optimize chart real estate. Typically placed opposite the matrix table for balanced visual layout.
Win Rate Smoothing Strength
Default: 5
Range: 1 to 20
Description: Controls Bayesian prior weighting for estimated win rate calculations. Acts as virtual sample size assuming 50% win rate baseline. Default 5 provides moderate smoothing preventing extreme win rate estimates from small samples. Lower values (1-3) reduce smoothing effect, allowing win rates to reflect raw data more directly. Higher values (10-20) increase conservatism, pulling win rate estimates toward 50% until substantial evidence accumulates.
✅Best Use Cases
Pattern-based discretionary trading where you want historical confirmation before entering setups that “look good” based on current technical alignment
Swing trading with holding periods matching your forward performance bar setting, using high-confidence bullish cells as entry filters
Risk assessment and position sizing, allocating larger size to trades originating from cells with strong positive average returns and high estimated win rates
Market regime identification by observing which RSI states and indicator combinations are currently producing the most reliable historical patterns
Backtesting validation by comparing your manual strategy signals against the historical performance of the corresponding matrix cells
Educational tool for developing intuition about which technical condition combinations have actually worked versus those that feel right but lack historical evidence
⚠️Limitations
Historical patterns do not guarantee future performance, especially during unprecedented market events or regime changes not represented in the lookback period
Small sample sizes (low occurrence counts) produce unreliable statistics despite Bayesian smoothing, requiring caution when acting on low-confidence cells
Matrix statistics lag behind rapidly changing market conditions, as the lookback period must accumulate new state occurrences before updating performance data
Forward return calculations use fixed bar periods that may not align with actual trade exit timing, support/resistance levels, or volatility-adjusted profit targets
💡What Makes This Unique
Multi-Dimensional State Space: Unlike single-indicator tools, simultaneously tracks 56 distinct market condition combinations providing granular pattern resolution unavailable in traditional technical analysis
Bayesian Statistical Rigor: Implements proper probabilistic smoothing to prevent overconfidence from limited data, a critical feature missing from most pattern recognition tools
Real-Time Contextual Feedback: The “NOW” marker and dedicated info panel instantly connect current market conditions to their historical performance profile, eliminating guesswork
Transparent Occurrence Counts: Displays sample sizes directly in each cell, allowing traders to judge statistical reliability themselves rather than hiding data quality issues
Fully Customizable Analysis Window: Complete control over lookback depth and forward return horizons lets traders align the tool precisely with their trading timeframe and strategy requirements
🔬How It Works
1. State Classification and Encoding
Each bar’s RSI value is evaluated and assigned to one of 7 discrete states based on threshold levels (0: <20, 1: 20-30, 2: 30-40, 3: 40-60, 4: 60-70, 5: 70-80, 6: >80)
Simultaneously, three binary conditions are evaluated: MACD line position relative to signal line, current volume relative to its moving average, and current close relative to previous close
These three binary conditions are combined into a single indicator state integer (0-7) using binary encoding, creating 8 possible indicator combinations
The RSI state and indicator state are stored together, defining one of 56 possible market condition cells in the matrix
2. Historical Data Accumulation
As each bar completes, the current state classification, closing price, and bar index are stored in rolling arrays maintained at the size specified by the lookback period
When the arrays reach capacity, the oldest data point is removed and the newest added, creating a sliding historical window
This continuous process builds a comprehensive database of past market conditions and their subsequent price movements
3. Forward Return Calculation and Statistics Update
On each bar, the indicator looks back through the stored historical data to find bars where sufficient forward bars exist to measure outcomes
For each historical occurrence, the price change from that bar to the bar N periods ahead (where N is the forward performance bars setting) is calculated as a percentage return
This percentage return is added to the cumulative return total for the specific matrix cell corresponding to that historical bar’s state classification
Occurrence counts are incremented, and wins are tallied for positive returns, building comprehensive statistics for each of the 56 cells
The Bayesian smoothing formula combines these raw statistics with prior assumptions (neutral 50% win rate) weighted by the smoothing strength parameter to produce estimated win rates that remain stable even with small samples
💡Note:
The Historical Matrix Analyzer is designed as a decision support tool, not a standalone trading system. Best results come from using it to validate discretionary trade ideas or filter systematic strategy signals. Always combine matrix insights with proper risk management, position sizing rules, and awareness of broader market context. The estimated win rate feature uses Bayesian statistics specifically to prevent false confidence from limited data, but no amount of smoothing can create reliable predictions from fundamentally insufficient sample sizes. Focus on high-confidence cells (green-colored confidence indicators) with occurrence counts well above your minimum threshold for the most actionable insights.
Tzotchev Trend Measure [EdgeTools]Are you still measuring trend strength with moving averages? Here is a better variant at scientific level:
Tzotchev Trend Measure: A Statistical Approach to Trend Following
The Tzotchev Trend Measure represents a sophisticated advancement in quantitative trend analysis, moving beyond traditional moving average-based indicators toward a statistically rigorous framework for measuring trend strength. This indicator implements the methodology developed by Tzotchev et al. (2015) in their seminal J.P. Morgan research paper "Designing robust trend-following system: Behind the scenes of trend-following," which introduced a probabilistic approach to trend measurement that has since become a cornerstone of institutional trading strategies.
Mathematical Foundation and Statistical Theory
The core innovation of the Tzotchev Trend Measure lies in its transformation of price momentum into a probability-based metric through the application of statistical hypothesis testing principles. The indicator employs the fundamental formula ST = 2 × Φ(√T × r̄T / σ̂T) - 1, where ST represents the trend strength score bounded between -1 and +1, Φ(x) denotes the normal cumulative distribution function, T represents the lookback period in trading days, r̄T is the average logarithmic return over the specified period, and σ̂T represents the estimated daily return volatility.
This formulation transforms what is essentially a t-statistic into a probabilistic trend measure, testing the null hypothesis that the mean return equals zero against the alternative hypothesis of non-zero mean return. The use of logarithmic returns rather than simple returns provides several statistical advantages, including symmetry properties where log(P₁/P₀) = -log(P₀/P₁), additivity characteristics that allow for proper compounding analysis, and improved validity of normal distribution assumptions that underpin the statistical framework.
The implementation utilizes the Abramowitz and Stegun (1964) approximation for the normal cumulative distribution function, achieving accuracy within ±1.5 × 10⁻⁷ for all input values. This approximation employs Horner's method for polynomial evaluation to ensure numerical stability, particularly important when processing large datasets or extreme market conditions.
Comparative Analysis with Traditional Trend Measurement Methods
The Tzotchev Trend Measure demonstrates significant theoretical and empirical advantages over conventional trend analysis techniques. Traditional moving average-based systems, including simple moving averages (SMA), exponential moving averages (EMA), and their derivatives such as MACD, suffer from several fundamental limitations that the Tzotchev methodology addresses systematically.
Moving average systems exhibit inherent lag bias, as documented by Kaufman (2013) in "Trading Systems and Methods," where he demonstrates that moving averages inevitably lag price movements by approximately half their period length. This lag creates delayed signal generation that reduces profitability in trending markets and increases false signal frequency during consolidation periods. In contrast, the Tzotchev measure eliminates lag bias by directly analyzing the statistical properties of return distributions rather than smoothing price levels.
The volatility normalization inherent in the Tzotchev formula addresses a critical weakness in traditional momentum indicators. As shown by Bollinger (2001) in "Bollinger on Bollinger Bands," momentum oscillators like RSI and Stochastic fail to account for changing volatility regimes, leading to inconsistent signal interpretation across different market conditions. The Tzotchev measure's incorporation of return volatility in the denominator ensures that trend strength assessments remain consistent regardless of the underlying volatility environment.
Empirical studies by Hurst, Ooi, and Pedersen (2013) in "Demystifying Managed Futures" demonstrate that traditional trend-following indicators suffer from significant drawdowns during whipsaw markets, with Sharpe ratios frequently below 0.5 during challenging periods. The authors attribute these poor performance characteristics to the binary nature of most trend signals and their inability to quantify signal confidence. The Tzotchev measure addresses this limitation by providing continuous probability-based outputs that allow for more sophisticated risk management and position sizing strategies.
The statistical foundation of the Tzotchev approach provides superior robustness compared to technical indicators that lack theoretical grounding. Fama and French (1988) in "Permanent and Temporary Components of Stock Prices" established that price movements contain both permanent and temporary components, with traditional moving averages unable to distinguish between these elements effectively. The Tzotchev methodology's hypothesis testing framework specifically tests for the presence of permanent trend components while filtering out temporary noise, providing a more theoretically sound approach to trend identification.
Research by Moskowitz, Ooi, and Pedersen (2012) in "Time Series Momentum in the Cross Section of Asset Returns" found that traditional momentum indicators exhibit significant variation in effectiveness across asset classes and time periods. Their study of multiple asset classes over decades revealed that simple price-based momentum measures often fail to capture persistent trends in fixed income and commodity markets. The Tzotchev measure's normalization by volatility and its probabilistic interpretation provide consistent performance across diverse asset classes, as demonstrated in the original J.P. Morgan research.
Comparative performance studies conducted by AQR Capital Management (Asness, Moskowitz, and Pedersen, 2013) in "Value and Momentum Everywhere" show that volatility-adjusted momentum measures significantly outperform traditional price momentum across international equity, bond, commodity, and currency markets. The study documents Sharpe ratio improvements of 0.2 to 0.4 when incorporating volatility normalization, consistent with the theoretical advantages of the Tzotchev approach.
The regime detection capabilities of the Tzotchev measure provide additional advantages over binary trend classification systems. Research by Ang and Bekaert (2002) in "Regime Switches in Interest Rates" demonstrates that financial markets exhibit distinct regime characteristics that traditional indicators fail to capture adequately. The Tzotchev measure's five-tier classification system (Strong Bull, Weak Bull, Neutral, Weak Bear, Strong Bear) provides more nuanced market state identification than simple trend/no-trend binary systems.
Statistical testing by Jegadeesh and Titman (2001) in "Profitability of Momentum Strategies" revealed that traditional momentum indicators suffer from significant parameter instability, with optimal lookback periods varying substantially across market conditions and asset classes. The Tzotchev measure's statistical framework provides more stable parameter selection through its grounding in hypothesis testing theory, reducing the need for frequent parameter optimization that can lead to overfitting.
Advanced Noise Filtering and Market Regime Detection
A significant enhancement over the original Tzotchev methodology is the incorporation of a multi-factor noise filtering system designed to reduce false signals during sideways market conditions. The filtering mechanism employs four distinct approaches: adaptive thresholding based on current market regime strength, volatility-based filtering utilizing ATR percentile analysis, trend strength confirmation through momentum alignment, and a comprehensive multi-factor approach that combines all methodologies.
The adaptive filtering system analyzes market microstructure through price change relative to average true range, calculates volatility percentiles over rolling windows, and assesses trend alignment across multiple timeframes using exponential moving averages of varying periods. This approach addresses one of the primary limitations identified in traditional trend-following systems, namely their tendency to generate excessive false signals during periods of low volatility or sideways price action.
The regime detection component classifies market conditions into five distinct categories: Strong Bull (ST > 0.3), Weak Bull (0.1 < ST ≤ 0.3), Neutral (-0.1 ≤ ST ≤ 0.1), Weak Bear (-0.3 ≤ ST < -0.1), and Strong Bear (ST < -0.3). This classification system provides traders with clear, quantitative definitions of market regimes that can inform position sizing, risk management, and strategy selection decisions.
Professional Implementation and Trading Applications
The indicator incorporates three distinct trading profiles designed to accommodate different investment approaches and risk tolerances. The Conservative profile employs longer lookback periods (63 days), higher signal thresholds (0.2), and reduced filter sensitivity (0.5) to minimize false signals and focus on major trend changes. The Balanced profile utilizes standard academic parameters with moderate settings across all dimensions. The Aggressive profile implements shorter lookback periods (14 days), lower signal thresholds (-0.1), and increased filter sensitivity (1.5) to capture shorter-term trend movements.
Signal generation occurs through threshold crossover analysis, where long signals are generated when the trend measure crosses above the specified threshold and short signals when it crosses below. The implementation includes sophisticated signal confirmation mechanisms that consider trend alignment across multiple timeframes and momentum strength percentiles to reduce the likelihood of false breakouts.
The alert system provides real-time notifications for trend threshold crossovers, strong regime changes, and signal generation events, with configurable frequency controls to prevent notification spam. Alert messages are standardized to ensure consistency across different market conditions and timeframes.
Performance Optimization and Computational Efficiency
The implementation incorporates several performance optimization features designed to handle large datasets efficiently. The maximum bars back parameter allows users to control historical calculation depth, with default settings optimized for most trading applications while providing flexibility for extended historical analysis. The system includes automatic performance monitoring that generates warnings when computational limits are approached.
Error handling mechanisms protect against division by zero conditions, infinite values, and other numerical instabilities that can occur during extreme market conditions. The finite value checking system ensures data integrity throughout the calculation process, with fallback mechanisms that maintain indicator functionality even when encountering corrupted or missing price data.
Timeframe validation provides warnings when the indicator is applied to unsuitable timeframes, as the Tzotchev methodology was specifically designed for daily and higher timeframe analysis. This validation helps prevent misapplication of the indicator in contexts where its statistical assumptions may not hold.
Visual Design and User Interface
The indicator features eight professional color schemes designed for different trading environments and user preferences. The EdgeTools theme provides an institutional blue and steel color palette suitable for professional trading environments. The Gold theme offers warm colors optimized for commodities trading. The Behavioral theme incorporates psychology-based color contrasts that align with behavioral finance principles. The Quant theme provides neutral colors suitable for analytical applications.
Additional specialized themes include Ocean, Fire, Matrix, and Arctic variations, each optimized for specific visual preferences and trading contexts. All color schemes include automatic dark and light mode optimization to ensure optimal readability across different chart backgrounds and trading platforms.
The information table provides real-time display of key metrics including current trend measure value, market regime classification, signal strength, Z-score, average returns, volatility measures, filter threshold levels, and filter effectiveness percentages. This comprehensive dashboard allows traders to monitor all relevant indicator components simultaneously.
Theoretical Implications and Research Context
The Tzotchev Trend Measure addresses several theoretical limitations inherent in traditional technical analysis approaches. Unlike moving average-based systems that rely on price level comparisons, this methodology grounds trend analysis in statistical hypothesis testing, providing a more robust theoretical foundation for trading decisions.
The probabilistic interpretation of trend strength offers significant advantages over binary trend classification systems. Rather than simply indicating whether a trend exists, the measure quantifies the statistical confidence level associated with the trend assessment, allowing for more nuanced risk management and position sizing decisions.
The incorporation of volatility normalization addresses the well-documented problem of volatility clustering in financial time series, ensuring that trend strength assessments remain consistent across different market volatility regimes. This normalization is particularly important for portfolio management applications where consistent risk metrics across different assets and time periods are essential.
Practical Applications and Trading Strategy Integration
The Tzotchev Trend Measure can be effectively integrated into various trading strategies and portfolio management frameworks. For trend-following strategies, the indicator provides clear entry and exit signals with quantified confidence levels. For mean reversion strategies, extreme readings can signal potential turning points. For portfolio allocation, the regime classification system can inform dynamic asset allocation decisions.
The indicator's statistical foundation makes it particularly suitable for quantitative trading strategies where systematic, rules-based approaches are preferred over discretionary decision-making. The standardized output range facilitates easy integration with position sizing algorithms and risk management systems.
Risk management applications benefit from the indicator's ability to quantify trend strength and provide early warning signals of potential trend changes. The multi-timeframe analysis capability allows for the construction of robust risk management frameworks that consider both short-term tactical and long-term strategic market conditions.
Implementation Guide and Parameter Configuration
The practical application of the Tzotchev Trend Measure requires careful parameter configuration to optimize performance for specific trading objectives and market conditions. This section provides comprehensive guidance for parameter selection and indicator customization.
Core Calculation Parameters
The Lookback Period parameter controls the statistical window used for trend calculation and represents the most critical setting for the indicator. Default values range from 14 to 63 trading days, with shorter periods (14-21 days) providing more sensitive trend detection suitable for short-term trading strategies, while longer periods (42-63 days) offer more stable trend identification appropriate for position trading and long-term investment strategies. The parameter directly influences the statistical significance of trend measurements, with longer periods requiring stronger underlying trends to generate significant signals but providing greater reliability in trend identification.
The Price Source parameter determines which price series is used for return calculations. The default close price provides standard trend analysis, while alternative selections such as high-low midpoint ((high + low) / 2) can reduce noise in volatile markets, and volume-weighted average price (VWAP) offers superior trend identification in institutional trading environments where volume concentration matters significantly.
The Signal Threshold parameter establishes the minimum trend strength required for signal generation, with values ranging from -0.5 to 0.5. Conservative threshold settings (0.2 to 0.3) reduce false signals but may miss early trend opportunities, while aggressive settings (-0.1 to 0.1) provide earlier signal generation at the cost of increased false positive rates. The optimal threshold depends on the trader's risk tolerance and the volatility characteristics of the traded instrument.
Trading Profile Configuration
The Trading Profile system provides pre-configured parameter sets optimized for different trading approaches. The Conservative profile employs a 63-day lookback period with a 0.2 signal threshold and 0.5 noise sensitivity, designed for long-term position traders seeking high-probability trend signals with minimal false positives. The Balanced profile uses a 21-day lookback with 0.05 signal threshold and 1.0 noise sensitivity, suitable for swing traders requiring moderate signal frequency with acceptable noise levels. The Aggressive profile implements a 14-day lookback with -0.1 signal threshold and 1.5 noise sensitivity, optimized for day traders and scalpers requiring frequent signal generation despite higher noise levels.
Advanced Noise Filtering System
The noise filtering mechanism addresses the challenge of false signals during sideways market conditions through four distinct methodologies. The Adaptive filter adjusts thresholds based on current trend strength, increasing sensitivity during strong trending periods while raising thresholds during consolidation phases. The Volatility-based filter utilizes Average True Range (ATR) percentile analysis to suppress signals during abnormally volatile conditions that typically generate false trend indications.
The Trend Strength filter requires alignment between multiple momentum indicators before confirming signals, reducing the probability of false breakouts from consolidation patterns. The Multi-factor approach combines all filtering methodologies using weighted scoring to provide the most robust noise reduction while maintaining signal responsiveness during genuine trend initiations.
The Noise Sensitivity parameter controls the aggressiveness of the filtering system, with lower values (0.5-1.0) providing conservative filtering suitable for volatile instruments, while higher values (1.5-2.0) allow more signals through but may increase false positive rates during choppy market conditions.
Visual Customization and Display Options
The Color Scheme parameter offers eight professional visualization options designed for different analytical preferences and market conditions. The EdgeTools scheme provides high contrast visualization optimized for trend strength differentiation, while the Gold scheme offers warm tones suitable for commodity analysis. The Behavioral scheme uses psychological color associations to enhance decision-making speed, and the Quant scheme provides neutral colors appropriate for quantitative analysis environments.
The Ocean, Fire, Matrix, and Arctic schemes offer additional aesthetic options while maintaining analytical functionality. Each scheme includes optimized colors for both light and dark chart backgrounds, ensuring visibility across different trading platform configurations.
The Show Glow Effects parameter enhances plot visibility through multiple layered lines with progressive transparency, particularly useful when analyzing multiple timeframes simultaneously or when working with dense price data that might obscure trend signals.
Performance Optimization Settings
The Maximum Bars Back parameter controls the historical data depth available for calculations, with values ranging from 5,000 to 50,000 bars. Higher values enable analysis of longer-term trend patterns but may impact indicator loading speed on slower systems or when applied to multiple instruments simultaneously. The optimal setting depends on the intended analysis timeframe and available computational resources.
The Calculate on Every Tick parameter determines whether the indicator updates with every price change or only at bar close. Real-time calculation provides immediate signal updates suitable for scalping and day trading strategies, while bar-close calculation reduces computational overhead and eliminates signal flickering during bar formation, preferred for swing trading and position management applications.
Alert System Configuration
The Alert Frequency parameter controls notification generation, with options for all signals, bar close only, or once per bar. High-frequency trading strategies benefit from all signals mode, while position traders typically prefer bar close alerts to avoid premature position entries based on intrabar fluctuations.
The alert system generates four distinct notification types: Long Signal alerts when the trend measure crosses above the positive signal threshold, Short Signal alerts for negative threshold crossings, Bull Regime alerts when entering strong bullish conditions, and Bear Regime alerts for strong bearish regime identification.
Table Display and Information Management
The information table provides real-time statistical metrics including current trend value, regime classification, signal status, and filter effectiveness measurements. The table position can be customized for optimal screen real estate utilization, and individual metrics can be toggled based on analytical requirements.
The Language parameter supports both English and German display options for international users, while maintaining consistent calculation methodology regardless of display language selection.
Risk Management Integration
Effective risk management integration requires coordination between the trend measure signals and position sizing algorithms. Strong trend readings (above 0.5 or below -0.5) support larger position sizes due to higher probability of trend continuation, while neutral readings (between -0.2 and 0.2) suggest reduced position sizes or range-trading strategies.
The regime classification system provides additional risk management context, with Strong Bull and Strong Bear regimes supporting trend-following strategies, while Neutral regimes indicate potential for mean reversion approaches. The filter effectiveness metric helps traders assess current market conditions and adjust strategy parameters accordingly.
Timeframe Considerations and Multi-Timeframe Analysis
The indicator's effectiveness varies across different timeframes, with higher timeframes (daily, weekly) providing more reliable trend identification but slower signal generation, while lower timeframes (hourly, 15-minute) offer faster signals with increased noise levels. Multi-timeframe analysis combining trend alignment across multiple periods significantly improves signal quality and reduces false positive rates.
For optimal results, traders should consider trend alignment between the primary trading timeframe and at least one higher timeframe before entering positions. Divergences between timeframes often signal potential trend reversals or consolidation periods requiring strategy adjustment.
Conclusion
The Tzotchev Trend Measure represents a significant advancement in technical analysis methodology, combining rigorous statistical foundations with practical trading applications. Its implementation of the J.P. Morgan research methodology provides institutional-quality trend analysis capabilities previously available only to sophisticated quantitative trading firms.
The comprehensive parameter configuration options enable customization for diverse trading styles and market conditions, while the advanced noise filtering and regime detection capabilities provide superior signal quality compared to traditional trend-following indicators. Proper parameter selection and understanding of the indicator's statistical foundation are essential for achieving optimal trading results and effective risk management.
References
Abramowitz, M. and Stegun, I.A. (1964). Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Washington: National Bureau of Standards.
Ang, A. and Bekaert, G. (2002). Regime Switches in Interest Rates. Journal of Business and Economic Statistics, 20(2), 163-182.
Asness, C.S., Moskowitz, T.J., and Pedersen, L.H. (2013). Value and Momentum Everywhere. Journal of Finance, 68(3), 929-985.
Bollinger, J. (2001). Bollinger on Bollinger Bands. New York: McGraw-Hill.
Fama, E.F. and French, K.R. (1988). Permanent and Temporary Components of Stock Prices. Journal of Political Economy, 96(2), 246-273.
Hurst, B., Ooi, Y.H., and Pedersen, L.H. (2013). Demystifying Managed Futures. Journal of Investment Management, 11(3), 42-58.
Jegadeesh, N. and Titman, S. (2001). Profitability of Momentum Strategies: An Evaluation of Alternative Explanations. Journal of Finance, 56(2), 699-720.
Kaufman, P.J. (2013). Trading Systems and Methods. 5th Edition. Hoboken: John Wiley & Sons.
Moskowitz, T.J., Ooi, Y.H., and Pedersen, L.H. (2012). Time Series Momentum. Journal of Financial Economics, 104(2), 228-250.
Tzotchev, D., Lo, A.W., and Hasanhodzic, J. (2015). Designing robust trend-following system: Behind the scenes of trend-following. J.P. Morgan Quantitative Research, Asset Management Division.
Recession Warning Model [BackQuant]Recession Warning Model
Overview
The Recession Warning Model (RWM) is a Pine Script® indicator designed to estimate the probability of an economic recession by integrating multiple macroeconomic, market sentiment, and labor market indicators. It combines over a dozen data series into a transparent, adaptive, and actionable tool for traders, portfolio managers, and researchers. The model provides customizable complexity levels, display modes, and data processing options to accommodate various analytical requirements while ensuring robustness through dynamic weighting and regime-aware adjustments.
Purpose
The RWM fulfills the need for a concise yet comprehensive tool to monitor recession risk. Unlike approaches relying on a single metric, such as yield-curve inversion, or extensive economic reports, it consolidates multiple data sources into a single probability output. The model identifies active indicators, their confidence levels, and the current economic regime, enabling users to anticipate downturns and adjust strategies accordingly.
Core Features
- Indicator Families : Incorporates 13 indicators across five categories: Yield, Labor, Sentiment, Production, and Financial Stress.
- Dynamic Weighting : Adjusts indicator weights based on recent predictive accuracy, constrained within user-defined boundaries.
- Leading and Coincident Split : Separates early-warning (leading) and confirmatory (coincident) signals, with adjustable weighting (default 60/40 mix).
- Economic Regime Sensitivity : Modulates output sensitivity based on market conditions (Expansion, Late-Cycle, Stress, Crisis), using a composite of VIX, yield-curve, financial conditions, and credit spreads.
- Display Options : Supports four modes—Probability (0-100%), Binary (four risk bins), Lead/Coincident, and Ensemble (blended probability).
- Confidence Intervals : Reflects model stability, widening during high volatility or conflicting signals.
- Alerts : Configurable thresholds (Watch, Caution, Warning, Alert) with persistence filters to minimize false signals.
- Data Export : Enables CSV output for probabilities, signals, and regimes, facilitating external analysis in Python or R.
Model Complexity Levels
Users can select from four tiers to balance simplicity and depth:
1. Essential : Focuses on three core indicators—yield-curve spread, jobless claims, and unemployment change—for minimalistic monitoring.
2. Standard : Expands to nine indicators, adding consumer confidence, PMI, VIX, S&P 500 trend, money supply vs. GDP, and the Sahm Rule.
3. Professional : Includes all 13 indicators, incorporating financial conditions, credit spreads, JOLTS vacancies, and wage growth.
4. Research : Unlocks all indicators plus experimental settings for advanced users.
Key Indicators
Below is a summary of the 13 indicators, their data sources, and economic significance:
- Yield-Curve Spread : Difference between 10-year and 3-month Treasury yields. Negative spreads signal banking sector stress.
- Jobless Claims : Four-week moving average of unemployment claims. Sustained increases indicate rising layoffs.
- Unemployment Change : Three-month change in unemployment rate. Sharp rises often precede recessions.
- Sahm Rule : Triggers when unemployment rises 0.5% above its 12-month low, a reliable recession indicator.
- Consumer Confidence : University of Michigan survey. Declines reflect household pessimism, impacting spending.
- PMI : Purchasing Managers’ Index. Values below 50 indicate manufacturing contraction.
- VIX : CBOE Volatility Index. Elevated levels suggest market anticipation of economic distress.
- S&P 500 Growth : Weekly moving average trend. Declines reduce wealth effects, curbing consumption.
- M2 + GDP Trend : Monitors money supply and real GDP. Simultaneous declines signal credit contraction.
- NFCI : Chicago Fed’s National Financial Conditions Index. Positive values indicate tighter conditions.
- Credit Spreads : Proxy for corporate bond spreads using 10-year vs. 2-year Treasury yields. Widening spreads reflect stress.
- JOLTS Vacancies : Job openings data. Significant drops precede hiring slowdowns.
- Wage Growth : Year-over-year change in average hourly earnings. Late-cycle spikes often signal economic overheating.
Data Processing
- Rate of Change (ROC) : Optionally applied to capture momentum in data series (default: 21-bar period).
- Z-Score Normalization : Standardizes indicators to a common scale (default: 252-bar lookback).
- Smoothing : Applies a short moving average to final signals (default: 5-bar period) to reduce noise.
- Binary Signals : Generated for each indicator (e.g., yield-curve inverted or PMI below 50) based on thresholds or Z-score deviations.
Probability Calculation
1. Each indicator’s binary signal is weighted according to user settings or dynamic performance.
2. Weights are normalized to sum to 100% across active indicators.
3. Leading and coincident signals are aggregated separately (if split mode is enabled) and combined using the specified mix.
4. The probability is adjusted by a regime multiplier, amplifying risk during Stress or Crisis regimes.
5. Optional smoothing ensures stable outputs.
Display and Visualization
- Probability Mode : Plots a continuous 0-100% recession probability with color gradients and confidence bands.
- Binary Mode : Categorizes risk into four levels (Minimal, Watch, Caution, Alert) for simplified dashboards.
- Lead/Coincident Mode : Displays leading and coincident probabilities separately to track signal divergence.
- Ensemble Mode : Averages traditional and split probabilities for a balanced view.
- Regime Background : Color-coded overlays (green for Expansion, orange for Late-Cycle, amber for Stress, red for Crisis).
- Analytics Table : Optional dashboard showing probability, confidence, regime, and top indicator statuses.
Practical Applications
- Asset Allocation : Adjust equity or bond exposures based on sustained probability increases.
- Risk Management : Hedge portfolios with VIX futures or options during regime shifts to Stress or Crisis.
- Sector Rotation : Shift toward defensive sectors when coincident signals rise above 50%.
- Trading Filters : Disable short-term strategies during high-risk regimes.
- Event Timing : Scale positions ahead of high-impact data releases when probability and VIX are elevated.
Configuration Guidelines
- Enable ROC and Z-score for consistent indicator comparison unless raw data is preferred.
- Use dynamic weighting with at least one economic cycle of data for optimal performance.
- Monitor stress composite scores above 80 alongside probabilities above 70 for critical risk signals.
- Adjust adaptation speed (default: 0.1) to 0.2 during Crisis regimes for faster indicator prioritization.
- Combine RWM with complementary tools (e.g., liquidity metrics) for intraday or short-term trading.
Limitations
- Macro indicators lag intraday market moves, making RWM better suited for strategic rather than tactical trading.
- Historical data availability may constrain dynamic weighting on shorter timeframes.
- Model accuracy depends on the quality and timeliness of economic data feeds.
Final Note
The Recession Warning Model provides a disciplined framework for monitoring economic downturn risks. By integrating diverse indicators with transparent weighting and regime-aware adjustments, it empowers users to make informed decisions in portfolio management, risk hedging, or macroeconomic research. Regular review of model outputs alongside market-specific tools ensures its effective application across varying market conditions.
Trend Gauge [BullByte]Trend Gauge
Summary
A multi-factor trend detection indicator that aggregates EMA alignment, VWMA momentum scaling, volume spikes, ATR breakout strength, higher-timeframe confirmation, ADX-based regime filtering, and RSI pivot-divergence penalty into one normalized trend score. It also provides a confidence meter, a Δ Score momentum histogram, divergence highlights, and a compact, scalable dashboard for at-a-glance status.
________________________________________
## 1. Purpose of the Indicator
Why this was built
Traders often monitor several indicators in parallel - EMAs, volume signals, volatility breakouts, higher-timeframe trends, ADX readings, divergence alerts, etc., which can be cumbersome and sometimes contradictory. The “Trend Gauge” indicator was created to consolidate these complementary checks into a single, normalized score that reflects the prevailing market bias (bullish, bearish, or neutral) and its strength. By combining multiple inputs with an adaptive regime filter, scaling contributions by magnitude, and penalizing weakening signals (divergence), this tool aims to reduce noise, highlight genuine trend opportunities, and warn when momentum fades.
Key Design Goals
Signal Aggregation
Merged trend-following signals (EMA crossover, ATR breakout, higher-timeframe confirmation) and momentum signals (VWMA thrust, volume spikes) into a unified score that reflects directional bias more holistically.
Market Regime Awareness
Implemented an ADX-style filter to distinguish between trending and ranging markets, reducing the influence of trend signals during sideways phases to avoid false breakouts.
Magnitude-Based Scaling
Replaced binary contributions with scaled inputs: VWMA thrust and ATR breakout are weighted relative to recent averages, allowing for more nuanced score adjustments based on signal strength.
Momentum Divergence Penalty
Integrated pivot-based RSI divergence detection to slightly reduce the overall score when early signs of momentum weakening are detected, improving risk-awareness in entries.
Confidence Transparency
Added a live confidence metric that shows what percentage of enabled sub-indicators currently agree with the overall bias, making the scoring system more interpretable.
Momentum Acceleration Visualization
Plotted the change in score (Δ Score) as a histogram bar-to-bar, highlighting whether momentum is increasing, flattening, or reversing, aiding in more timely decision-making.
Compact Informational Dashboard
Presented a clean, scalable dashboard that displays each component’s status, the final score, confidence %, detected regime (Trending/Ranging), and a labeled strength gauge for quick visual assessment.
________________________________________
## 2. Why a Trader Should Use It
Main benefits and use cases
1. Unified View: Rather than juggling multiple windows or panels, this indicator delivers a single score synthesizing diverse signals.
2. Regime Filtering: In ranging markets, trend signals often generate false entries. The ADX-based regime filter automatically down-weights trend-following components, helping you avoid chasing false breakouts.
3. Nuanced Momentum & Volatility: VWMA and ATR breakout contributions are normalized by recent averages, so strong moves register strongly while smaller fluctuations are de-emphasized.
4. Early Warning of Weakening: Pivot-based RSI divergence is detected and used to slightly reduce the score when price/momentum diverges, giving a cautionary signal before a full reversal.
5. Confidence Meter: See at a glance how many sub-indicators align with the aggregated bias (e.g., “80% confidence” means 4 out of 5 components agree ). This transparency avoids black-box decisions.
6. Trend Acceleration/Deceleration View: The Δ Score histogram visualizes whether the aggregated score is rising (accelerating trend) or falling (momentum fading), supplementing the main oscillator.
7. Compact Dashboard: A corner table lists each check’s status (“Bull”, “Bear”, “Flat” or “Disabled”), plus overall Score, Confidence %, Regime, Trend Strength label, and a gauge bar. Users can scale text size (Normal, Small, Tiny) without removing elements, so the full picture remains visible even in compact layouts.
8. Customizable & Transparent: All components can be enabled/disabled and parameterized (lengths, thresholds, weights). The full Pine code is open and well-commented, letting users inspect or adapt the logic.
9. Alert-ready: Built-in alert conditions fire when the score crosses weak thresholds to bullish/bearish or returns to neutral, enabling timely notifications.
________________________________________
## 3. Component Rationale (“Why These Specific Indicators?”)
Each sub-component was chosen because it adds complementary information about trend or momentum:
1. EMA Cross
o Basic trend measure: compares a faster EMA vs. a slower EMA. Quickly reflects trend shifts but by itself can whipsaw in sideways markets.
2. VWMA Momentum
o Volume-weighted moving average change indicates momentum with volume context. By normalizing (dividing by a recent average absolute change), we capture the strength of momentum relative to recent history. This scaling prevents tiny moves from dominating and highlights genuinely strong momentum.
3. Volume Spikes
o Sudden jumps in volume combined with price movement often accompany stronger moves or reversals. A binary detection (+1 for bullish spike, -1 for bearish spike) flags high-conviction bars.
4. ATR Breakout
o Detects price breaking beyond recent highs/lows by a multiple of ATR. Measures breakout strength by how far beyond the threshold price moves relative to ATR, capped to avoid extreme outliers. This gives a volatility-contextual trend signal.
5. Higher-Timeframe EMA Alignment
o Confirms whether the shorter-term trend aligns with a higher timeframe trend. Uses request.security with lookahead_off to avoid future data. When multiple timeframes agree, confidence in direction increases.
6. ADX Regime Filter (Manual Calculation)
o Computes directional movement (+DM/–DM), smoothes via RMA, computes DI+ and DI–, then a DX and ADX-like value. If ADX ≥ threshold, market is “Trending” and trend components carry full weight; if ADX < threshold, “Ranging” mode applies a configurable weight multiplier (e.g., 0.5) to trend-based contributions, reducing false signals in sideways conditions. Volume spikes remain binary (optional behavior; can be adjusted if desired).
7. RSI Pivot-Divergence Penalty
o Uses ta.pivothigh / ta.pivotlow with a lookback to detect pivot highs/lows on price and corresponding RSI values. When price makes a higher high but RSI makes a lower high (bearish divergence), or price makes a lower low but RSI makes a higher low (bullish divergence), a divergence signal is set. Rather than flipping the trend outright, the indicator subtracts (or adds) a small penalty (configurable) from the aggregated score if it would weaken the current bias. This subtle adjustment warns of weakening momentum without overreacting to noise.
8. Confidence Meter
o Counts how many enabled components currently agree in direction with the aggregated score (i.e., component sign × score sign > 0). Displays this as a percentage. A high percentage indicates strong corroboration; a low percentage warns of mixed signals.
9. Δ Score Momentum View
o Plots the bar-to-bar change in the aggregated score (delta_score = score - score ) as a histogram. When positive, bars are drawn in green above zero; when negative, bars are drawn in red below zero. This reveals acceleration (rising Δ) or deceleration (falling Δ), supplementing the main oscillator.
10. Dashboard
• A table in the indicator pane’s top-right with 11 rows:
1. EMA Cross status
2. VWMA Momentum status
3. Volume Spike status
4. ATR Breakout status
5. Higher-Timeframe Trend status
6. Score (numeric)
7. Confidence %
8. Regime (“Trending” or “Ranging”)
9. Trend Strength label (e.g., “Weak Bullish Trend”, “Strong Bearish Trend”)
10. Gauge bar visually representing score magnitude
• All rows always present; size_opt (Normal, Small, Tiny) only changes text size via text_size, not which elements appear. This ensures full transparency.
________________________________________
## 4. What Makes This Indicator Stand Out
• Regime-Weighted Multi-Factor Score: Trend and momentum signals are adaptively weighted by market regime (trending vs. ranging) , reducing false signals.
• Magnitude Scaling: VWMA and ATR breakout contributions are normalized by recent average momentum or ATR, giving finer gradation compared to simple ±1.
• Integrated Divergence Penalty: Divergence directly adjusts the aggregated score rather than appearing as a separate subplot; this influences alerts and trend labeling in real time.
• Confidence Meter: Shows the percentage of sub-signals in agreement, providing transparency and preventing blind trust in a single metric.
• Δ Score Histogram Momentum View: A histogram highlights acceleration or deceleration of the aggregated trend score, helping detect shifts early.
• Flexible Dashboard: Always-visible component statuses and summary metrics in one place; text size scaling keeps the full picture available in cramped layouts.
• Lookahead-Safe HTF Confirmation: Uses lookahead_off so no future data is accessed from higher timeframes, avoiding repaint bias.
• Repaint Transparency: Divergence detection uses pivot functions that inherently confirm only after lookback bars; description documents this lag so users understand how and when divergence labels appear.
• Open-Source & Educational: Full, well-commented Pine v6 code is provided; users can learn from its structure: manual ADX computation, conditional plotting with series = show ? value : na, efficient use of table.new in barstate.islast, and grouped inputs with tooltips.
• Compliance-Conscious: All plots have descriptive titles; inputs use clear names; no unnamed generic “Plot” entries; manual ADX uses RMA; all request.security calls use lookahead_off. Code comments mention repaint behavior and limitations.
________________________________________
## 5. Recommended Timeframes & Tuning
• Any Timeframe: The indicator works on small (e.g., 1m) to large (daily, weekly) timeframes. However:
o On very low timeframes (<1m or tick charts), noise may produce frequent whipsaws. Consider increasing smoothing lengths, disabling certain components (e.g., volume spike if volume data noisy), or using a larger pivot lookback for divergence.
o On higher timeframes (daily, weekly), consider longer lookbacks for ATR breakout or divergence, and set Higher-Timeframe trend appropriately (e.g., 4H HTF when on 5 Min chart).
• Defaults & Experimentation: Default input values are chosen to be balanced for many liquid markets. Users should test with replay or historical analysis on their symbol/timeframe and adjust:
o ADX threshold (e.g., 20–30) based on instrument volatility.
o VWMA and ATR scaling lengths to match average volatility cycles.
o Pivot lookback for divergence: shorter for faster markets, longer for slower ones.
• Combining with Other Analysis: Use in conjunction with price action, support/resistance, candlestick patterns, order flow, or other tools as desired. The aggregated score and alerts can guide attention but should not be the sole decision-factor.
________________________________________
## 6. How Scoring and Logic Works (Step-by-Step)
1. Compute Sub-Scores
o EMA Cross: Evaluate fast EMA > slow EMA ? +1 : fast EMA < slow EMA ? -1 : 0.
o VWMA Momentum: Calculate vwma = ta.vwma(close, length), then vwma_mom = vwma - vwma . Normalize: divide by recent average absolute momentum (e.g., ta.sma(abs(vwma_mom), lookback)), clip to .
o Volume Spike: Compute vol_SMA = ta.sma(volume, len). If volume > vol_SMA * multiplier AND price moved up ≥ threshold%, assign +1; if moved down ≥ threshold%, assign -1; else 0.
o ATR Breakout: Determine recent high/low over lookback. If close > high + ATR*mult, compute distance = close - (high + ATR*mult), normalize by ATR, cap at a configured maximum. Assign positive contribution. Similarly for bearish breakout below low.
o Higher-Timeframe Trend: Use request.security(..., lookahead=barmerge.lookahead_off) to fetch HTF EMAs; assign +1 or -1 based on alignment.
2. ADX Regime Weighting
o Compute manual ADX: directional movements (+DM, –DM), smoothed via RMA, DI+ and DI–, then DX and ADX via RMA. If ADX ≥ threshold, market is considered “Trending”; otherwise “Ranging.”
o If trending, trend-based contributions (EMA, VWMA, ATR, HTF) use full weight = 1.0. If ranging, use weight = ranging_weight (e.g., 0.5) to down-weight them. Volume spike stays binary ±1 (optional to change if desired).
3. Aggregate Raw Score
o Sum weighted contributions of all enabled components. Count the number of enabled components; if zero, default count = 1 to avoid division by zero.
4. Divergence Penalty
o Detect pivot highs/lows on price and corresponding RSI values, using a lookback. When price and RSI diverge (bearish or bullish divergence), check if current raw score is in the opposing direction:
If bearish divergence (price higher high, RSI lower high) and raw score currently positive, subtract a penalty (e.g., 0.5).
If bullish divergence (price lower low, RSI higher low) and raw score currently negative, add a penalty.
o This reduces score magnitude to reflect weakening momentum, without flipping the trend outright.
5. Normalize and Smooth
o Normalized score = (raw_score / number_of_enabled_components) * 100. This yields a roughly range.
o Optional EMA smoothing of this normalized score to reduce noise.
6. Interpretation
o Sign: >0 = net bullish bias; <0 = net bearish bias; near zero = neutral.
o Magnitude Zones: Compare |score| to thresholds (Weak, Medium, Strong) to label trend strength (e.g., “Weak Bullish Trend”, “Medium Bearish Trend”, “Strong Bullish Trend”).
o Δ Score Histogram: The histogram bars from zero show change from previous bar’s score; positive bars indicate acceleration, negative bars indicate deceleration.
o Confidence: Percentage of sub-indicators aligned with the score’s sign.
o Regime: Indicates whether trend-based signals are fully weighted or down-weighted.
________________________________________
## 7. Oscillator Plot & Visualization: How to Read It
Main Score Line & Area
The oscillator plots the aggregated score as a line, with colored fill: green above zero for bullish area, red below zero for bearish area. Horizontal reference lines at ±Weak, ±Medium, and ±Strong thresholds mark zones: crossing above +Weak suggests beginning of bullish bias, above +Medium for moderate strength, above +Strong for strong trend; similarly for bearish below negative thresholds.
Δ Score Histogram
If enabled, a histogram shows score - score . When positive, bars appear in green above zero, indicating accelerating bullish momentum; when negative, bars appear in red below zero, indicating decelerating or reversing momentum. The height of each bar reflects the magnitude of change in the aggregated score from the prior bar.
Divergence Highlight Fill
If enabled, when a pivot-based divergence is confirmed:
• Bullish Divergence : fill the area below zero down to –Weak threshold in green, signaling potential reversal from bearish to bullish.
• Bearish Divergence : fill the area above zero up to +Weak threshold in red, signaling potential reversal from bullish to bearish.
These fills appear with a lag equal to pivot lookback (the number of bars needed to confirm the pivot). They do not repaint after confirmation, but users must understand this lag.
Trend Direction Label
When score crosses above or below the Weak threshold, a small label appears near the score line reading “Bullish” or “Bearish.” If the score returns within ±Weak, the label “Neutral” appears. This helps quickly identify shifts at the moment they occur.
Dashboard Panel
In the indicator pane’s top-right, a table shows:
1. EMA Cross status: “Bull”, “Bear”, “Flat”, or “Disabled”
2. VWMA Momentum status: similarly
3. Volume Spike status: “Bull”, “Bear”, “No”, or “Disabled”
4. ATR Breakout status: “Bull”, “Bear”, “No”, or “Disabled”
5. Higher-Timeframe Trend status: “Bull”, “Bear”, “Flat”, or “Disabled”
6. Score: numeric value (rounded)
7. Confidence: e.g., “80%” (colored: green for high, amber for medium, red for low)
8. Regime: “Trending” or “Ranging” (colored accordingly)
9. Trend Strength: textual label based on magnitude (e.g., “Medium Bullish Trend”)
10. Gauge: a bar of blocks representing |score|/100
All rows remain visible at all times; changing Dashboard Size only scales text size (Normal, Small, Tiny).
________________________________________
## 8. Example Usage (Illustrative Scenario)
Example: BTCUSD 5 Min
1. Setup: Add “Trend Gauge ” to your BTCUSD 5 Min chart. Defaults: EMAs (8/21), VWMA 14 with lookback 3, volume spike settings, ATR breakout 14/5, HTF = 5m (or adjust to 4H if preferred), ADX threshold 25, ranging weight 0.5, divergence RSI length 14 pivot lookback 5, penalty 0.5, smoothing length 3, thresholds Weak=20, Medium=50, Strong=80. Dashboard Size = Small.
2. Trend Onset: At some point, price breaks above recent high by ATR multiple, volume spikes upward, faster EMA crosses above slower EMA, HTF EMA also bullish, and ADX (manual) ≥ threshold → aggregated score rises above +20 (Weak threshold) into +Medium zone. Dashboard shows “Bull” for EMA, VWMA, Vol Spike, ATR, HTF; Score ~+60–+70; Confidence ~100%; Regime “Trending”; Trend Strength “Medium Bullish Trend”; Gauge ~6–7 blocks. Δ Score histogram bars are green and rising, indicating accelerating bullish momentum. Trader notes the alignment.
3. Divergence Warning: Later, price makes a slightly higher high but RSI fails to confirm (lower RSI high). Pivot lookback completes; the indicator highlights a bearish divergence fill above zero and subtracts a small penalty from the score, causing score to stall or retrace slightly. Dashboard still bullish but score dips toward +Weak. This warns the trader to tighten stops or take partial profits.
4. Trend Weakens: Score eventually crosses below +Weak back into neutral; a “Neutral” label appears, and a “Neutral Trend” alert fires if enabled. Trader exits or avoids new long entries. If score subsequently crosses below –Weak, a “Bearish” label and alert occur.
5. Customization: If the trader finds VWMA noise too frequent on this instrument, they may disable VWMA or increase lookback. If ATR breakouts are too rare, adjust ATR length or multiplier. If ADX threshold seems off, tune threshold. All these adjustments are explained in Inputs section.
6. Visualization: The screenshot shows the main score oscillator with colored areas, reference lines at ±20/50/80, Δ Score histogram bars below/above zero, divergence fill highlighting potential reversal, and the dashboard table in the top-right.
________________________________________
## 9. Inputs Explanation
A concise yet clear summary of inputs helps users understand and adjust:
1. General Settings
• Theme (Dark/Light): Choose background-appropriate colors for the indicator pane.
• Dashboard Size (Normal/Small/Tiny): Scales text size only; all dashboard elements remain visible.
2. Indicator Settings
• Enable EMA Cross: Toggle on/off basic EMA alignment check.
o Fast EMA Length and Slow EMA Length: Periods for EMAs.
• Enable VWMA Momentum: Toggle VWMA momentum check.
o VWMA Length: Period for VWMA.
o VWMA Momentum Lookback: Bars to compare VWMA to measure momentum.
• Enable Volume Spike: Toggle volume spike detection.
o Volume SMA Length: Period to compute average volume.
o Volume Spike Multiplier: How many times above average volume qualifies as spike.
o Min Price Move (%): Minimum percent change in price during spike to qualify as bullish or bearish.
• Enable ATR Breakout: Toggle ATR breakout detection.
o ATR Length: Period for ATR.
o Breakout Lookback: Bars to look back for recent highs/lows.
o ATR Multiplier: Multiplier for breakout threshold.
• Enable Higher Timeframe Trend: Toggle HTF EMA alignment.
o Higher Timeframe: E.g., “5” for 5-minute when on 1-minute chart, or “60” for 5 Min when on 15m, etc. Uses lookahead_off.
• Enable ADX Regime Filter: Toggles regime-based weighting.
o ADX Length: Period for manual ADX calculation.
o ADX Threshold: Value above which market considered trending.
o Ranging Weight Multiplier: Weight applied to trend components when ADX < threshold (e.g., 0.5).
• Scale VWMA Momentum: Toggle normalization of VWMA momentum magnitude.
o VWMA Mom Scale Lookback: Period for average absolute VWMA momentum.
• Scale ATR Breakout Strength: Toggle normalization of breakout distance by ATR.
o ATR Scale Cap: Maximum multiple of ATR used for breakout strength.
• Enable Price-RSI Divergence: Toggle divergence detection.
o RSI Length for Divergence: Period for RSI.
o Pivot Lookback for Divergence: Bars on each side to identify pivot high/low.
o Divergence Penalty: Amount to subtract/add to score when divergence detected (e.g., 0.5).
3. Score Settings
• Smooth Score: Toggle EMA smoothing of normalized score.
• Score Smoothing Length: Period for smoothing EMA.
• Weak Threshold: Absolute score value under which trend is considered weak or neutral.
• Medium Threshold: Score above Weak but below Medium is moderate.
• Strong Threshold: Score above this indicates strong trend.
4. Visualization Settings
• Show Δ Score Histogram: Toggle display of the bar-to-bar change in score as a histogram. Default true.
• Show Divergence Fill: Toggle background fill highlighting confirmed divergences. Default true.
Each input has a tooltip in the code.
________________________________________
## 10. Limitations, Repaint Notes, and Disclaimers
10.1. Repaint & Lag Considerations
• Pivot-Based Divergence Lag: The divergence detection uses ta.pivothigh / ta.pivotlow with a specified lookback. By design, a pivot is only confirmed after the lookback number of bars. As a result:
o Divergence labels or fills appear with a delay equal to the pivot lookback.
o Once the pivot is confirmed and the divergence is detected, the fill/label does not repaint thereafter, but you must understand and accept this lag.
o Users should not treat divergence highlights as predictive signals without additional confirmation, because they appear after the pivot has fully formed.
• Higher-Timeframe EMA Alignment: Uses request.security(..., lookahead=barmerge.lookahead_off), so no future data from the higher timeframe is used. This avoids lookahead bias and ensures signals are based only on completed higher-timeframe bars.
• No Future Data: All calculations are designed to avoid using future information. For example, manual ADX uses RMA on past data; security calls use lookahead_off.
10.2. Market & Noise Considerations
• In very choppy or low-liquidity markets, some components (e.g., volume spikes or VWMA momentum) may be noisy. Users can disable or adjust those components’ parameters.
• On extremely low timeframes, noise may dominate; consider smoothing lengths or disabling certain features.
• On very high timeframes, pivots and breakouts occur less frequently; adjust lookbacks accordingly to avoid sparse signals.
10.3. Not a Standalone Trading System
• This is an indicator, not a complete trading strategy. It provides signals and context but does not manage entries, exits, position sizing, or risk management.
• Users must combine it with their own analysis, money management, and confirmations (e.g., price patterns, support/resistance, fundamental context).
• No guarantees: past behavior does not guarantee future performance.
10.4. Disclaimers
• Educational Purposes Only: The script is provided as-is for educational and informational purposes. It does not constitute financial, investment, or trading advice.
• Use at Your Own Risk: Trading involves risk of loss. Users should thoroughly test and use proper risk management.
• No Guarantees: The author is not responsible for trading outcomes based on this indicator.
• License: Published under Mozilla Public License 2.0; code is open for viewing and modification under MPL terms.
________________________________________
## 11. Alerts
• The indicator defines three alert conditions:
1. Bullish Trend: when the aggregated score crosses above the Weak threshold.
2. Bearish Trend: when the score crosses below the negative Weak threshold.
3. Neutral Trend: when the score returns within ±Weak after being outside.
Good luck
– BullByte
TUF_LOGICThe TUF_LOGIC library incorporates three-valued logic (also known as trilean logic) into Pine Script, enabling the representation of states beyond the binary True and False to include an 'Uncertain' state. This addition is particularly apt for financial market contexts where information may not always be black or white, accommodating scenarios of partial or ambiguous data.
Key Features:
Trilean Data Type: Defines a tri type, facilitating the representation of True (1), Uncertain (0), and False (-1) states, thus accommodating a more nuanced approach to logical evaluation.
Validation and Conversion: Includes methods like validate, to ensure trilean variables conform to expected states, and to_bool, for converting trilean to boolean values, enhancing interoperability with binary logic systems.
Core Logical Operations: Extends traditional logical operators (AND, OR, NOT, XOR, EQUALITY) to work within the trilean domain, enabling complex conditionals that reflect real-world uncertainties.
Specialized Logical Operations:
Implication Operators: Features IMP_K (Kleene's), IMP_L (Łukasiewicz's), and IMP_RM3, offering varied approaches to logical implication within the trilean framework.
Possibility, Necessity, and Contingency Operators: Implements MA ("it is possible that..."), LA ("it is necessary that..."), and IA ("it is unknown/contingent that..."), derived from Tarski-Łukasiewicz's modal logic attempts, enriching the library with modal logic capabilities.
Unanimity Functions: The UNANIMOUS operator assesses complete agreement among trilean values, useful for scenarios requiring consensus or uniformity across multiple indicators or conditions.
This library is developed to support scenarios in financial trading and analysis where decisions might hinge on more than binary outcomes. By incorporating modal logic aspects and providing a framework for handling uncertainty through the MA, LA, and IA operations, TUF_LOGIC bridges the gap between classical binary logic and the realities of uncertain information, making it a valuable tool for developing sophisticated trading strategies and analytical models.
Library "TUF_LOGIC"
3VL Implementation (TUF stands for True, Uncertain, False.)
method validate(self)
Ensures a valid trilean variable. This works by clamping the variable to the range associated with the trilean type.
Namespace types: tri
Parameters:
self (tri)
Returns: Validated trilean object.
method to_bool(self)
Converts a trilean object into a boolean object. True -> True, Uncertain -> na, False -> False.
Namespace types: tri
Parameters:
self (tri)
Returns: A boolean variable.
method NOT(self)
Negates the trilean object. True -> False, Uncertain -> Uncertain, False -> True
Namespace types: tri
Parameters:
self (tri)
Returns: Negated trilean object.
method AND(self, comparator)
Logical AND operation for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The first trilean object.
comparator (tri) : The second trilean object to compare with.
Returns: `tri` Result of the AND operation as a trilean object.
method OR(self, comparator)
Logical OR operation for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The first trilean object.
comparator (tri) : The second trilean object to compare with.
Returns: `tri` Result of the OR operation as a trilean object.
method EQUALITY(self, comparator)
Logical EQUALITY operation for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The first trilean object.
comparator (tri) : The second trilean object to compare with.
Returns: `tri` Result of the EQUALITY operation as a trilean object, True if both are equal, False otherwise.
method XOR(self, comparator)
Logical XOR (Exclusive OR) operation for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The first trilean object.
comparator (tri) : The second trilean object to compare with.
Returns: `tri` Result of the XOR operation as a trilean object.
method IMP_K(self, comparator)
Material implication using Kleene's logic for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The antecedent trilean object.
comparator (tri) : The consequent trilean object.
Returns: `tri` Result of the implication operation as a trilean object.
method IMP_L(self, comparator)
Logical implication using Łukasiewicz's logic for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The antecedent trilean object.
comparator (tri) : The consequent trilean object.
Returns: `tri` Result of the implication operation as a trilean object.
method IMP_RM3(self, comparator)
Logical implication using RM3 logic for trilean objects.
Namespace types: tri
Parameters:
self (tri) : The antecedent trilean object.
comparator (tri) : The consequent trilean object.
Returns: `tri` Result of the RM3 implication as a trilean object.
method MA(self)
Evaluates to True if the trilean object is either True or Uncertain, False otherwise.
Namespace types: tri
Parameters:
self (tri) : The trilean object to evaluate.
Returns: `tri` Result of the operation as a trilean object.
method LA(self)
Evaluates to True if the trilean object is True, False otherwise.
Namespace types: tri
Parameters:
self (tri) : The trilean object to evaluate.
Returns: `tri` Result of the operation as a trilean object.
method IA(self)
Evaluates to True if the trilean object is Uncertain, False otherwise.
Namespace types: tri
Parameters:
self (tri) : The trilean object to evaluate.
Returns: `tri` Result of the operation as a trilean object.
UNANIMOUS(self, comparator)
Evaluates the unanimity between two trilean values.
Parameters:
self (tri) : The first trilean value.
comparator (tri) : The second trilean value.
Returns: `tri` Returns True if both values are True, False if both are False, and Uncertain otherwise.
method UNANIMOUS(self)
Evaluates the unanimity among an array of trilean values.
Namespace types: array
Parameters:
self (array) : The array of trilean values.
Returns: `tri` Returns True if all values are True, False if all are False, and Uncertain otherwise.
tri
Three Value Logic (T.U.F.), or trilean. Can be True (1), Uncertain (0), or False (-1).
Fields:
v (series int) : Value of the trilean variable. Can be True (1), Uncertain (0), or False (-1).
SML SuiteIntroducing the "SML Suite" Indicator
The "SML Suite" is a powerful and easy-to-use trading indicator designed to help traders make informed decisions in the world of financial markets. Whether you're a seasoned trader or a novice, this indicator is your trusty sidekick for evaluating market trends.
Key Features:
Three Moving Averages: The indicator employs three different moving averages, each with a distinct length, allowing you to adapt to various market conditions.
Customizable Parameters: You can easily customize the moving average lengths and source data to tailor the indicator to your specific trading strategy.
Standard Deviation Multiplier: Adjust the standard deviation multiplier to fine-tune the indicator's sensitivity to market fluctuations.
Binary Results: The indicator provides clear binary signals (1 or -1) based on whether the current price is above or below certain bands. This simplifies your decision-making process.
SML Calculation: The SML (Short, Medium, Long) calculation is a smart combination of the binary results, offering you an overall sentiment about the market.
Color-Coded Visualization: Visualize market sentiment with color-coded bars, making it easy to spot trends at a glance.
Interactive Table: A table is displayed on your chart, giving you a quick overview of the binary results and the overall SML sentiment.
With the "SML Suite" indicator, you don't need to be a coding expert to harness the power of technical analysis. Stay ahead of the game and enhance your trading strategy with this user-friendly tool. Make your trading decisions with confidence and clarity, backed by the insights provided by the "SML Suite" indicator.
fuson DEMA-->it does not need to be played with any settings. so I did not add a period.
-->l shows bearish trends very well but not as good as bearish trends in bullish trends
-->vdub can be used with binary options v3 and increases the leakage rates very high
-->If used for forex, it can be used in periods of 1 hour and longer
-->vdub binary option v3can be used for 1 minute verification with binary option if it will be used in binary option
Liquidity Void Zone Detector [PhenLabs]📊 Liquidity Void Zone Detector
Version: PineScript™v6
📌 Description
The Liquidity Void Zone Detector is a sophisticated technical indicator designed to identify and visualize areas where price moved with abnormally low volume or rapid momentum, creating "voids" in market liquidity. These zones represent areas where insufficient trading activity occurred during price movement, often acting as magnets for future price action as the market seeks to fill these gaps.
Built on PineScript v6, this indicator employs a dual-detection methodology that analyzes both volume depletion patterns and price movement intensity relative to ATR. The revolutionary 3D visualization system uses three-layer polyline rendering with adaptive transparency and vertical offsets, creating genuine depth perception where low liquidity zones visually recede and high liquidity zones protrude forward. This makes critical market structure immediately apparent without cluttering your chart.
🚀 Points of Innovation
Dual detection algorithm combining volume threshold analysis and ATR-normalized price movement sensitivity for comprehensive void identification
Three-layer 3D visualization system with progressive transparency gradients (85%, 78%, 70%) and calculated vertical offsets for authentic depth perception
Intelligent state machine logic that tracks consecutive void bars and only renders zones meeting minimum qualification requirements
Dynamic strength scoring system (0-100 scale) that combines inverted volume ratios with movement intensity for accurate void characterization
Adaptive ATR-based spacing calculation that automatically adjusts 3D layering depth to match instrument volatility
Efficient memory management system supporting up to 100 simultaneous void visualizations with automatic array-based cleanup
🔧 Core Components
Volume Analysis Engine: Calculates rolling volume averages and compares current bar volume against dynamic thresholds to detect abnormally thin trading conditions
Price Movement Analyzer: Normalizes bar range against ATR to identify rapid price movements that indicate liquidity exhaustion regardless of instrument or timeframe
Void Tracking State Machine: Maintains persistent tracking of void start bars, price boundaries, consecutive bar counts, and cumulative strength across multiple bars
3D Polyline Renderer: Generates three-layer rectangular polylines with precise timestamp-to-bar index conversion and progressive offset calculations
Strength Calculation System: Combines volume component (inverted ratio capped at 100) with movement component (ATR intensity × 30) for comprehensive void scoring
🔥 Key Features
Automatic Void Detection: Continuously scans price action for low volume conditions or rapid movements, triggering void tracking when thresholds are exceeded
Real-Time Visualization: Creates 3D rectangular zones spanning from void initiation to termination, with color-coded depth indicating liquidity type
Adjustable Sensitivity: Configure volume threshold multiplier (0.1-2.0x), price movement sensitivity (0.5-5.0x), and minimum qualifying bars (1-10) for customized detection
Dual Color Coding: Separate visual treatment for low liquidity voids (receding red) and high liquidity zones (protruding green) based on 50-point strength threshold
Optional Compact Labels: Toggle LV (Low Volume) or HV (High Volume) circular labels at void centers for quick identification without visual clutter
Lookback Period Control: Adjust analysis window from 5 to 100 bars to match your trading timeframe and market volatility characteristics
Memory-Efficient Design: Automatically manages polyline and label arrays, deleting oldest elements when user-defined maximum is reached
Data Window Integration: Plots void detection binary, current strength score, and average volume for detailed analysis in TradingView's data window
🎨 Visualization
Three-Layer Depth System: Each void is rendered as three stacked polylines with progressive transparency (85%, 78%, 70%) and calculated vertical offsets creating authentic 3D appearance
Directional Depth Perception: Low liquidity zones recede with back layer most transparent; high liquidity zones protrude with front layer most transparent for instant visual differentiation
Adaptive Offset Spacing: Vertical separation between layers calculated as ATR(14) × 0.001, ensuring consistent 3D effect across different instruments and volatility regimes
Color Customization: Fully configurable base colors for both low liquidity zones (default: red with 80 transparency) and high liquidity zones (default: green with 80 transparency)
Minimal Chart Clutter: Closed polylines with matching line and fill colors create clean rectangular zones without unnecessary borders or visual noise
Background Highlight: Subtle yellow background (96% transparency) marks bars where void conditions are actively detected in real-time
Compact Labeling: Optional tiny circular labels with 60% transparent backgrounds positioned at void center points for quick reference
📖 Usage Guidelines
Detection Settings
Lookback Period: Default: 10 | Range: 5-100 | Number of bars analyzed for volume averaging and void detection. Lower values increase sensitivity to recent changes; higher values smooth detection across longer timeframes. Adjust based on your trading timeframe: short-term traders use 5-15, swing traders use 20-50, position traders use 50-100.
Volume Threshold: Default: 1.0 | Range: 0.1-2.0 (step 0.1) | Multiplier applied to average volume. Bars with volume below (average × threshold) trigger void conditions. Lower values detect only extreme volume depletion; higher values capture more moderate low-volume situations. Start with 1.0 and decrease to 0.5-0.7 for stricter detection.
Price Movement Sensitivity: Default: 1.5 | Range: 0.5-5.0 (step 0.1) | Multiplier for ATR-normalized price movement detection. Values above this threshold indicate rapid price changes suggesting liquidity voids. Increase to 2.0-3.0 for volatile instruments; decrease to 0.8-1.2 for ranging or low-volatility conditions.
Minimum Void Bars: Default: 10 | Range: 1-10 | Minimum consecutive bars exhibiting void conditions required before visualization is created. Filters out brief anomalies and ensures only sustained voids are displayed. Use 1-3 for scalping, 5-10 for intraday trading, 10+ for swing trading to match your time horizon.
Visual Settings
Low Liquidity Color: Default: Red (80% transparent) | Base color for zones where volume depletion or rapid movement indicates thin liquidity. These zones recede visually (back layer most transparent). Choose colors that contrast with your chart theme for optimal visibility.
High Liquidity Color: Default: Green (80% transparent) | Base color for zones with relatively higher liquidity compared to void threshold. These zones protrude visually (front layer most transparent). Ensure clear differentiation from low liquidity color.
Show Void Labels: Default: True | Toggle display of compact LV/HV labels at void centers. Disable for cleaner charts when trading; enable for analysis and review to quickly identify void types across your chart.
Max Visible Voids: Default: 50 | Range: 10-100 | Maximum number of void visualizations kept on chart. Each void uses 3 polylines, so setting of 50 maintains 150 total polylines. Higher values preserve more history but may impact performance on lower-end systems.
✅ Best Use Cases
Gap Fill Trading: Identify unfilled liquidity voids that price frequently returns to, providing high-probability retest and reversal opportunities when price approaches these zones
Breakout Validation: Distinguish genuine breakouts through established liquidity from false breaks into void zones that lack sustainable volume support
Support/Resistance Confluence: Layer void detection over key horizontal levels to validate structural integrity—levels within high liquidity zones are stronger than those in voids
Trend Continuation: Monitor for new void formation in trend direction as potential continuation zones where price may accelerate due to reduced resistance
Range Trading: Identify void zones within consolidation ranges that price tends to traverse quickly, helping to avoid getting caught in rapid moves through thin areas
Entry Timing: Wait for price to reach void boundaries rather than entering mid-void, as voids tend to be traversed quickly with limited profit-taking opportunities
⚠️ Limitations
Historical Pattern Indicator: Identifies past liquidity voids but cannot predict whether price will return to fill them or when filling might occur
No Volume on Forex: Indicator uses tick volume for forex pairs, which approximates but doesn't represent true trading volume, potentially affecting detection accuracy
Lagging Confirmation: Requires minimum consecutive bars (default 10) before void is visualized, meaning detection occurs after void formation begins
Trending Market Behavior: Strong trends driven by fundamental catalysts may create voids that remain unfilled for extended periods or permanently
Timeframe Dependency: Detection sensitivity varies significantly across timeframes; settings optimized for one timeframe may not perform well on others
No Directional Bias: Indicator identifies liquidity characteristics but provides no predictive signal for price direction after void detection
Performance Considerations: Higher max visible void settings combined with small minimum void bars can generate numerous visualizations impacting chart rendering speed
💡 What Makes This Unique
Industry-First 3D Visualization: Unlike flat volume or liquidity indicators, the three-layer rendering with directional depth perception provides instant visual hierarchy of liquidity quality
Dual-Mode Detection: Combines both volume-based and movement-based detection methodologies, capturing voids that single-approach indicators miss
Intelligent Qualification System: State machine logic prevents premature visualization by requiring sustained void conditions, reducing false signals and chart clutter
ATR-Normalized Analysis: All detection thresholds adapt to instrument volatility, ensuring consistent performance across stocks, forex, crypto, and futures without constant recalibration
Transparency-Based Depth: Uses progressive transparency gradients rather than colors or patterns to create depth, maintaining visual clarity while conveying information hierarchy
Comprehensive Strength Metrics: 0-100 void strength calculation considers both the degree of volume depletion and the magnitude of price movement for nuanced zone characterization
🔬 How It Works
Phase 1: Real-Time Detection
On each bar close, the indicator calculates average volume over the lookback period and compares current bar volume against the volume threshold multiplier
Simultaneously measures current bar's high-low range and normalizes it against ATR, comparing the result to price movement sensitivity parameter
If either volume falls below threshold OR movement exceeds sensitivity threshold, the bar is flagged as exhibiting void characteristics
Phase 2: Void Tracking & Qualification
When void conditions first appear, state machine initializes tracking variables: start bar index, initial top/bottom prices, consecutive bar counter, and cumulative strength accumulator
Each subsequent bar with void conditions extends the tracking, updating price boundaries to envelope all bars and accumulating strength scores
When void conditions cease, system checks if consecutive bar count meets minimum threshold; if yes, proceeds to visualization; if no, discards the tracking and resets
Phase 3: 3D Visualization Construction
Calculates average void strength by dividing cumulative strength by number of bars, then determines if void is low liquidity (>50 strength) or high liquidity (≤50 strength)
Generates three polyline layers spanning from start bar to end bar and from top price to bottom price, each with calculated vertical offset based on ATR
Applies progressive transparency (85%, 78%, 70%) with layer ordering creating recession effect for low liquidity zones and protrusion effect for high liquidity zones
Creates optional center label and pushes all visual elements into arrays for memory management
Phase 4: Memory Management & Display
Continuously monitors polyline array size (each void creates 3 polylines); when total exceeds max visible voids × 3, deletes oldest polylines via array.shift()
Similarly manages label array, removing oldest labels when count exceeds maximum to prevent memory accumulation over extended chart history
Plots diagnostic data to TradingView’s data window (void detection binary, current strength, average volume) for detailed analysis without cluttering main chart
💡 Note:
This indicator is designed to enhance your market structure analysis by revealing liquidity characteristics that aren’t visible through standard price and volume displays. For best results, combine void detection with your existing support/resistance analysis, trend identification, and risk management framework. Liquidity voids are descriptive of past market behavior and should inform positioning decisions rather than serve as standalone entry/exit signals. Experiment with detection parameters across different timeframes to find settings that align with your trading style and instrument characteristics.
