Okay, this time I'm going to show one pair where this setup performs in an awesome manner and how the algo does it...
First, take a look at the SCSM III, settings 80,2,30 making it more responsive short term, use what numbers work best for your strategy or your pairs. The strategy presented does the following: When the pair crosses over or crosses below it opens or sell, no other condition is taken into account ( in real life it would be better to do Multiple time frame analysis or to avoid selling in overbought markets or oversold markets, probably it would be best...) Now, the pair crosses and we open the trade long or short, this particular case has 2500 points SL (a HUGE SL) and a trailing stop of 350 points that will activate once 400 points of profit are made. This marks the start of the trade, now, once the algorithm detects that a candle has closed with profit, it moves the SL to -20 points, meaning if price reverses it would hit the SL with a small win and we would break even and call it a day (I want to experiment a bit with this setup, but it's been proving fairly hard so far to code in pine) That's it, the algo does nothing else, NOTHING, if price crosses over again in the other direction it won't do anything, anything at all, just wait until either the TP or SL get hits (it would be interesting to test hedging going in both directions as the crossover hits, but I can't do it in Pine, sorry) See in the chart how I marked three bad longs and 2 bad shorts that broke even with the one candle of profit rule? See how in the last one to the right the algorithm bought at the close of a shooting star and had to wait 3 days to breakeven. See also how it sold and then got stopped out at the end of the rally for a very bad trade (this is one of these cases where hedging would have been useful, since we would have gone long on the crossover too for even more profit than the loss).
Now for some math, 2500 points of SL for a trailing stop activation of 350 points seems hardly good, too small of a risk reward?, 97.8% chance means a LOT of more 350 than the occasional 2500, add to that the fact that the algo completely ignores other factors such as overall trend or news and I'm sure we could get somewhere near 99%
Now as for some disclaimers: This algo uses 1:1 leverage, overleverage and overtrading kills any algo, period. This algo doesn't account for spread, but it should take out a substantial chunk of profit, lowering it to somewhere near 26 or 27% This algo doesn't account for lack of liquidity or precise slippage, this isn't so important when trading bigger time frames but it should also reduce (or increase) the net profit. This algo doesn't repaint, look at the same algo in 30 min charts: The idea would be the same but 2500 points of SL is too much for 30 min charts and 400 points of profit is too much yet again... we would need to adjust these numbers to more realistic ones... In this case the algo runs for only 3 years, I've done test on daily candles all the way to 1990 and the results are less spectaculars, but still good.
Ways to improve it: 1- Hedging (not tested) 2- Buy or sell only when there is a certain separation between bars after the crossover (but only near the crossover, not 10 or 12 bars later, that IS unprofitable, tested already) 3- Multiple time frame analysis so you ignore trades against the prevailing trend in the higher TF 4- Range detection methods so you avoid selling at the bottom and buying at the top 5- Fundamental analysis and news to avoid those nasty spikes 6- Have luck, patience and a lot of trust in you and your strategy...
Just remember in order to mantain a 2500 points SL, you need no more than 1:3 leverage or you'll blow your account sooner than later, just saying, 97.8% means it fails 2 out of 100 times, if those two take out 50% of your account, then you're done...
Good luck to everyone, hope this helps get better results in your trading.
Comentário
⋅
Just noticed that when posting a snapshot the strategy results aren't posted, in the 30 min charts the algo is profitable 96.8% of the time but has a net profit of -1.8% due to excessive size of SL and TP.
Looks really interesting. I'd be interested in checking out the algo if you ever decide to release it.
Hope you don't mind me adding this here, but I've been working on a semi-similar idea for a few days since you released the new SCSM. After studying it I noticed that crossovers were powerful, but like any indicator it can struggle in ranging/whipsaw action. So I wanted to come up with some sort of simple strength based indicator that was still based on the relationship between the two currencies as plotted by the SCSM. I mashed together an indicator I'm fond of (Money Flow Index), and came up with this. It needs some tweaking as it lags by a bar or two to my liking, and I'll need to come up with some rules around it, but I like the simplicity of it so far, and it provides some interesting signals (which I've pointed out in the screenshot).
Let me know your thoughts! Happy to send you a copy if you're interested. Thanks :)
DreamsDefined
⋅
@DreamsDefined, That was literally the first currency and timeframe I look at by the way. Haven't really cherry picked the results.
DreamsDefined
⋅
There's a typo in the bit around retracements not reaching 60, that should be 80 :)
EilderJorgeGarcia
⋅
@DreamsDefined, Thanks for the comments, and yes, I would like to see it, I'm probably not going to be around TV much time next week, but it looks good, what I'm doing now is picking the crossover when it has a bit of momentum and crosses a small threshold, probably similar to the idea you're using of price not reaching 40 so the movement should continue up... I was trying to make a divergence indicator as well lately, but have failed miserably two times already, yours is looking good, probably better than mine, we could improve on your idea and see if we can fine tune and make it useful for all TF and pairs.
So far the best way I've found to resist ranges and whipsaws are BB and I'm going to test hedging once I come back from my vacation. OANDA allows for hedging inside the same pair, so I'm going to use their demo account to test the possibility of going both directions at the same time... with the rules defined above it SHOULD:
- Profit in sideways markets
- Profit during spikes
- Profit during range markets
- Profit during breakouts
-Small loss on interest for open positions (depending on your TF)
-Loss on small rallies that take out an SL but don't continue much longer, limiting the profit of the position in the same direction
-Loss on trades that trigger trailing stops too early but price continues taking the bigger SL.
@Bakker80, I'll continue work on it this week to make sure it's compatible across timeframe and markets. Keep an eye out on my profile or I'll PM you once it's done.