Adaptive Volume Confluence OscillatorWhat it is
One pane that fuses seven different reads of the bar into a single 0–100 confluence score, gates that score by a trend-vs-chop regime filter, confirms it against an auto-mapped higher timeframe, and — most importantly — forward-calibrates its own Buy/Sell signals against an unconditional base rate, so you can see whether the construction actually carries an edge on your instrument.
The seven votes: momentum sign · momentum vs its signal · money flow · trend structure (MA fan) · price location vs VWAP · trend slope · higher-timeframe bias.
The displayed wave is a volume-flow ribbon; the votes drive the score, the signals and the verdict. A plain-language verdict and a subtle pane tint make it readable at a glance (Simple view); a full analytic layer is available for advanced users (Pro view).
Why these are combined (mashup rationale)
A single oscillator whipsaws and a single signal over-fires. Combining helps only when the inputs key on different quantities and their agreement is checked. Each vote reads a different thing — momentum, momentum-vs-signal, volume flow, multi-MA structure, location vs a session mean, slope, and a higher-timeframe read — so the count that agrees carries more information than any one of them alone. A Kaufman Efficiency-Ratio regime gate suppresses conviction in chop, and a forward-calibration harness ties the whole construction back to realised forward outcomes.
An honest caveat, stated up front: the votes are not statistically independent. The oscillator itself embeds money flow, and vote 2 is derived from vote 1's series. Treat the score as a weight-of-evidence read, not as seven independent confirmations. The harness exists precisely so you can check whether the construction earns its keep on your instrument rather than taking the claim on faith.
How it works
Score — how many of the seven votes are bullish, scaled 0–100.
Regime — Kaufman Efficiency Ratio. Below the chop threshold, conviction dims, signals are withheld, and the verdict reads "WAIT – choppy".
HTF — the chart timeframe auto-maps to a confirming higher timeframe (~4–6×), requested with lookahead_off and offset by one bar while the live bar forms.
Signals — Buy/Sell fire only when the oscillator crosses its signal at a statistical OB/OS extreme and the score agrees and the regime isn't choppy and the visible wave isn't already at the opposite extreme.
Climax — a volume spike at an OB/OS extreme prints a Possible Bottom/Top exhaustion mark.
Divergence (Pro) — regular + hidden, from confirmed pivots on the momentum oscillator.
Calibration — each Buy/Sell is queued and resolved a fixed horizon later, then compared with the unconditional same-horizon base rate. The dashboard shows, per side: Hit %, Edge = Hit − Base, sample size, and a Wilson-gated star.
How to use it
Read the verdict and the score. Above the gate = bullish weight of evidence; below = bearish; in between, or in chop, the tool says WAIT — and it means it.
Treat Buy/Sell marks as context, not triggers. They already require the score, the regime and the wave to agree, but they remain a description of conditions — not a recommendation.
Read the Edge row before you weight any signal. If Buy/Sell Edge isn't clearly positive with an adequate sample and a star, this construction is not carrying an edge on this instrument — weight it down or ignore it. Do not tune the parameters until the Edge turns green: that is curve-fitting, and the harness is there to catch it, not to be defeated.
Combine with your own levels, structure and risk rules.
Universal across markets
Price / high / low are inputs, so the engine runs on any symbol or timeframe. The volume votes (money flow, climax, VWAP location) need real volume — prefer a futures contract or a stock. On a symbol with no volume the tool degrades gracefully: money flow is neutralised, the score falls back to the price-only votes, and the dashboard says "NO VOLUME", so you're never misled by a blank or a phantom reading.
Non-repainting
Votes read confirmed closes. The HTF series uses lookahead_off and is offset by one bar while the live bar forms. Divergences come from ta.pivot* and confirm a few bars after the pivot; once printed they don't move. The calibration harness logs and resolves only on confirmed bars, so its statistics never inflate intrabar. The live oscillator updates each bar, like any oscillator.
Concept credits
Super Smoother and Ultimate Smoother low-lag filters — John Ehlers. Chebyshev Type-I filter — classical DSP. Recursive (Kalman) smoothing — R. E. Kalman. Volume Zone Oscillator — Walid Khalil & David Steckler. Accumulation/Distribution money-flow multiplier — Marc Chaikin. Efficiency Ratio — Perry J. Kaufman. ATR — J. Welles Wilder. Wilson score interval — Edwin B. Wilson. VWAP, Hull MA and percentile rank — standard public methods.
Original implementation; not affiliated with, nor endorsed by, any third party. No third-party code is reused.
Honest limits
The score is context, not a guarantee, and the votes are correlated (see the caveat above). The Edge figures are in-sample, close-to-close, with overlapping forward windows and no costs — descriptive context, not a verified backtest. An Edge near zero, negative, or unstable across timeframes is the harness honestly telling you the signal has no reliable edge on that instrument. Nothing here predicts price.
Disclaimer
Research and educational tool only. Not financial advice and no guarantee of profitability or accuracy. Indicators describe past behaviour; they do not predict the future. Trading carries risk of loss. Test out-of-sample and make your own decisions. The author accepts no liability for any use of this script.
Indicador






















