// Methodology · Whitepaper
The Confluence Terminal: a dual-gate approach to pattern qualification
Why we separate direction from quality — and how a two-independent-gate design filters weak setups before they ever reach the trader.
Abstract
Most signal tools compress everything into a single blended score, hiding the trade-off between where a move is going and how clean it is. TradrQuant instead runs two fully independent gates — a 0–6 Confluence score for structural alignment and a 0–100 Quality score for the character of the move. A setup must clear both before it prints. This paper explains the design rationale, each gate's inputs, and why decoupling direction from quality produces a more honest, tunable qualification pipeline.
01The problem with a single score
A blended "AI score" answers a question no trader actually asked: it fuses direction, momentum, volume and cleanliness into one number, then asks you to trust the weighting. When that number is 78, you cannot tell whether the setup is strongly directional but choppy, or clean but barely trending. The weighting is hidden, fixed, and unfalsifiable.
Our starting premise is that direction and quality are different axes and should be measured — and shown — separately.
02Gate one — Confluence (0–6)
Six independent structural checks, each contributing at most one point:
- ›EMA trend direction (fast/slow ribbon alignment).
- ›Major-trend alignment against the 200-EMA bias filter.
- ›Quant Force baseline direction (see the Quant Force paper).
- ›RSI momentum side (with, not against, the setup).
- ›Volume expansion versus recent average.
- ›Non-overextension against the reversal bands (not already stretched).
03Gate two — Quality (0–100)
The Quality gate ignores direction entirely and measures how clean the move is: momentum strength, current volume ratio versus its own average, candle body-to-range ratio (conviction versus indecision), and baseline efficiency. A strong-but-messy move and a clean-but-weak move can share the same Confluence score yet score very differently on Quality — which is exactly the information a single number destroys.
04Why two gates beat one
Independent gates are debuggable. If your results skew toward false starts, you raise the Quality floor; if you are missing clean continuations, you relax Confluence. With a single blended score you can only turn one dial and hope. Two gates also make the tool honest: every printed setup can state exactly which checks it passed, so the trader grades it rather than obeys it.
05Tuning the gates
The thresholds are exposed on purpose. Higher-volatility instruments (crypto alts) generally warrant a higher Quality floor to filter noisy false starts; smoother instruments (indices intraday) can run a lower Confluence minimum. The engine ships with sensible per-asset starting points, but they are a starting line, not an optimisation.
// Key findings
- 01Direction and quality are distinct axes; blending them into one score discards actionable information.
- 02Confluence (0–6) measures structural alignment; Quality (0–100) measures the cleanliness of the move.
- 03Independent gates are individually tunable and individually explainable.
- 04A setup prints only when both gates clear — and, optionally, multi-timeframe agreement.
Frequently asked
Is the Confluence score an AI model?
No. It is six deterministic, rule-based structural checks that each contribute one point. There is no opaque model weighting — you can see exactly which checks passed.
Why 0–6 and 0–100 instead of one number?
Because direction and quality are different questions. Keeping them separate lets you see a clean-but-weak setup differently from a strong-but-choppy one, and lets you tune each gate independently.
Can I change the thresholds?
Yes. Both the minimum Confluence and minimum Quality are exposed inputs, with per-asset-class starting points documented in the indicator.
// References & further reading
- Bulkowski, T. — Encyclopedia of Chart Patterns / thepatternsite.com (pattern reliability statistics).
- Wilder, J. W. — New Concepts in Technical Trading Systems (RSI, ATR).
- Kaufman, P. — Trading Systems and Methods (efficiency & adaptive baselines).
See the engine in action. Meet the Scanner.
Explore the Scanner// Related papers
// Educational & methodology content only — not financial advice. Figures are conceptual and illustrative and do not represent verified trading results. Past performance does not guarantee future results.