Statistical Trend-Pullback Hybrid (STPH)
Family: pullback · Regime: trending · Complexity: high · Asset classes: Equities, Forex, Crypto Pairs · Timeframes: H1, H4
Thesis
Markets exhibit short-term mean reversion (oversold/overbought) within long-term trends. By combining a structural trend filter (t_ma/BB) with a statistical relative-value filter (AI Pair Tool), we can enter pullbacks only when the asset is also statistically 'cheap' relative to a benchmark, increasing the probability of a trend resumption.
Components
- Bollinger Bands (regime) — Acts as a regime filter; only allows long trades when price is above the midline (SMA) to ensure a bullish volatility environment.
- t_ma (Smoothed LWMA) (direction) — Provides the primary trend direction by requiring Buffer 1 (LWMA) to be above Buffer 2 (Smoothed) for longs.
- Williams %R (entry) — Identifies short-term oversold/overbought conditions within the established trend. Entry occurs on the cross back from extremes.
- Risk Reward Indicator Tool (exit) — Determines the exit price levels (Take Profit) based on a fixed 2:1 ratio relative to the Fractal-based stop loss.
- Williams Fractals (risk) — Defines objective structural support and resistance for hard stop-loss placement, accounting for the 2-bar lag.
- AI Pair Trading System v1.0 (volatility_filter) — Used as a volatility/correlation filter; trades are only valid if the Z-Score of the base asset vs. its primary benchmark is within a mean-reverting threshold (+/- 2.0).
Known failure conditions
- If the Williams Fractals consistently result in stop losses greater than 5% of the account balance due to high volatility.
- If the 't_ma' 11-bar hardcoded limit causes the EA to stop updating signals during extended trend periods.
- If the Kalman Filter Z-score remains pinned at extremes (>3.0) for more than 50 bars, indicating a decoupling of the pair relationship.
Explore the full interactive blueprint, parameter ranges and evidence on WOBR StrategyVerse, or generate this strategy as an MT4/MT5 Expert Advisor with QuantMogul AI Engine.