Adaptive Kernel Flow RLS Strategy
Family: trend_following · Regime: trending · Complexity: high · Asset classes: FX, Equities, Crypto · Timeframes: H1, H4
Thesis
This strategy assumes that high-conviction trends are characterized by 'volume turnover cycles' (Float) and multi-timeframe structural alignment. By using adaptive filters (RLS) instead of static averages, we can identify pullbacks (Stochastic) within a trend with lower lag, while the Nadaraya-Watson kernel provides a non-parametric confirmation that the underlying 'fair value' trend is still intact. The edge exists in filtering noise through dual-domain confirmation (Adaptive Filter + Kernel Regression).
Components
- Multi Timeframe Moving Average Dashboard (regime) — Establishes a 'Global Regime' by ensuring the majority of the SMA stack across H1-D1 timeframes is aligned, reducing the probability of trading against structural flow.
- Recursive Least Squares Adaptive Filter (RLS) (direction) — Acts as the primary trend direction filter. Unlike static SMAs, the RLS adapts to non-stationary price signals with minimal phase lag.
- Stochastic Oscillator (entry) — Identifies local oversold/overbought conditions to time entries on pullbacks within the established adaptive trend.
- Bollinger Bands (Standard) (exit) — Provides a volatility-based target for profit-taking when price reaches an statistical extreme relative to the short-term mean.
- GAS Price Levels (risk) — Utilizes dynamic structural levels (KAMA/BB hybrid) to determine logical stop-loss placement and size positions based on current ATR.
- Nadaraya-Watson Kernel Regression (confirmation) — Provides a non-parametric confirmation of the trend. The Gaussian kernel smooths noise without the extreme lag of high-period moving averages.
- Float Indicator (volatility_filter) — Ensures the current price move has not yet exhausted its cumulative volume turnover, filtering out late-stage trend entries.
Known failure conditions
- Numerical instability in the RLS P-matrix causing the filter to 'explode' or return NaN values.
- Extended sideways price action where the NW Kernel slope oscillates frequently around zero.
- Market regimes where volume (Float Indicator) does not correlate with price expansion (e.g., low-liquidity gap environments).
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.