Kalman-HP Cyclical Trend System
Family: hybrid · Regime: trending · Complexity: high · Asset classes: Crypto/Altcoin Spreads, Major Forex Pairs, Equities/Indices · Timeframes: 1H, 4H, 1D
Thesis
Market trends are most tradable when the relative-value relationship between the asset and its primary benchmark (e.g., ETHBTC) is in a state of equilibrium (Z-score < 2), suggesting that price movement is driven by systematic trends rather than idiosyncratic shocks. By entering only when cyclical momentum (Bandpass Filter) aligns with a structurally decomposed trend (HP Filter) and high momentum force (TDFI), the strategy captures high-probability trend continuations.
Components
- AI Pair Trading System v1.0 (regime) — Acts as the global regime filter; trades are only permitted when the relative value relationship between the target and a benchmark (e.g., ETH vs BTC) is within statistical norms (Z-Score < 2), indicating a stable market structure for trend-following.
- Hodrick-Prescott Filter (HP) (direction) — Provides the underlying structural trend direction by decomposing noise from the macro-cycle without the look-ahead bias of traditional HP filters.
- Bandpass Filter (BPF) (entry) — Identifies the specific cyclical 'swing' within the HP trend, entering as price momentum enters the preferred frequency band (the zero-cross).
- Parabolic SAR (exit) — Provides a time-price trailing exit that accelerates to lock in profits as the cyclical swing reaches maturity.
- Extrem SuperTrend (risk) — Used for dynamic stop-loss placement and position sizing based on ATR-derived volatility bands.
- Trend Direction and Force Index (TDFI) (confirmation) — Filters out weak cyclical moves by ensuring the trend intensity (cubed momentum) is high enough to overcome the 'dead zone'.
Known failure conditions
- The ML Score remains below 0.5 for extended periods while price trends, indicating the AI model cannot find patterns in the current volatility regime.
- The HP Trend and BPF cycle diverge consistently (e.g., price making new highs while HP trend flattens), suggesting the causal approximation is failing.
- Realized spread volatility exceeds the Z-score thresholds consistently, leading to 'regime lockout'.
Explore the full interactive blueprint with parameter ranges and evidence on WOBR StrategyVerse, or generate this strategy as an MT4/MT5 Expert Advisor with QuantMogul AI Engine (free download).