Strategies
Gordon runs a multi-strategy trend-following portfolio. Multiple uncorrelated strategies run simultaneously, each capturing different market phases. Drawdowns do not fully overlap, and the combined portfolio achieves a higher Sharpe than any individual strategy.
All strategies share a common execution pipeline:
- Entry signal — strategy-specific breakout or crossover logic
- Signal overlays — CVD veto, OI regime, funding z-score
- Sentiment regime — Fear & Greed, correlation density, liquidation risk, SSR
- Volatility targeting — position sized by
target_vol / realized_vol - Risk management — five circuit breakers, VPIN kill switch, mandatory stop loss
Active strategies (validated)
These three passed the full promotion flow: Level 1 signal quality, Level 2 realistic simulation, walk-forward validation, ablation analysis, and multi-asset validation across BTC, ETH, and SOL.
| Strategy | Type | Timeframes | Sharpe | Status |
|---|---|---|---|---|
| Supertrend 3.0 | ATR-based trend following | D1, 8h | 0.90 | Validated |
| EWMAC Ensemble | Moving average crossover (4 speed pairs) | 6h, 4h, 2h, 1h | 0.90 | Validated |
| PSAR Tight | Parabolic Stop & Reverse | W1, D1 | 0.69–1.43 | Validated |
Strategies under validation (gordon-strategy 4.6.x)
These eleven strategies were added in the 4.6.x development cycle. All are implemented and registered in StrategyRegistry, available for backtesting. Walk-forward validation is in progress.
| Strategy | Type |
|---|---|
| BasisRevertStrategy | Futures basis mean reversion |
| BollingerBreakoutStrategy | Bollinger Band volatility breakout |
| FundingArbStrategy | Funding rate arbitrage signal |
| HurstRegimeSwitcherStrategy | Hurst exponent regime detection |
| OpeningRangeBreakoutStrategy | Session opening range breakout |
| PairsMrStrategy | Cross-asset pairs mean reversion |
| RsiMeanReversionStrategy | RSI-based mean reversion |
| TsMomStrategy | Time-series momentum |
| VolumeSpikeBreakoutStrategy | Volume spike confirmation breakout |
| VpinImpulseStrategy | VPIN toxic-flow impulse |
| VwapStretchMrStrategy | VWAP deviation mean reversion |
None of these are currently deployed to production. Promotion to live requires the full validation gate — see Adding a Strategy.
Retired
| Strategy | Sharpe | Reason |
|---|---|---|
| Donchian Ensemble | 0.08 (median) | Retired in v4 retrospective — too thin an edge to justify complexity in the portfolio |
Validation gate
Every strategy must pass before promotion to live:
- Walk-forward testing: Sharpe > 0.5, consistency > 50% of out-of-sample windows
- Maximum drawdown < 30%
- Ablation analysis: every component must demonstrate lift
- Multi-asset validation: BTC, ETH, SOL minimum
- Binance testnet validation before any live capital is risked
See Adding a Strategy for the full flow from implementation to production.