Model white papers and methodology documentation
Rolling OLS regression estimates fair value from dynamically discovered correlated drivers across 16 FX pairs. Trades z-score deviations with rapid exit on reversion. Margin-based sizing, volatility filtering, and regime-conditional entry gating.
Detects accumulated capital flow through EMA crossover on cumulative beta-adjusted residuals against the S&P 500 driver. 10 FX pairs, 1-minute data, US session. MC-validated regime gating across 120 pair-regime-direction combinations.
Four signal types (Bollinger, RSI, Range, Session) assigned per pair through walk-forward validation. Trade management adapted per volatility regime. 12 FX pairs with expanding-window walk-forward and Monte Carlo validation.
Multi-dimensional market environment classifier: per-pair ATR regimes (QUIET through STRESS) with direction tracking, global VIX/DXY/dispersion indicators, Schmitt trigger anti-flicker. Shared across all trading systems, improves Sharpe 30-40%.
50+ autonomous agents across technical, fundamental, flow, and cross-asset categories. Bayesian linear regression learns which agents predict which pairs at which horizons. Regime-conditional models with FV, ER, and RT signals integrated.