Not All Hours Are Created Equal: Session-Based Entry Filtering

The Session Effect

Markets have distinct personality changes across trading sessions. London session EURUSD behaves differently from Asian session EURUSD. S35 tracks win rate, average R, and signal quality by hour of day and trading session for each instrument.

The findings are clear: FOREX instruments perform best during London and London/NY overlap (6:00-16:00 UTC). Asian session signals have a 54% win rate versus 62% during London. INDEX instruments are best during their home market sessions. CRYPTO shows the least session dependency because it trades 24/7 with more distributed liquidity.

How S35 Filters

For each instrument-session combination, S35 maintains a rolling win rate. If the trailing 100-trade win rate for a specific session drops below 50%, new entries during that session are blocked. This is a dynamic filter. Sessions that underperform get shut off. Sessions that recover get restored.

In practice, Asian session forex is blocked about 70% of the time. London/NY overlap is never blocked. The dynamic nature means the system adapts to structural changes in session behavior rather than relying on static rules that might become outdated.

The 8-Point Win Rate Edge

The 8-point gap between session-filtered (62%) and all-hours (54%) win rate is substantial. It means that session timing is one of the most impactful filters in the entire system, comparable to S04's ADX filter and S28's multi-timeframe confluence. The practical effect is that V7 concentrates trading during the most liquid, most predictable hours. This aligns with the system's design philosophy: be patient, be selective, and only trade when conditions favor the strategy. Session filtering removes the worst trading hours and lets the model operate where its training data is most representative of live conditions.