Gotobi Tokyo Fix

Methodology

Genetic optimization on the in-sample window with a net-of-costs robustness criterion; parameter-plateau selection (top candidates must have profitable neighborhoods, not just a best pass); out-of-sample window run once per frozen candidate on real ticks; verdicts gated on net profit factor, drawdown, and IS-to-OOS retention.

Key Takeaway

Three lessons. First, the anomaly exists: every optimized configuration was profitable in-sample (PF 2.0+ net of costs), consistent with the academic literature. Second, in-sample robustness is not enough - the 4h-before-fix entry sat on a genuine parameter plateau yet lost money out-of-sample (PF 0.79), and the tight-stop variant decayed to breakeven. Third, the 9h-before-fix entry kept a real out-of-sample edge (PF 1.36, +21.7% over 4 years at 1% risk per trade, 6.8% max drawdown, still profitable with commission stressed to $8.50/side) but retained only 67% of its in-sample profit factor against our 70% bar. We reject strategies that degrade that much rather than rationalize them - the surviving configuration goes to a demo forward test, the only validation that cannot be overfit.

Disclaimer: Past performance is not indicative of future results. These backtest results are based on historical data with realistic commission assumptions. Real trading involves additional risks including execution delays, variable spreads, and emotional decision-making.