Lab
Trading Lab
A futures-trading research pipeline built to kill bad strategies before they cost money: pre-registered specs, multi-gate walk-forward validation, and a Deflated-Sharpe overfitting verdict engine. Thirteen candidates tested, thirteen killed, zero dollars risked. The asset is the discipline, not a holy grail.
01Gates, not vibes
Every strategy is spec’d and git-frozen before its first backtest, then walked through gates: a three-year cost-aware backtest, a one-shot walk-forward, and only then paper trading. A separate verdict engine runs Deflated Sharpe Ratio and bootstrap checks against overfitting.
02The results table is all red — and that is the point
ORB v1 failed Gate 1. ORB-D passed provisionally, then failed the walk-forward and was marked dead. MIM-30 failed. An overnight batch of ten pre-registered strategies produced zero survivors. No live account was ever opened; nothing was risked on an unproven edge. A paper-only forward-test harness with prop-firm risk enforcement runs the next candidates.