FORWARD TRIAL
The paper clock continues
Existing paper period recorded since 2026-07-23. 3 closed trades entered before that date; their whole-trade returns are shown in history but are separate from complete forward trades. Session evidence: 63 recorded at the time, 12 recovered later, 6 missing. Earliest contemporaneous paper-state checkpoint retained in GitHub; earlier VPS start is unknown.
Complete forward trade metrics: win rate — · mean excess versus SPY —.
Historical validation
Compare fixed rules on earlier and later windows in Research. Historical tests help evaluate generalization; reviewed test windows become exploratory.
Compare chronological windows →Forward observation
The paper period keeps the stored path, including recoveries. Upgrades preserve its start; complete forward trades and coverage gaps stay separate.
How the evaluation checkpoint works
Evaluation is due after 30 complete forward trades or 9 recorded months. At that checkpoint, negative mean trade excess versus SPY or a win rate below 40% produces a halt; excess above 2% and wins above 52% produces a pass. Relative drawdown below −15% over the trailing 126 sessions produces a halt at any time. A recent recovery gap requires review before a funding pass. These checkpoints are decision rules, not statistical proof of an edge. Trades opened before the recorded start remain in history; their full returns are not attributed entirely to this period.