Forward-tested leaderboard

Only strategies that were validated with a held-out out-of-sample window(walk-forward or CPCV) appear here. Single-pass backtests don't — the number you'd see would be the one that's already been optimised against. Rows are bucketed by symbol and period so you're always comparing like with like, and each Sharpe carries a 95% confidence interval so rows that are statistically tied with the leader are visually grouped.

0 published strategies considered

0 excluded for missing out-of-sample data

0 ranked · 0 demoted for < 10 OOS trades

No forward-tested strategies yet

Strategies need a walk-forward or CPCV run with at least 10 out-of-sample trades to appear here. Run a walk-forward, publish to the marketplace, and you'll show up.

Methodology: out-of-sample Sharpe = mean of per-window Sharpes across the walk-forward or CPCV folds reported by scripts/run_backtest.py. 95% CI uses the Lo (2002) asymptotic formula 1.96 × √((1 + 0.5·SR²) / N) where N is the OOS trade count. Buckets are symbol × period(year); ranks are per-bucket. Refreshes every 5 minutes.

Want to compete? Run a walk-forward, publish your strategy, and your row appears here once a 10+ trade OOS window exists.