Research repository · receipts for everything
Every test we've run — including the failures.
Passed, failed, null, archived — every study is listed here with its full receipt: sample size, dataset and date range, method, the engine file that ran it, the SHA256 of the actual source artifact, and the data itself to download. You shouldn't have to believe us about anything on this site — you should be able to check.
Run like a repository, not a marketing page: verdicts are append-only, nulls are first-class results, and a claim without its n does not ship.
5-gate gauntlet (quant/lab/validation_engine.py): sample ≥100 · permutation p<0.05 · deflated Sharpe ≥0.95 vs configs searched (Bailey & López de Prado 2014) · drop-best-5% expectancy >0 · minus-best-era still significant. Ledger is append-only. VALIDATED count is currently ZERO — that is the standard working, not a gap in it.
Education & analytics only — not signals, not advice, not FCA-regulated. Past results, including tested failures, do not indicate future performance. Figures are measured from real historical data with the sample size shown; where a result is qualitative — or a field was never recorded — we say so rather than invent a number.