Backtest
Compare historical composite signals vs. benchmarks.
How to read these numbers. Every methodology IC and hit rate below is computed against snapshots scored with latest-restated fundamentals (not as-of-knowable). That inflates the absolute predictive numbers across all methodologies by a similar amount, so the relative ranking between methodologies is still informative — but treat the absolute values as upper bounds. PIT-clean backtests will become available approximately 12 months after PIT data collection began.
Pre-cutover window uses non-PIT data. Snapshots before 2026-04-19 were scored against the latest restated fundamentals, not the as-of-knowable figures. Backtest returns on that portion may be inflated by restatement bias. See ADR-001.
No predictive signal at 90 days. PIT-corrected validation (2026-05-02) found a composite decile spread of −6.25pp at 90 days, non-monotonic ordering, and only 2 of 7 methodologies producing positive IC on the US ≥$1B universe 2020–2025. The current methodology composite does not predict forward returns. Methodology freeze in effect until 2026-07-31. See PIT validation report and decision rule.
CAGR
…
Strong Buy tier
Sharpe Ratio
…
Strong Buy tier
Max Drawdown
…
Strong Buy tier
Win Rate
…
Strong Buy hit rate
Signal Tier Performance
Average return, hit rate and W/L ratio by composite signal tier
Methodology Ranking
Sorted by Information Coefficient (IC) — higher is better
Performance Over Time
Weekly hit-rate trend for Strong Buy and Hold signal tiers
Loading chart…
Suggested Composite Weights
Optimised methodology weights based on backtest IC history