Loading track record...
Public Proof
Historical alert outcomes across exact forward windows. Missing price coverage is counted per horizon, not hidden.
Self-managed health check on every score the dashboard surfaces. Recomputes win-rate-by-bucket every hour from forward-returns data and flags drift before it shows up in alerts.
| bucket | n | avg 24h | median | win rate |
|---|---|---|---|---|
| 80+ | 35 | -1.91% | -3.62% | 37.1% |
| 70-79 | 115 | -5.29% | -4.10% | 37.4% |
| 60-69 | 254 | 5.24% | -0.34% | 46.1% |
| 50-59 | 789 | 1.17% | -0.97% | 43.1% |
| 30-49 | 2,195 | 0.31% | -1.09% | 38.5% |
| <30 | 1,647 | 0.13% | -0.50% | 30.5% |
| bucket | n | avg 24h | median | win rate |
|---|---|---|---|---|
| noise | 66,558 | 0.32% | -5.83% | 37.5% |
Long horizons stay visible with their own sample badge so sparse data does not masquerade as a mature proof set.
INTERNET is the cleanest internal example of why the product exists: multiple exchange drain signals, clustering context, and a major subsequent move.
| Alert | Count | Coverage 24h | Win 24h | Median 24h | IQR 24h |
|---|---|---|---|---|---|
| squeeze | 167400 | 95.6% | 58.5% | 2.9% | -6.8% / 21.6% |
| score_mover | 50162 | 97.9% | 48.7% | -0.2% | -18.1% / 22.5% |
| supply_change | 34948 | 76.2% | 34.3% | -1.3% | -6.6% / 1.9% |
| fresh_setup | 5296 | 99.7% | 41.8% | -7.2% | -30.1% / 22.1% |
| exchange_cliff | 605 | 99.7% | 27.2% | -29.1% | -51.0% / 2.5% |
Performance breakdown by SSS stage at alert time
Performance breakdown by entry timing at alert time