Public Track Record
Every signal is logged to an append-only ledger with its realized outcome. We publish the hit rate, the calibration, and the misses, not just the wins. The ledger fills as live signals resolve.
Sample-size note: 62 resolved calls so far. We don't treat any hit rate as meaningful until at least 200 have resolved; below that, chance alone moves it by several points.
Dot size = number of resolved calls behind that point (n=). A point built on 5 calls means less than one built on 50 — we show both honestly.
We surface losing calls, not just winners. This is the honesty competitors hide.
- HALO: -13.8% vs benchmark -1.0% over 20 sessions
- PSX: -14.4% vs benchmark -3.6% over 20 sessions
- PLTR: -23.3% vs benchmark -5.6% over 20 sessions
- OKTA: -27.3% vs benchmark -5.6% over 20 sessions
- VLO: -18.7% vs benchmark -4.8% over 20 sessions
| Score | Resolved | Hit rate | Mean excess |
|---|---|---|---|
| 70–79 | 54 | 27.8% | −8.59% |
| 80–89 | 8 | 50.0% | +3.14% |
0 calls closed as misses because the price stopped updating (halts, delistings, data failures). We count them; we don't drop them. 0 calls are past due and awaiting a fresh price.
A call is logged when a stock's displayed Signal Score reaches 70, recording the probability we assigned, the entry price and the S&P 500 and sector-ETF baselines. Entry is measured at the next session's opening price after the call is logged (falls back to the logged price only when no next-session open was captured). At its ~20-session horizon we mark it a win only if it beat the benchmark, a higher bar than simply going up. Calls are only scored against a fresh price. Hit rate alone is misleading, so we also publish calibration error: a well-calibrated 70% call should come true ~70% of the time. Nothing on this page is a price target or financial advice.