Anyone can publish a hit rate. In a strong tape everything runs, so "most of our calls went green" can be true and tell you nothing. The number that means something is how our calls did against comparable stocks on the same day, through the same trigger, from the same minute.
An alerted name was 4.4 times more likely to run 20%+ than a comparable name we passed on.95% confidence interval 3.0x to 5.9x · bootstrap, 4,000 resamples
The control group is every name that met the same screen and triggered the same setup on the same day that we did not call. Same sessions, same universe, same measurement. The only difference is our filters and our ranking.
| Outcome | Our calls | Control | Lift | 95% CI |
|---|---|---|---|---|
| Ran 10%+ | 52.7% | 31.2% | 1.7x | 1.4 to 2.0 |
| Ran 20%+ | 29.0% | 6.7% | 4.4x | 3.0 to 5.9 |
| Ran 50%+ | 7.5% | 0.7% | 11.0x | 3.6 to 24.0 |