5-min pooled window returns around the 14:00 ET statement (UTC). Random-time placebo x2000; Bonferroni bar p < 0.013 across 4 windows. ~30 events (2022-12+).
| window | n | mean bps | t | IS bps | OOS bps | placebo p | survives |
|---|
| pre24h | 30 | +18.9 | +0.48 | +31.8 | +6.0 | 0.3980 | no |
| pre2h | 30 | +20.3 | +2.12 | +38.1 | +2.5 | 0.0585 | no |
| reaction | 30 | -19.0 | -1.56 | -23.0 | -15.0 | 0.9985 | no |
| post | 30 | -52.9 | -1.81 | -94.1 | -11.7 | 1.0000 | no |
| window | n | mean bps | t | IS bps | OOS bps | placebo p | survives |
|---|
| pre24h | 30 | +56.9 | +0.89 | +38.7 | +75.1 | 0.1490 | no |
| pre2h | 30 | +30.7 | +1.74 | +19.1 | +42.3 | 0.0350 | no |
| reaction | 30 | -14.7 | -0.98 | -3.0 | -26.3 | 0.9635 | no |
| post | 30 | -49.7 | -1.87 | -77.2 | -22.2 | 0.9985 | no |
- Directional adoption needs mean > 0, p below the Bonferroni bar and IS/OOS same sign. The reaction window is the sharpest test.
- Underpowered by construction (~30 events): NONE is a lack of evidence, not proof of absence.