Market consensus

Track record · accruing

When ≥1 tracked sharps back the same outcome, here's how it has resolved:

Hit rate

40%

Implied

42%

Edge

-2.2pp

Sample

981

ROI, equal-stake

-32%

ROI, $-weighted

+3%

But the 95% interval is -5.2 to +0.9pp — it still spans zero, so this is within noise. Not yet proven.

Ground-truth, captured at resolution — hit-rate alone misleads (smart money loads cheap longshots), so we measure edge over the implied price. We publish the interval, not just the point: the sample is young and growing, and we won't call it edge until the numbers earn it.