Every call, graded in public
Track Record
The calls we got right, the ones we got wrong, and the Brier score that keeps us honest. Refuted calls get the same billing as confirmed ones — that's the whole point.
5
Resolved Calls
4
Confirmed
1
Refuted
0.02
Brier Score
Brier score measures forecast accuracy: 0.00 is perfect, 0.25 is coin-flipping. Computed at render time from final probabilities — never stored, so it can't drift.
How Resolution Works
Every forecast publishes its resolution criteria up front — the exact conditions under which we'll call it confirmed or refuted. When the evidence lands, we grade the call against those criteria, freeze community voting, and record the verdict with a dated changelog entry. Individual expert predictions are graded on our assessed probability at the time, not the author's stated confidence. Full methodology on the About page.