Probs
Don't take our word for it — we grade ourselves, in public, against real market prices. This page can't be gamed: every projection is written down before the outcome is known.
Every night we save each market's projected price range — the crowd's call, sealed with a date.
When that date arrives, the real closing price is attached. No edits, no do-overs.
The scores below update automatically as projections come due — good weeks and bad.
The key test: our "80% likely range" should contain the real price about 80% of the time — no more (too timid), no less (overconfident). Same idea for the 50% range.
| Horizon | Graded | In 80% range (target 80%) | In 50% range (target 50%) |
|---|---|---|---|
| 1 week | 3,965 | 88.1% | 69.1% |
| 2 weeks | 214 | 85% | 60.7% |
4,179 projections graded so far. Small samples swing — treat early numbers as directional until each row has hundreds of grades.
Every member's forecast is scored against the real outcome — accuracy, calibration and a beat-the-crowd record, all public on their profile. The rankings are built from those scores, and consistently sharp forecasters carry more weight in the numbers you see.