The Crossbook Record #1

10–16 Aug 2026 · Published scorecard

How to read this. Crossbook states its own probability for a market before that market resolves, and freezes the market’s price at the same instant so the two can be compared afterwards. A Brier score measures how far a probability sat from what actually happened: lower is better, 0 is perfect, and 0.25 is what flipping a coin earns. The market’s own price is scored on the identical calls, and that comparison is the only one that means anything, because a forecast that merely agrees with the price inherits the market’s accuracy without adding to it.

Every call below was logged before its market resolved. Wins and losses count the same, nothing is deleted, and calls that turn out not to have been forecasts at all are kept and labelled rather than removed. This is a small sample and is not evidence of an edge either way.

Restated, and the score here is worse than the one that was emailed. This edition went out quoting a Brier of 0.1594 over 12 resolved calls. It now reads 0.1677 over 11, because 8 of its 20 calls were logged after the event they were about had already started. Those are not forecasts, so they count toward nothing. They are still listed below, labelled, and the score shown here is the one the current rules produce rather than the one that was sent. The full account.