Skip to content
CALIBRE
Graded calls

CALIBRE's own record

Every verdict CALIBRE has sealed on Walrus before the result was known, with its grade once the market resolves. The grade compares CALIBRE's forecast with the market price at the time of the verdict, on the same result. Nothing here is edited after the fact: a verdict that turns out wrong stays on the list.

Sealed verdicts
143
Graded
108
of 143 sealed
CALIBRE Brier
0.165
Lower is better
Market-price Brier
0.166
At verdict time

What the record says so far

Over 108 graded verdicts CALIBRE's forecast and the market price are not clearly different (gain 0.0012, standard error 0.0033).

Mean gain (price Brier minus CALIBRE Brier) +0.0012 ± 0.0033 (one standard error); the forecast was closer than the price in 31% of graded verdicts.

Brier score is the squared gap between a probability and what happened (1 if the first outcome won, else 0). A forecast that beats the price must be closer to the result on average, not just once. Verdicts from model v1 and v2 are kept as sealed, including their known weaknesses; the model changes are explained on the Method page.

Results by label

Graded verdicts only, grouped by the label the verdict was sealed with. A group with a handful of verdicts says nothing yet.

LabelGradedCALIBRE BrierPrice BrierGainCloser than price
contested380.19020.1977+0.007539%
mixed210.12460.1203-0.004329%
noise390.18160.1818+0.000231%
sharp-led100.09200.0846-0.007410%
all1080.16520.1664+0.001231%

Every sealed verdict

Read from Walrus 2026-11-08 16:05 UTC