> In my dashboard, any market above 50% that resolves to “Yes” is counted as correct and any market below 50% that results in “No” is counted as correct.
...Why not use a Brier score? This is kind of a weirdly innumerate version of accuracy-tracking for a prediction market.
I mean simply being able to “Polymarket predicted X but question resolved to Y.” (instead of expressing that in terms of Brier score) feels easier most of the time.
> In my dashboard, any market above 50% that resolves to “Yes” is counted as correct and any market below 50% that results in “No” is counted as correct.
...Why not use a Brier score? This is kind of a weirdly innumerate version of accuracy-tracking for a prediction market.
I mean simply being able to “Polymarket predicted X but question resolved to Y.” (instead of expressing that in terms of Brier score) feels easier most of the time.
I suppose this is an effect of the difficulties with communicating Brier score to a wider public.