At first I thought the "equal probability assumption" wasn't too bad an approximation, but it has an average Brier score of 0.217. Much worse than I thought.