Of the 9 cabinet selection picks that I made (see yesterday's post), none of the candidates that I said was most likely to win won. On average, I gave the favourite 23% odds, and there was a 10% chance that none of the favourites would have won according to my odds. This suggests that I slightly overestimated the chance that the most likely-to-win candidate would win.
Of the 9 picks, 6 of them (or 66.6%) were not on my list of candidates. On average, I gave "Someone Else" a 48% chance of being minister. According to my odds, "Someone else" would be expected to win by chance 6 or more times about 21% of the time. Again, I may have been slightly overconfident that my top candidates would win. It is evidently even harder to predict who will will be in cabinet than I thought.
My Brier score for the cabinet picks was about 0.68, while I would have expected a Brier score of 0.66. The agreement between these numbers is a sign that my forecasts were reasonably well-calibrated overall. This comparison is less useful with a "Someone else" category, because "Someone else" winning actually decreases my Brier score (since "Someone else" was always most likely to win), but is hardly the sign of a great prediction.
No comments:
Post a Comment