A small add-on to the BCB Over/Under Contest stats

If he had picked all the unders he’d be the furthest out instead, beating your score by about .03. :)

Compared to the maximum and minimum RJ’s actually fairly middle of the pack (1.98-2.70-3.64), which is technically where you’d expect an all-the-same ballot to be when the majority overs and unders are split evenly as they are – 15 questions ended up majority over and the other 15 majority under. The reason his score isn’t quite in the center is that the 15 majority overs happen to be somewhat more lopsided than the 15 unders. (70% over on average vs. just 62% under on average.)

But yes, it did take quite some contrarianness on the most lopsided questions to get a worse consensus score than RJ did.

I’m tempted to find out where the average coin-flipper would place, relative to the same baseline…

I didn’t mean to quantify how optimistic or pessimistic somebody is compared to the average opinion, just how agreeable or contrarian they are. For that, all distances were (and should be) considered equal, i.e. it doesn’t matter whether you’re optimistic on a question that got 80% pessimistic responses, or pessimistic on a question that got 80% optimistic responses.

The background to that is that being on the agreeable side tends to give you a leg up on scoring comparatively well in such contests (i.e. very likely to finish at least above the median score) and thus a somewhat better shot at actually winning, at least under the assumption that most participants know what they’re doing and the picks aren’t extremely biased by wide-eyed fan optimism.

Incidentally, it’s somewhat difficult to calculate a meaningful “average level of optimism” from those over/unders, as that’s dependent on where KL set the various mid-points (ideally all questions should have yielded a split opinion, which of course they didn’t, even though they were placed very well overall), the question of what’s considered optimistic (say, on the number of Mat Gamel’s games), and the question of how those 30 values should be weighted (e.g. the 100+ K’s are arguably much less important than the team wins).