Thanks for hosting this contest. The overconfidence thing in particular is a fascinating data point. When I was done with my function that output final probabilities, I deliberately made it way more agnostic, thinking that at least now my estimates are modest—but it turns out that I should have gone quite a bit farther with that adjustment.
I’m also intrigued by the variety of approaches for analyzing strings. I solely looked at frequency of monotone groups (i.e., how many single 0′s, how many 00′s, how many 000′s, how many 111′s, etc.), and as a result, I have widely different estimates (compared to the winning submissions) on some of the strings where other methods were successful.
Thanks for hosting this contest. The overconfidence thing in particular is a fascinating data point. When I was done with my function that output final probabilities, I deliberately made it way more agnostic, thinking that at least now my estimates are modest—but it turns out that I should have gone quite a bit farther with that adjustment.
I’m also intrigued by the variety of approaches for analyzing strings. I solely looked at frequency of monotone groups (i.e., how many single 0′s, how many 00′s, how many 000′s, how many 111′s, etc.), and as a result, I have widely different estimates (compared to the winning submissions) on some of the strings where other methods were successful.