March Madness
I recently met someone through LinkedIn that I had the pleasure of seeing in real life. He met me on the OSU campus, and we walked and talked for about an hour. His primary interest in data science is related to sports and we had a fascinating discussion about the use of AI models in sports betting, particularly the NCAA basketball tournaments. Every year, ESPN has a men's and women's tournament challenge. (There are $135k in prizes to be won, but ONLY for the men's tournament. Go figure.) What I learned from my new friend is that AI models were able to predict the winners in this year's tournament with an accuracy of 93%. Getting it 100% correct will win some cash, so there's definitely an incentive to get that last 7%! His comment that really struck me is that an upset is not *really* an upset. It's usually bad seeding. That is, when a #12 team beats a #5 team it's usually because the #12 team should have been seeded higher and the #5 team...