Well, I am still relatively new to modeling on Python – but I figure why not practice. I have all this great basketball data.

I remember from past tournaments that Bill Kahn would tell me about the Bradley-Terry models that he created. And he always did very well in the pool. So, I definitely don’t have the experience that Bill has with building these models, it was at least an opportunity to learn a little about Python.

So, I figure I have an extra day thanks to the unusual this schedule – lets see what I get. And it is really possible that I made some mistakes. But I also don’t think I can use the model. First of all, it predicted Winthrop would beat both Villanova and Purdue. I can live with the first upset but I can’t pick the second one against my Boilers. Second, it created a prediction that I suspect won’t happen.

That is because its prediction for the Final 4 are all 4 of the #1 seeds. March Madness is anything but predictable – so while I have no doubt that Gonzaga, Baylor, Illinois and Michigan are the top 4 teams in the country based on the regular season, I don’t think all 4 of them will survive the regionals. One of the top seeds always goes down. It is inevitable.

If you are wondering, it finishes with Gonzaga having a 65.7% chance of beating Baylor.