March Madness Predictions

2021 Predictions

Pre Tournament Analysis

Here are the steps I used to build the model:
(1) Used input data from Kaggle competition and created a feature set
(2) Combined Kaggle feature set with scraped Kenpom Data
(3) Trained a Gradient Boosting Algorithm on past tournament results
(4) Used trained model to predict the win probability of each potential game in the 2021 tournament
(5) Plotted results in bracket

GitHub Link:

The results of my model were mostly chalk (picking top seeds to win), however this makes sense because many of the same input factors used in my model are used in the seeding process. But my model did predict Michigan to come on top when they are the fourth ranked number one seed. However, please don’t use this as gambling advice because this model is flawed in that it does not take into account the injury of Isaiah Livers.