pauldevos / kaggle_ncaa_womens_2019

Solution for 3rd place finish in Kaggle's NCAA March Madness 2019 Tournament

Kaggle_NCAA_Womens_2019

Description

Solution for 3rd place finish in Kaggle's NCAA March Madness 2019 Tournament

Data Files

All input data files were sourced from https://www.kaggle.com/c/womens-machine-learning-competition-2019/data

Model Overview

A linear regression model was put together using pymc3. All data pre-processing and post-processing steps were handled using both numpy and pandas.

• FG attempts as team-level Gammas vs Normals (or Poissons for counts) to allow using loc/scale parameters instead of mu/sd
• FG percentages as team-level Betas vs Bounded (0,1) Normals to allow using loc/scale parameters instead of mu/sd
• League-level Half-Cauchy error term to soak up some variance
• Assume team's scoring propensity is normally-distributed and treat as observed variable
• Assume points spread between two teams is normally-distributed and treat as observed variable

Evaluation of odds of one team winning over another was done via simulation of team-level posterior scoring propensities (calculated from above).

• Simulation for each match up produces an output distribution representing the spread of each simulated match
• The win probability is determined by the number of simulated matches won over the total number of simulated matches
• 5000 simulations were run for each potential match up
• Matches exempt from this process for overridding odds are high seed/low seed tournament matches