NCAA Predictions

This project scrapes NCAA Division I college basketball game data, uses various prediction algorithms to compare different teams, and produces a filled-in bracket with table of probabilities.

The current algorithms are:

• Naive Seed (i.e. "chalk")
• Bradley-Terry
• Elo
• PageRank
• Path Weight
• Resistance

You can read more about each in the corresponding module's docstrings.

This project currently only works on Windows. You could probably modify some of the commands in simulator.py to get it working on other operating systems.

Prerequisities

• Python 3 and pip
• A LaTeX distribution that has xelatex. I recommend MiKTeX.

Usage

1. Install Python dependencies with pip install -r requirements.txt
2. Modify the tournament variable in simulator.py to include whichever arrangement of teams you'd like. You can find the list of teams in teams/teams.txt.
3. Modify the final lines of simulator.py to include whichever comparators you'd like.
4. Run with python simulator.py. After running, the result(s) will be .pdf and .tex file(s) in predictions/. The PDF(s) should open for viewing automatically.