Home About Stats

The Perfect Bracket Simulator: How The Stats Are Processed

Summary

All statistics for this website are sourced from ESPN's college basketball data. After retrieving the stats, Python was used to clean and format the data for this specific use case. To ensure balanced metric comparisons, the stats were scaled so that no single metric, such as points per game, outweighed others like assists per game. This scaling process adjusted each stat so that the highest value for a team was set to 1, the lowest to 0, and all other values were linearly scaled in between. For instance, if Kansas led all tournament teams in points per game, their scaled value for that metric would be 1.

For stats such as points per game (PPG), where higher values are desirable, assigning a higher weight to the stat means teams with higher PPG averages are predicted to perform better. Conversely, for stats like turnovers (TOV), where lower values are preferable, the data was inverted so that the team with the fewest turnovers had their TOV value scaled to 1. As a result, if the TOV weight is increased, teams that took care of the ball will be predicted to perform better.

Once you've created the combination of weights you believe will create the perfect bracket and hit "Run," each team's score will be displayed next to their name in the first round. This allows you to see how closely matched two teams are and identify areas where a team excels or falls short. Scores are only shown for the first round but remain consistent throughout all rounds. To fine-tune a stat's weight and see how it affects team scores, you can adjust the slider using the left/right arrow keys for increments of 1 or type your desired weight directly into the input field.

Formulas

“Perfection is not attainable, but if we chase perfection we can catch excellence” - Vince Lombardi

Questions? Contact: perfectbracketsim@gmail.com