BizarroMath™ Sports Analysis System | FAQ | The Numbers | Last Updated: 22:02 on Saturday, August 20, 2022
BizarroMath™ is a predictive statistical model for NCAA football games. It is written in the Perl programming language and uses data from CFBD.
The program looks at every game played between Division 1 teams during the season to calculate and opponent-adjusted measure of three statistics on both offense and defense: points per game, yards per game, and yards per play. It then projects how each team's offense can be expected to perform against its remaining schedule, and vice versa, and projects how many points each will score. The spread is then mapped to a win percentage based on a linear regression of the system's past performanc eover the prior two years of games.
Before the season is played, we have no data. Even during the early weeks, we have very limited data, and its predictive value is questionable because most teams haven't played much of their conference schedule. Statistically, last year's data is the strongest predictor of this year's data.
I'd really like to but the point of this system is to eliminate subjectivity, and assessing roster talent is really subjective. Roster composition is implicitly included by basing early season data based on last year's roster, which will consist of most (or at least many) of the same players. As we get into the current season, we weight last year's data less, and this year's data more, which also implicitly reflects this year's roster. That said, I am looking at incorporation some measures, such as returning starts, or differential in assessed roster talent. Historically, I don't think roster changes have made massive differences, but rosters turn over a lot faster now.
It's pure data based on on-field performance. It can (and will) make incorrect predictions, but it can't be biased.
I haven't followed Bill's work since he went to ESPN, but yes, it's similar.
Nope. Bill is a sports journalist. I'm a nerd who likes college football and programming.
It's fun. I like fun.
Enjoy the season!