I’m sort of falling in love with football’s Pythagorean Wins Theorem.
Before we get into that data, there are two questions worth answering quickly:
1) How accurate are Pythagorean Wins in college football? – This concept works very well in the NFL but there’s some debate, or at least there was a few years back based on Google searches, over how well it applies to the college game. Based on my numbers, the football formula predicted win totals within one game of the actual win totals nearly two-thirds (63.07%) of the time between 2007 and 2012. The formula projected 35.96% of the teams within a half game of the actual win totals over the same span. It’s reasonably accurate.
2) How predictive is it? – Pythagorean Wins does a pretty good job of identifying teams that drastically over or underperformed. Over the last five seasons, teams that were +/- 1.5 or more wins typically played to their actual form the next season, particularly the underperforming teams. Between 2007 and 2012, 64.8% of teams that were 1.5 games or more below their expected win total improved their win total the next season, while just 14.8% got worse. On the other end of the spectrum, teams that were 1.5 games or more above their expected win total saw their win total decline 62.2% of the time the following season, while 31.1% improved.
Read the entire thing over at HailVarsity.com.