## Saturday, October 24, 2009

### WASEAN Explained

WASEAN is a college football team rating system. The acronym stands for Winning% Algorithm for Scoring ExpectAtioN. The name not only describes the system, but honors Toledo's star running back during the 1990s, Wasean Tait.

The system uses the Pythagorean theory (developed by baseball author Bill James), which says that you can predict team winning percentage knowing only how many points they scored and allowed. It's an incredibly useful tool. The basic formula (for baseball) is that R^2 / ((R^2)+(RA^2)) = WPct.

Over the years the formula has been refined. Instead squaring the runs and runs allowed, it's been determined that using (R/G + RA/G) ^ .286 as a dynamic exponent based on run environment works even better. For football, the best exponent appears to be (PF/G + PA/G) ^ .26.

Basically I do this on a game by game basis and factor in a home field advantage as well.

However, there is one major issue with this formula. It sees a 30-0 game and a 3-0 game as the same thing.

It sees a 70-30 win as the same as a 7-3 win. So we make an adjustment that fixes this.

The next thing you have to do is work in a strength of schedule adjustment.

Again, I'm borrowing from James, who uses an iterative process to compute his power ratings at billjamesonline.com (which is well worth the \$2.99 every 3 months).

Each team starts with the average of their raw individual game Pythagorean scores. If you post a .932 (a 55-7 win at home) but it's against a team whose average score is only .167, that's not nearly as impressive as posting that win against a good team.

Special thanks to the guys at The Hardball Times for helping work through the mechanics of this. Read their site and buy their books if you love baseball. Paul Kislanko and Greg Tamer were also a big help to me throughout this process.

Anyway, after adjusting for strength of opponent, the score for the 55-7 win becomes .735.

We do that once for everyone. After doing that once, the .167 team's new average becomes .211. So we adjust based on .211 instead of .167 and your 55-7 win now becomes a .787 instead of a .735.

We keep doing this over and over, until the numbers stop moving. I run 100 iterations, but after 7 weeks, the last round where the numbers move enough to change a ranking position is the 51st round.

We then make one final adjustment for unbeaten/untied (remember we don't count OT) and winless/tieless teams. We recognize a limitation of the system, and move the unbeaten teams above the best team they beat, and winless teams below the worst team they lost to.

One note . . . WASEAN does not include overtime. The reasoning is that a 21-20 regulation win should score higher than a 27-20 overtime win. This is somewhat ironic, considering who the system is named for . . .