Do the math: IU football's record will improve this year

IU finished the year 4-8 last season, but should have finished 5-7 according to the Pythagorean Expectation Model.

The model, first developed by statistical pioneer Bill James for baseball and later adapted for football, evaluates how good a team is relative to points the team scored and points scored against the team.

Looking at teams in the Big Ten last year, the PEM number suggests teams with a higher expected win total than actual win total will have more wins the following year.

Essentially, the PEM is a better tool for predicting how a team will do the next season than its actual win-loss record.

In four contests last year, IU lost by a combined total of 10 points. The Hoosiers were in all of those games late into the fourth quarter but couldn’t manage to win.

But even though they were within a touchdown of winning each of those games, the win-loss record still stands at 0-4. The win-loss record fails to show how close the Hoosiers were to a victory in each instance.

This is where the PEM can more clearly illustrate what IU’s record should have been.

Last year, IU scored 369 points and gave up 423. Using the equation which uses elements similar to the Pythagorean Theorem, the Hoosiers should have won 5.03 games relative to their points scored and points against.

They won four.

The difference of 1.03 expected wins versus actual wins was tied for the third-highest figure in the conference last season.

Consider Ohio State for example. In 2011, the Buckeyes had a 6-7 record. However, the PEM showed the team should have won 7.7 wins. This is a full 1.7 difference between the PEM and actual win-loss record.

The next year, Ohio State went 12-0.

Not all analysis is as drastic as this, but the PEM is a useful tool for predicting a team’s record.

Using the PEM retroactively in IU’s history, it shows its usefulness once again in predicting improvement or regression. The 2009 IU team went 4-8 but had a positive PEM number of 0.42, so they should have won 4.42 games.
In 2010, they improved to 5-7.

That year, IU had a PEM number of -0.56 (so they should have won 4.44 games), and the next year their total number of wins decreased from five to one.

In 2011, the Hoosiers finished 1-11 but had a positive PEM number of 2.54, meaning they should have won 3.54 games instead of one.

So by considering the PEM, the 2012 Hoosiers should have seen an increase in wins, which they did. The 2012 squad finished 4-8.

Looking at the past three seasons for IU, they have improved each time they had a positive PEM the year before and regressed in the years they have had a negative PEM the year before.

So what is IU expected to do this year?

Last season they had a PEM of 1.04, meaning they should have won roughly five games, but they only went 4-8.

The PEM is not always correct in its predictions and is not the only factor one should look at when deciphering whether or not a team will improve.

However, if this model holds true for the fourth consecutive year, IU will win five or more games this year.


Follow reporter Evan Hoopfer on Twitter @EvanHoopfer.