The Hoosiers stormed back from a 22-point deficit Nov. 2 at Memorial Stadium last season to reach within three points of the Gophers.
Threatening to score again, IU faced 2nd and goal from the Minnesota 9-yard line with 25 seconds remaining.
But on a backward swing pass from then-sophomore Nate Sudfeld to then-sophomore Tevin Coleman, the Hoosiers lost possession on a fumble that Coleman thought was an incomplete pass and didn’t attempt to recover.
Minnesota went on to run out the clock and hang onto the three-poimt victory.
Some called it an unfortunate break. A freak play.
Bad luck that proved just how close IU football was to returning to a bowl for the first time since 2007.
“Just an unfortunate play and kind of really bad timing,” Sudfeld said after the game.
But when predicting individual stats and team records, intangible concepts like luck and “bad timing” don’t exist.
In fact, it gets completely tossed out the window. Luck gets replaced by series of formulas and equations.
After IU was seemingly one “unfortunate play” away from an elusive bowl appearance last season, the Indiana Daily Student will once again attempt to predict IU football’s 2014 record using the Pythagorean Expectation Model, or PEM.
The model was first developed by statistical pioneer Bill James in the early 1980s when analyzing Major League Baseball.
PEM has since been made popular by the likes of Oakland A’s general manager Billy Beane, who has used James’ research as a cornerstone of his analytical managing style, which was made famous by the book and movie “Moneyball.”
PEM tries to take luck out of the equation when predicting the amount of wins a team will have.
The model evaluates how good a team is relative to points scored, and points allowed.
The IDS used PEM this time last year to accurately predict that IU would win at least five games last season. The Hoosiers finished 5-7.
Last season, IU scored 461 points and gave up 466. Based on the formula, the Hoosiers should have a winning percentage of .600, rounded to three decimal points.
That equates to last year’s team winning 7.2 games.
More than seven wins for last year’s IU football team seems fairly high. One of the downfalls of the PEM formula is that it fails to account for blowouts, which is where the problem here lies.
Blowouts, especially extreme blowouts such as IU’s 51-3 loss against Wisconsin, can artificially inflate or deflate a team’s PEM.
The not-so-simple solution is to compute the PEM winning percentage on a game-by-game basis and average the results, giving a more accurate reflection of how the team fared according to the PEM.
When computing IU’s PEM with this method, the Hoosiers’ win percentage last season is a projected .487, which equates to 5.848 wins.
This figure is eerily similar to about five wins and a nail-biting loss prompted by a fumble on the 9-yard line against Minnesota.
The PEM is not always correct in its predictions like other intensive statistical formulas, as it is a reflective model and not a predictive one.
However, if the model holds true for the fifth consecutive year, IU should see another slight improvement in its record this season.
And based on IU’s predicted 5.848 wins last season, even the slightest improvement could have bowl ?implications.
