Projecting Wins – 2018 Edition

Bill Connelly of Football Study Hall recently released his initial S&P+ ranks for 2018.  As he puts it, “S&P+ projections are a simple mix of three factors: recent history, returning production, and recruiting.”  For this rating, Indiana ranks 58th.  On offensive returning production, Indiana ranks 83rd with a 60% offensive return.  This percentage is not the simple returning starters metric.  It essentially looks at the amount of returning statistics. Defensively Indiana is 126th at 51%.

Before diving into the numbers, it is important to remember a few things.  First of all, it’s March.  We don’t know who the starting quarterback will be for this team.  We don’t even know if he is currently on the roster.  Secondly, these are projections.  In March.  Don’t cancel your season ticket order over them.

Using the S&P+ ratings, I was able to translate that into the probability to win each game, adjusting for home or road games.  This is my best calculation,[ref]Connelly will have his actual win probabilities in his team profile previews[/ref] but it should be very close to his.

The probabilities have Indiana with an expected win total of 5.5.  The Hoosiers are favored in 6 games and projected to lose the other 6.  The chart to the left is sorted in order of most likely to win to least likely.

When projecting wins prior to the season[ref]This is what we’ll be referring to, projections prior to the season.  Obviously as the season progresses we have new information to make more accurate projections[/ref], a couple problems often befall many Indiana fans.  These problems often result in a best-case scenario projection that is nonetheless viewed as a midpoint estimate.  It is common for fans to divide the season into three categories: winnable games, games where you hope IU can pull off an upset, and toss-up games.  For simplicity, I counted winnable games as games with a greater than 67% change of victory, upset games as games with less than a 33% chance of victory, and toss-up games as ones between 33 and 67%.

The first problem for fans is generally marking all the winnable games as wins.  The statement generally goes something like “Indiana should win all the 3 non-conference games and this 1 Big Ten game”.  And it is often true that Indiana should win all those games.  But what is the likelihood that Indiana will win all of those games?  It isn’t 100%.

Taken individually, Indiana should win each of them.  But in total, to win all of winnable games, is entirely different.  Last year, for instance, prior to the season Purdue was generally listed as a win.  Connelly had IU with a 73% win probability over Purdue.  It was fair to assume that IU should win that game.  However, simply counting these games as wins without taking into the possibility that Indiana stumbles or one of the winnable game opponents improves, generally over-inflates Indiana’s expected win total.

The other issue comes in projecting upsets.  This one generally has more reason and typically goes something like “Indiana should win one of these 4 games”.  Last season maybe it was IU should beat one of Ohio State, Penn State, and Michigan.  Making these assumptions isn’t necessarily wrong as long as expectations are tempered.  It can also be dependent on the number of games classified as an upset game.  Some seasons there are 2 such games.  In other years there are 4.  Winning 1 of 4 is obviously a much easier task than winning 1 of 2.

Everything in between a winnable game and an upset game is generally considered a toss-up game.  While there is generally a favorite in these matchups, it wouldn’t shock you if either team won.  Last year you were probably looking at three such games.  Virginia and Maryland were both games you would have called IU the favorite, and Michigan State was considered a game where the Spartans had the advantage.  It’s quite common, and probably reasonable, to conclude that Indiana should win about 50% of these toss-up games.

Unfortunately Indiana has been unable to run the table on the winnable games and pull off the necessary upsets and split or win more than half of the toss-up games all in the same season to have that breakout campaign.  Last year multiple issues arose.  Purdue emerged as bowl-caliber team and Indiana was unable to win all 5 of the pre-season winnable games. Indiana failed to pull off an upset despite coming so close in the Michigan overtime game. Finally, the Hoosiers won just 1 of 3 toss-up games.  In each instance, Indiana failed to meet or exceed the expectation.

In 2016, Indiana won 3 of the obvious winnable games, neither of the 2 upset games, but only 3 of 7 toss up games.  The result was a 6 win season.

The 2015 team collapsed against Rutgers to ruin what would have been a clean sweep of winnable games.  Indiana was able to win on the road in a toss-up game at Maryland to help ensure a bowl game, but the inability to beat Rutgers or pull off the giant Ohio State upset prevented that elusive 7 win season.

These instances illustrate how hard it is to win all your winnable games, pull off the upset, and split the toss-up games all in the same season.

Getting back to this year, using our definitions the schedule sets up with 2 winnable games, 4 upset games, and 6 toss-up games.

Starting off, the non-conference slate has 2 winnable games[ref]FIU and Ball State[/ref] and 1 toss-up[ref]Virginia[/ref].  Indiana should win all 3 non-conference games.  They are favored in all of them.  But the probability of winning all 3 of them is around just 40%.  Think about that.  It is more likely than not that Indiana is unable to sweep their non-conference schedule.  That may seem really low, but Indiana is headed to Miami to take on a Florida International team that won 8 games a year ago.  A road game with, at best, a relatively inexperienced QB isn’t a sure thing.  They follow that game up against another 2017 bowl game participant in Virginia.  While this isn’t necessarily a difficult non-conference slate, it isn’t a slam dunk either.

Singling out just the 2 winnable games, Indiana has a 64% chance of winning both.  It is more likely than not that they win both, but it is important to take into account the possibility that IU stumbles.  Indiana projects at 1.6 wins in these contests.

On the flip side, I’m sure someone is thinking that Indiana should steal 1 game against the Big 4 of Michigan State, Ohio State, Penn State, and Michigan.  And actually there is about a 50% chance that Indiana will win at least 1 of those games.  A home game with the Spartans is most likely at 22%.  There’s also a 50% chance of losing all 4.  Indiana projects for 0.6 total wins in these games.

In the middle are the toss-up games.  This year there are more toss-ups and less winnable games which could pose a challenge to Indiana.  One positive is Indiana is the favorite, however small, in 4 of the 6 games.  There’s roughly a 74% chance at 3+ wins and about a 43% chance at 4+ wins in the toss-up games.  Indiana projects for 3.3 wins in these 6 games.

Adding up the wins, Indiana comes out with a projected 5.5 wins.  This gives the Hoosiers essentially a 50% chance of winning 6+ games.  There is as good of opportunity as not that the Hoosiers go bowling.  Unfortunately it leaves a slim margin of error.

Winning against both Florida International and Ball State would help.  That increases IU’s odds of 6+ wins to 60%.  If IU could also steal an upset game, let’s use MSU since that game is most likely, then there is an 80% chance of 6+ wins and a 54% chance at 7+.

Having said that, it’s been a while since IU was able to pull off an upset win.  That doesn’t mean they can’t, but it’s important to at least evaluate that possibility.  If IU hypothetically won both their winnable games but lost their 4 upset games, the season would come down to needing 4 of their 6 toss-up games.  That puts the Hoosiers back to about a 43% chance of a bowl game.

What is the most likely scenario?  Indiana probably should win both winnable games.  The upset games are a little trickier.  An optimist would hope that Indiana could win 1 of them, but it is just as safe to assume Indiana strikes out here. That leaves the 6 toss-up games.  If Indiana could hold serve on the 4 games in which they are favored, they would reach the requisite six win plateau.  These games will likely define the season.  Winning 4 of the 6 won’t be easy.

If you are looking for a breakthrough season, the numbers have IU with about a 25% chance at 7+ wins, about a 9% chance of 8+ wins, and about a 2% chance at 9+ wins.  Seven wins would be an incredible accomplishment for this team given the schedule, the quarterback uncertainty, and the defensive rebuild. Let’s not concern ourselves with anything beyond that.

Getting off to a hot start is also important.  If Indiana can start 3-0 with wins over FIU, Virginia, and Ball State, they jump to a 72% probability of 6+ wins.

All this said, it fun to look at the schedule and project wins and losses.  It is somewhat less fun—but maybe better for the fanbase’s collective mental health—to manage expectations.  However, teams exceed expectations all the time, both from a pre-season perspective[ref]For example, Purdue last year[/ref] and winning an unusually high amount of toss-up or upset games.  Come this fall, we’ll find out what the Hoosiers do.