Using Statistics to Predict Penn State's Fate and the Outcome of the Big Ten Tournament

By Nate Wilmot on March 10, 2020 at 8:00 am
Penn State guard Myreon Jones attempts a layup against Northwestern in the regular season finale in Evanston, Illinois.
David Banks-USA TODAY Sports

If anyone, with any level of confidence, tells you they know or think they know what will happen when the Big Ten Tournament tips off on Wednesday, they are either lying or delusional. This year's conference was the model for parity - outside of Nebraska and Northwestern that is. Three teams tied for the top spot at 14 - 6. Five others were within three games. So, what you're about to read is a guess. It's a guess based on the best statistical analysis we have but it's still a guess.

Offensive and Defensive Efficiencies

Ken Pomeroy and others use efficiency (points per 100 possessions) as the model to predict basketball outcomes. The graph below shows the offensive efficiency numbers for the entire conference. In conference play, every team was within 10 PPP of the top team (Iowa) except for Nebraska and Northwestern. Considering a normal college game has about 70 possessions, this means that teams are all scoring within about two possessions of each other. Looking at the differences between the blue (out of conference) and red circles also shows us that most teams (WTH Wisconsin!?) had a pretty big drop-off once the conference play hit. 

Offensive efficiencies - in and out of conference

The defensive efficiencies are below. This represents points allowed per 100 possessions. For conference play, again, with the exception of a few outliers – Nebraska, Northwestern, and Iowa – everyone is fairly bunched up and basically giving up 100 points per 100 possessions (give or take a 3-pointer).

Defensive Efficiencies - In and out of conference.

Putting it all together we get the chart below which is a plot of offensive and defensive efficiencies for Big Ten play only (top right means great on offense and defense). This puts into perspective how bad Nebraska and Northwestern are. Neither of them does much well (except in Northwestern's case, finding a way to beat Penn State). Iowa has the league's best offense but plays defense as poorly as the two bottom feeders. Then, there's pretty much everyone else, save Michigan State. There is a group of ten schools that play some variety of good defense and score a fair amount of points. When they play each other, the games are basically coin-flips.

The brightest star in the conference is clearly Michigan State. They combine the league's best defense (<95 points per 100 possessions) with the league's 2nd best offense (about 1.06 PPP).

off and def efficiencies.

So what's going to happen? As I mentioned, to say anyone knows for sure would be folly. However, it doesn't mean we can't try to guess. So here we go... 

Background: The model and calculations are based only on the Big Ten games. I threw away all of the non-conference games because even Northwestern is not as bad as Maryland-Eastern Shore. I used only my own calculated efficiencies and possessions which may differ slightly from KenPom. Then, with some assumption for error, I did 5,000 Monte Carlo simulations per game. Win probabilities may not add to 100% because of the possibility of ties/overtime. You may comment or contact me for further details if interested!

First Round

No upsets here (probably). I see Minnesota and Indiana each winning by about a five-point margin. While the average point difference is pretty low for both games, the favorites each have a >80% win probabillity.

first round prediction

Second Round

This makes me a bit sick actually. Michigan by 0.2, Iowa by 0.4, and Ohio State by 0.6 points, none of which have a 50% win probability. Obviously someone will win but I left the ties in as "overtime". They're all coin flips.

But would you look at that, which second-round team has the best win probability? That's right, it's Dear Old State with a near two-point projected advantage! In the simulations, they win six out of ten against Indiana. Here's to landing on the correct side of 6-4 this time.

second round predictions.


In the quarters we are going to likely see a few more close games. Everything is within 2.4 points with Michigan State having the highest win probability at 68%. No game should be a blowout but by the numbers, Penn State's run could come to an end at the hands of Maryland. The Lions do have a win over the Terps already this year though, so hopefully, they'll be able to beat the odds.

Quarterfinal predictions

Semi-Finals and Finals

semi and finals

Obviously you had to see this coming. I told you above what the projections were based on AND that one team stands out. Sparty takes home the championship in the simulation after beating Wisconsin in the finals.

All in all, there is just one game in the entire simulation with a projected scoring margin of greater than 2.3 points. After round one, every game seems primed to come down to which team is more engaged and more energized, which would be very fitting after the Big Ten season we just saw. Still, the parity also means opportunity. Penn State may have ended the regular season on a cold streak, but anyone can win in this league, and now's as good a time as any to prove it.