By TSB Bracketology, November 27, 2023
Big Ten basketball is upon us. Our Boilers start off the conference season with a game against Northwestern on December 1st and follow it up with the home opener the Monday after against Iowa on December 4th. Other notable matchups this week include Michigan State-Wisconsin, and Indiana-Michigan, which should be compelling matchups this early in the season.
Coming off of a strong week of MTEs with Feast Week, we’re going to take a look at how the Big Ten season may shape up by using KenPom Efficiency Data to form our forecasts. Before this interlude of Big Ten matchups, we’re also going to compare projections from data from the beginning of the season and projections from the most updated KenPom numbers. In doing so, we’re going to see how projections have changed already in this short season and have a better understanding of what could be to come for the upcoming conference season.
KenPom’s ratings can be used to predict the outcome of future games, but, for forecasting purposes, we’ll be using the data in order to calculate win probabilities. With the win probabilities, we can effectively assign wins to teams based on randomly generated numbers ranging from 0 to 1, with each number corresponding to the result of the simulated game. For example, if a home team in a matchup had a win probability of 58 percent and the randomly generated number was less than or equal to 58, then the home team would be assigned a win.
If we repeat this process for all 140 Big Ten conference games and we assume that all of the games are independent events, then we can complete a simulation for the upcoming conference slate. Running through 10,000 simulations of the Big Ten season, we can get decently accurate forecasts for the probability that a team reaches a certain threshold of wins across the season. We can also get fairly precise numbers for a given team’s projected wins for the season.
The decision to include every single conference game in a simulation was made in order to ensure accuracy for the probabilities that the teams have to win the Big Ten regular season title. The individual win probabilities and the expected win totals for each team could have been collected by simulating a team’s conference schedule individually (e.g., only Purdue’s schedule, only Illinois’ schedule, etc.), and the subsequent numbers would be the same, if not very similar to, the resulting numbers.
However, in order to obtain title forecast numbers from simulations for individual teams, teams would be compared by randomly generating the wins based on the win probabilities for the team, and the team (or teams) with the most conference wins in a simulation would be given the title. The issue with this method is the sum of all of the simulated win totals could be above or below the 140 conference games since results are not decided game by game, which means that both teams for a given game could be unknowingly given a win or a loss. By including every single conference in a simulation, we ensure more accurate regular season title probabilities.
With the bulk of the methodology explained, let’s dig into the initial numbers. The projected wins for each team from the preseason data were largely consistent with what was expected from the media. At the beginning of the season, Purdue was the clear favorite for the Big Ten regular season crown at about 15 wins, with Michigan State expected to be elite competition for the Boilermakers with an average of just about 13 wins. Wisconsin, Maryland, and Illinois were expected to be competitive at about 12 conference wins each, with each team having about a 10 percent chance for at least a share of the title and a 5 percent chance at sole position of the title. Penn State at about 6 wins and Minnesota at about 5 wins were expected to be the bottom dwellers of the conference, while the rest of the teams filled out the rest of the standings, averaging anywhere between 8 to 11 conference wins.
However, as the season has progressed, these projections have fluctuated, just as KenPom’s efficiency numbers have fluctuated as well. Purdue had the largest jump in Efficiency ratings after a Maui Invitational tournament win that included wins against Gonzaga, Tennessee, and Marquette. In fact, at one point, Purdue had 5 of the top 8 KenPom teams on their schedule, showing just how difficult their non-conference schedule has been. Nebraska has jumped 12 positions in the KenPom rankings after a hot start against lesser opponents, while Ohio State’s upset win over Alabama in the Emerald Coast Classic catapulted them up in the rankings. Even with losses to Creighton and Oklahoma, Iowa’s dominant wins over North Dakota and Alabama State have moved them up as well.
For the most part, most other Big Ten teams have had a tough non-conference schedule. Maryland was the largest falling team, having been upset by Davidson and UAB, and dominated by Villanova. Even though Indiana has only lost one game to UConn, they have largely struggled in all of their other games, as Florida Gulf Coast, Army, Wright State, and Harvard have proved difficult opponents for the Hoosiers. Northwestern hasn’t lived up to preseason expectations, and even Michigan State has dropped a little bit, after an opening night upset at home against James Madison, and having lost two neutral games to Duke and Arizona.
After only three weeks of basketball so far, Purdue has run away with the forecasted Big Ten title. With an 89 percent chance of at least a shared title, the Boilermakers is the team to beat not only in the Big Ten, but across the nation. The expected ~17 wins for Purdue is miles above any other conference foe, and the team has about a 2 percent chance to go undefeated in the Big Ten according to the current data. Of course it’s still a long shot for the Boilers to finish unbeaten in the Big Ten, but even still, Purdue will be a huge mountain to climb for any opponent (literally and figuratively with Zach Edey on the court), especially in Mackey Arena.
Aside from Purdue, the rest of the teams have more or less been broken up into three separate tiers. There’s the better half that is composed of Ohio State, Michigan State, Wisconsin, Illinois, Iowa, and Nebraska. Then, there’s the worse half with Maryland, Michigan, Northwestern, Rutgers, Indiana, and Penn State. And then there’s still Minnesota towards the bottom.
Ohio State and Michigan State actually ended up with the same amount of conference wins to the hundredths place with the efficiency data at this point in the season, but the teams have gotten to this number in different ways. At 5-1, Ohio State probably has the best resume in the conference behind Purdue. The Buckeyes have a very strong neutral site win over Alabama and a close home loss to a good Texas A&M team. On the other hand, Michigan State currently sits at 3-3 and has had tough opponents in Duke, Arizona, and apparently James Madison. The schedule for the Spartans doesn’t get much easier, with a gritty Wisconsin team, a hot Nebraska team, and Baylor in coming weeks. Even so, Coach Izzo always seems to find a way every season, and if Tyson Walker can get some consistent help offensively, Michigan State will be a scary team come March.
The biggest thing to gather from this forecast is how much parity there is in this league. Aside from Purdue at the top and Minnesota at the bottom, teams have shuffled in and out of projected positions as the short season has gone on. At the beginning of the season, Maryland was near the top at about 12 expected wins, but now sits in the very middle at a projected conference record of 9-11. Wisconsin has stayed about stable from these two forecasts, but they have only recently moved back up to the upper echelon of the projected standings after their wins over Virginia and SMU in Fort Myers. Nebraska moved up nearly 2 whole conference wins since the start of the season, but has only moved from 11th to 7th. There really is no predicting just which way the Big Ten season will take us.
Of course, both of these forecasts are preliminary and teams go on hot and cold streaks throughout the season. Even with Big Ten conference games this weekend, many teams still have tough non-conference opponents coming up too. At the end of the day, these projections should only be used to gather a general understanding of the expectations for each team at this stage. All in all, I am very excited for the start of Big Ten basketball and very interested to see how accurate these forecasts end up.
