The last time I provided a metrics and odds update for Michigan State and the Big Ten, the situation was much different. The Spartans were undefeated and in the top-five. The Big Ten campaign was about to start and many prognosticators were starting to predict that Michigan State might just win a fourth consecutive Big Ten regular season title.
But, Michigan State took a trip to Evanston, Illinois and came back with an “L.” Then, the Spartans’ Christmas Day house guest left MSU fans feeling blue. Now, MSU is 0-2 to start Big Ten play. How bad is it? I have once again crunched the number and I think that I have the answer.
Before looking at the data, however, I will note that I have made a significant improvement to my simulation. I found a few old mid-season Kenpom data files on my hard drive for the 2019 season. This gave me a way to estimate the how much the adjusted efficiency data is likely to change over time. Figure 1 below compares these changes for all Division One teams between Jan. 4th, 2019 and the final pre-tournament data in 2019 relative to the adjusted efficiency margin values in early January.
As the Figure shows, the standard deviation of efficiency margin movement over this timeframe is about 3.5 and the range is about plus-or-minus roughly 10. The distribution also appears to be basically Gaussian/Normal and it does not depend on the quality of any given team. Furthermore, this range and standard deviation appears to decrease linearly down to zero on Selection Sunday.
With these parameters, it is straightforward to inject the proper amount of uncertainty into the simulation. I am making the assumption that the final pre-tournament efficiency margin for every team is the “correct” value for that team and that it just takes the full season of accumulated data to figure that out. In this way I can use a random number generator in each simulation cycle to assign “true” efficiency margin to each team based on the current efficiency margin value and the number of days left until Selection Sunday. I will use this method going forward.
Current Standings and Odds Update
With this in mind, let’s take a look at the current position of MSU and the rest of the Big Ten, starting with my enhanced Big Ten standings. Table 1 below shows the current Kenpom adjusted efficiency margin and rank of all 14 Big Ten teams, along with the current record and winning percentage. In addition, I have included each team’s “plus/minus” rating, which is simply the number of road wins minus the number of home losses. Finally there are two columns for “luck” and expected wins, which are related.
Basically, expected wins are the number of expected wins a team should have up to this point based on the retroactive projected win probabilities for the games that are already finished. The parameter I call “luck” is the different between the current number of wins and the expected number of wins.
Despite the fact that the most Big Ten games that any team has played is four, there are only three total undefeated teams in conference play: Wisconsin, Michigan, and Northwestern. Furthermore, the road team has only won only five of the current 17 games played so far.
As for “luck,” it is still early for these numbers to mean much, but so far MSU seems pretty unlucky. Based on the updated expected win values for the Northwestern and Wisconsin game, more likely than not, MSU should have won one (0.86, exactly) of the two games so far. Now, perhaps this is due to “luck” and perhaps it is “grit,” or perhaps it is just that the Spartans are still overrated. In any event, MSU is still 0-2 and near the bottom of the Big Ten standings.
The updated Big Ten win matrix is shown below in Table 2.
As of now, the Spartans’ expected win total has dropped to 8.77 and the odds of finishing at .500 or better is at only 39 percent. The most likely record is 9-11, which would be the worst finish in Tom Izzo’s tenure, as he has never finished under .500 in league play. Meanwhile, Wisconsin really solidified its position as the Big Ten front-runner. The Badgers’ expected win total in at 14.28 which is a full game greater than the next best team, Illinois.
As for the impact on the odds to win or share a Big Ten title, that is summarized below in Table 3.
MSU’s odds to share the Big Ten title have dropped to just 1.2 percent. This includes the scenarios where the Spartans’ “true” Kenpom efficiency margin is actually much higher than the current value of 17.87.
The reason for this is due in large measure to the current position of Wisconsin. As the table shows, the Badgers now project to have almost a 50 percent chance to win the Big Ten. Furthermore, the table suggests that there is now almost a 60 percent chance that the final record of the Big Ten champ(s) will be 16-4 or better. In order the MSU to reach this record, the Spartans would need to close with a 16-2 record, which seems very, very difficult.
Strength of Schedule Update
I spent a lot of time a month ago analyzing the idea of strength of schedule using the preseason Kenpom data. The general method that I used calculates the number of expected wins for each schedule assuming that it is played by an average power five team. As teams play and the data shows that they are either better or worse than expected, the relative strengths of schedule also change. Below I show the updated strength of schedule data for the Big Ten.
In the preseason, Wisconsin had a notable schedule advantage over the rest of the conference but the updated numbers now show Illinois and Purdue neck-and-neck with the Badgers. The Boilermakers now actually have a slightly easier schedule overall. MSU’s schedule remains in the middle of the pack.
Now that a few a games are in the books, it is also possible to calculate the remaining strength of schedule, which is the same normalized expected win calculation that only sums of victory probabilities for all remaining games.
However, not all teams have the same number of game remaining. To normalize for this imbalance, I divided the expected number of remaining wins for each schedule by the total number of remaining games to arrive at a the win percentage that an average Power Five would expect. The results of that calculation are shown below in Figure 3.
In this case, it is actually Purdue that still has the easiest remaining schedule, but Rutgers and Illinois are not far behind. Michigan State is still in the middle of the pack. It is also notable that the other two undefeated teams, Michigan and especially Northwestern, have the two hardest remaining conference schedules. This helps to explain why Wisconsin has such a commanding lead in the regular season race.
Big Ten Tournament Projection
If the season ended today, that would be weird, because only 17 total Big Ten games have been played and the teams have not played the same number of games. That said, it is possible even now to make projections about the Big Ten tournament based on the simulated results. Table 4 below provides that update.
Based on the combined odds for the entire simulation, the top four seeds of the Big Ten Tournament are currently projected to be Wisconsin, Illinois, Iowa, and Michigan. MSU currently is checking in as the No. 11 seed, which would again be the worst seed that Tom Izzo has ever drawn in the Big Ten post season, by a long shot. The previous worst seed was a No. 7 seed which has happened twice, once in 2007 and once in 2011. Since 2011, MSU has been no worse than the No. 3 seed once, which was in 2017 when the Spartans were the No. 5 seed in the Big Ten Tournament.
That said, Michigan State still projects to have a 10 percent chance to make it into the top four (which would earn the coveted double bye) and a 60 percent chance to at least avoid the opening day. Also note, in the scenario where the current projected favorite wins all the remaining Big Ten games, MSU would finish at only 8-12 in Big Ten play, which would also earn MSU just the No. 11. In both cases, MSU would earn a date to play on Wednesday of the Big Ten Tournament for the first time in history.
Michigan State’s Current Position and Upcoming Schedule
Following MSU’s loss to Wisconsin, I can update the Kenpom scatter plot to show the current position of MSU relative to past MSU teams, previous champions, and the current field of national contenders.
As we can see and we have experienced, the trajectory of the Spartans’ season has continued downward (as noted by the solid green line). In general, the adjusted offensive efficiency for MSU has increased slightly, but the adjusted defensive efficiency has been dropping steadily. Shockingly, however, MSU is still one of only 16 teams whose current combination of offensive and defensive efficiency remains in the blue shaded “championship zone.”
That said, the current position of the 2020-21 Michigan State team is rapidly approaching that of past Izzo teams that did not make it out of the first weekend of the NCAA tournament. Right now, the current profile most closely resembles that of the 2002 and 2017 MSU team.
I have also added an additional feature to this plot, which is the green “error bar” oval around the MSU data point. This oval represents one standard deviation in both the adjusted offensive and defensive efficiency expected this many days from Selection Sunday, based on the analysis of the 2019 Kenpom data discussed above (in Figure 1).
If we allow ourselves to be optimistic, MSU still has a reasonable chance, from a historical perspective, to improve. Just from a pure basketball standpoint, this MSU team has shown flashes of what they can be during long stretches of the Notre Dame and Duke games. I believe that the Spartans have it in them. But, there are only 18 more regular season chances to prove it. Time is already running out.
Just to provide a recent and relevant example of how things can improve, I would like to bring up the specific case of the 2019 Purdue team. The historical data that I found showed that on Jan. 4 of 2019, Purdue was unranked with a 9-5 record that would soon become 9-6 due to a thrashing at the hands of the Spartans in East Lansing. The Boilermakers were ranked No. 19 in Kenpom with an adjusted efficiency margin of 19.6.
After the loss to Michigan State, Purdue would only lose three more games prior to the NCAA tournament. The Boilermakers entered the tournament as a No. 3 seed and a No. 13 national ranking. At this point the Boilermakers’ Kenpom adjusted efficiency margin had climbed to 26.0, good for No. 10 overall. That team would go on to come within an eyelash of beating eventual National Champion Virginia in the Regional Final. MSU’s current efficiency margin is worse than that of Purdue in 2019, but there is also a little more time left.
Finally, Figure 5 below gives the updated projected game-by-game odds for the Spartans’ remaining regular season schedule.
Several of the remaining games on the Spartans’ schedule that previously were projecting as wins have crossed the center line and now project as losses, including MSU’s next game at Minnesota. The three games after that (at Nebraska, versus Rutgers, and versus Purdue) still all project as likely wins, but they all more accurately might be described as “must wins.” The back end of the schedule still looks brutal and MSU needs to start racking up wins quickly... before it’s too late.
In total, that was a pretty depressing update for this Holiday Weekend. Sorry. While the numbers are currently apt to make us feel a bit blue, there are still a lot of games left to be played with one of the best coaches of all time on the bench trying to figure this all out. As stated above Coach Izzo has never finished with a losing record in Big Ten play and I do not expect that streak (or the Tournament streak) to end this year. I have faith that this team will turn things around. But, the odds that things will get fixed in time to hang a banner are decreasing rapidly.
That is all for today. As always, enjoy, and Go State, Beat the Gophers!