By Kurt Bullard
Come March Madness, there are few things that one should undoubtedly expect to unfold. But over the last decade or so, one thing that people have grown accustomed to is Michigan State outperforming its seed. The Spartans have seemed to make several sustained runs in the NCAA Tournament as of late regardless of seed, whether it be making the Final Four as a seven seed last year, or the Elite Eight as a four seed the year before.
Several people, including Neil Paine of FiveThirtyEight, have suggested that Tom Izzo is the best coach in modern NCAA Tournament history. And, at a surface glance, it’s tough to argue. Paine measured relative performance by looking at actual performance against the expectation for that team’s respective seed, similar to how I did so recent article about previous NCAA Tournament experience not mattering much for coaches.
There are two reasons MSU could be outperforming its seed. For one, Tom Izzo could actually adding extra value in March. However, another reason could be that Michigan State has be consistently underseeded, which would make MSU’s outperforming expectations in March Madness seem less impressive. The Spartans have a reputation for never having a “star” player (except perhaps this year, with Denzel Valentine at the helm), and the Big Ten has shown slight (albeit not statistically significant) trend of being underseeded. Both could have an affect on the way the Selection Committee seeds the teams, which, as John Calipari calmly explains, few really understand, and could leave Michigan State underseeded.
Rather than measure relative performance based on performance compared to the team’s actual seed, I wanted to measure performance based on how the team should have been seeded. How to do so is not crystal clear, but I settled with the following methodology: I would reseed teams based on their rankings according Ken Pomeroy, with the four highest teams in the field being “projected” as one seeds, and so on with the rest of the teams in the NCAA Tournament. Here are the actual and projected rankings for MSU from 2002-2015.
There were three times when the team was underseeded by at least three seeds, suggesting that the team has been severely underseeded in the past. I then calculated the average of the wins by “projected” seed. The following is a chart comparing the average wins for projected seeding versus actual seeding:
I then re-measured relative performance by measuring NCAA Tournament wins against these two measures of expectations for Izzo from 2002-2015, and added the residuals (Tournament Wins minus Expectation) to find the wins above expectation that Izzo added over these 14 years at MSU.
Tom Izzo Wins Above Expectation (Actual Seed): 11.20 WAE
Tom Izzo Wins Above Expectation (Projected Seed): 5.98 WAE
The difference in the WAE is pretty noticeable, with Tom adding five wins less using the second measure. To see if Izzo had outperformed expectation based on projected seed, I ran a t-test on the residuals of the projected seed model. The results are as follows:
The p-value is not under the significance threshold, meaning that over the last 14 years, Izzo hasn’t significantly outperformed his team’s “projected” seed. However, the test for outperforming his actual seed is significant, meaning that Izzo indeed has led his team to better results than one would expect after only looking blindly at seed.
I ran the same analysis for all coaches, and below is a table for some big-name coaches from 2002-2014, showing the WAE that the coaches have had per year:
Over the last thirteen years, Roy Williams has exceeded the projected seed expectation more so than his other famous counterparts, while Coach K—even with winning one championship over this stretch—has fared far worse than the others. Even after including last year’s championship run in 2015, Coach K would still be the lowest on the list (-.49 WAE/Year).
Tom Izzo may be a great coach. But, with his team once again underseeded, he might not be the God of March.