Is Coach Self a choker in big games?

I really wanted see if the data could lend some sway to answering this question one way or another. I sliced and diced and spent a good deal of analysis using my formidable database to see if there is any data to give insight into answering this question. My conclusion is..............mostly no but a little yes. Let me explain. Coach Self, at Kansas, has played in 128 games when both teams are ranked. I developed a probability array based on historical trend of each ranking matchup. Then I calculated the number of games Self should have won in all 128 games based on this probability. Here are the results.

1. In 36 non-conference games, Self has gone 21-15 when the probability says he should have gone 21-15. No choking there but nothing outstanding either.

2. In 76 regular season conference games he went 54-22 when the probability says he should have gone 44-32. Clearly just the opposite of choking. He has caused others to choke and has excelled in regular season conference games. Self has owned the Big 12.

3. In Big 12 Tourney games he has gone 11-5 when the probability says he should have gone 9-7. He continues to own the Big 12.

4. Against Blue Bloods he has gone 10-8 when probability projects 10-8. He certainly held his own.

5. Against HOF coaches he has gone 32-20 when probability projects 30-22. Certainly no choking there and actually some clutch performance.

So overall it looks like in non NCAA tourney games, if anything, Self has been more clutch and certainly not a choker. At least that is my conclusion.

Now here is where the tables turn. When we look at the NCAA tourney, the picture is not quite s rosy. Overall, in 21 years Self has gone 48-20. Not a bad record at all. That is until you look at what he should have done based on historical net zero probabilities. Based on seeding, Self should have won 49.6 games and lost 18.4. So he is 1.6 games under where he should be so he has not performed quite up to expectation in the NCAA tourney. That does not sound too dreadfully awful until you compare that to all coaches who have been to the tourney at least 10 times. There are 28 coaches who have been to the NCAA at least 10 times. It reads like a whose who in college basketball coaching. I am guessing that almost all are in the HOF. When you look at the data, it jumps out that these 28 coaches have done far far better than the average coach. These coaches average as follows

1. 19.5 trips to the NCAA
2. 57.6 total games
3. 39.6 total wins and 18.0 total losses
4. 55 national championships (1.96 per coach)
5. 37.5 projected wins and 15.9 projected losses
6. 2.1 wins above projection.
7. Only 3 have won fewer games than projected (Lute Olson -1.1, Bob Knight -1.4 and Bill Self -1.6)

Clearly, Self trails his peers (meaning HOF or great coaches) in performance in the NCAA. If you break it down further, you will see that his nemesis is not the NCAA in general, but the Elite 8. Here is a breakdown by round.

1. 1st RD - Went 19-2, should have gone 18.6-2.4 so slightly better than projected.
2. Rd of 32 - Went 13-6 should have gone 13.7 and 5.3. Okay so the second round has been a problem too.
3. Sweet 16 - Went 10-3 should have gone 8.7-4.3. He has been a slayer in the Sweet 16, 15% better than projected.
4. Elite 8 - Went 3-7 should have gone 6.1-3.9. He has been terrible in the Elite 8. Really the only data that can be brought to bear to claim he is a choker. But it seems to be limited to the Elite 8.
5. Final Four - Went 2-1 should have gone 1.5-1.5. Way to go coach.
6. NC -- Went 1-1 should have gone 0.9-1.1. Done his job.

Overall, Coach Self has held his own in what could be described as "Big" games with the singular exception of the Elite 8 and there it is nothing short or terrible.

FWIW, Tom Izzo has the most net wins above projection in the NCAA at 7.9. But his record in the 1st 4 rounds is 10 wins above projection but he is -2.1 in the Final Four and NC.

If anyone is interested, you can find this data for the NCAA here --> NCAA Coach Data

wld be no way to really measure this corpus but arent two other potential explanations at work here? one is that hes a victim of his own in season success meaning that we go into the tournaments consistently “overrated”. in addition, when youre seeded one more often than not isnt it harder to overachieve against that just statistically?

Invalid assumption. If you say we are over rated in our tourney seed, you would have to make a case that that is unique to KU and Self to the exclusion of all other coaches. The data over 51 years of tourney's since they began seeding has proven to be extremely consistent in seed outcomes by matchup. I looked for anomalies in the data in regard to seed, year, team, coach, conference and several other variables and the data was all internally quite consistent and well within a projected deviation. I even looked at chronological year in the tourney by team and coach thinking that while a Coach K or someone like that was seeded differently early in his career versus later. But i could find nothing. I think Self's KU teams have been seeded by and large quite fairly.

Having said all that, here is where the dataset gets a little quirky. Self has been to the Elite 8 10 times. 7 of those times he was seeded No.1, once as a 2 seed, once as a 4 seed and once as a 7 seed. Now remember, theoretically, the Elite 8 should be a 1 or 2 seed. But Self's average opponent has been a 4.4 seed. In other words, Self has been put in a bracket where a big underdog has made it to the Elite 8. Now the good news is you would think that means he has a much better probability of winning and you would be right. But the bad news is, if you lose to a much lower seed it really hits your actual vs. expectation hard. One game, the VCU game, KU was projected with a 88% probability. That hurts. Oregon was a 68% probability. As a No. 1 seed, Self has lost to 4 No. 2 seeds, a No. 3 seed and a No. 11 seed. And we barely beat a No. 10 seed (Davidson). That is not bad seeding or a victim of regular season success. That underperformance. And none of those games was particularly close. I love coach Self, but there is no getting around the fact he has been a slight underperformer in the NCAA as compared to all coaches but even more so when compared to other elite coaches.

i know better than to argue math with you corpus but i wld respectfully accept your stipulation that bill self and ku are truly different as reflected by the nearly impossible 14 straight conf titles which of course none of these coaches or programs have even sniffed at. so that accomplishment is the undeniable result of great coaching over a sustained period because after all the players change every year. so bill coaches these teams to seed levels hard to sustain when the game switches away from what works in season to what works in a one and done tourney. when you add to that my own eye test that what prevails post season is talent ( one and doners are essentially a year older) so the better recruiting teams ( not us since we clearly lag in one and doners, none on this years team for example) tend to win more often. so while the numbers may suggest slight underperformance post season i do i think it may just be a statistical tautology that says we overperform in season and then achieve exactly what we shld post season.

If you slice and dice your "formidable database" enough times you are BOUND to find something. It's called a Type I error.