Quazzum69
Stable Genius
- Joined
- Oct 27, 2012
- Messages
- 3,921
- Like
- 6,107
Ok, here it is, folks: the hard, empirical, scientific (kind of) evidence that proves Jim Boeheim does bot underachieve in March. Some methods may not be appropriate and most definitely violate some statistical assumptions, but it does give reasonable results that align with most conventional wisdom.
If you take the results of the past 28 years, you can compute the average and standard deviation of the number of wins for each seed. Obviously, the one seeds have averaged the most wins/tournament (3.4), then the two seeds (2.4 wins/tournament) and so on. You can do some tests to see which seeds are significantly different and it turns out seeds 1, 2, 3 (1.9 win/tourney) and 4 (1.5 wins/tourney) are all significantly different from each other, 5 and 6 are the same (1.2).
So, with the means and standard deviations from each seed, you can kind of create a percentile score, which is the probability of teams of equal seeds that perform worse; basically, how the particular year's team fared relative to other teams of the same seed from the past 28 years. This has nothing to do with whether you underachieve during the regular season.
Coach K has had 20 one/two seeds out of 27 tournaments compared to Boeheim's 7 one/two seeds out of 23 tournaments; of course K is expected to achieve more in the tournament and he has. You can argue all day long about player talent, regular season success, overall legacy, etc. but this is irrelevant if you want to have an unbiased comparison of tournament success (i.e you need to account for seeding, which is just a proxy for how good a team has been up to the tournament and, most likely, during the tournament).
For each tournament, you can get a score based on the number of wins and the seed. For instance, if you win two games as a one seed, you'd get a score of .19 but as a seven seed you'd get a score of .91. There are problems with this simplified approach (especially with non-normal distributions of discrete counts) but, whatever. It's better than some talking heads say how crappy JB has been while giving minimal evidence. Granted, he is lower than many "great" coaches but he higher than some, too. Preparing for the tournament and positioning yourself for higher seeds during the regular season is also part of it, which is why what Coach K has done at Duke is pretty amazing.
Here's the complete list (since 1985, minimum of ten tournaments):
Coach/average tournament percentile
Izzo 0.626
Smith
0.6
Crum 0.595
Pitino 0.578
Calhoun 0.573
Donovan 0.56
Tubby 0.554
Richardson 0.551
Calipari 0.548
K 0.545
Self 0.543
Roy 0.54
Thompson 0.532
Matta 0.521
Sutton 0.502
Boeheim 0.497
Fisher 0.473
Knight 0.409
Olsen 0.395
Barnes 0.388
If you take the results of the past 28 years, you can compute the average and standard deviation of the number of wins for each seed. Obviously, the one seeds have averaged the most wins/tournament (3.4), then the two seeds (2.4 wins/tournament) and so on. You can do some tests to see which seeds are significantly different and it turns out seeds 1, 2, 3 (1.9 win/tourney) and 4 (1.5 wins/tourney) are all significantly different from each other, 5 and 6 are the same (1.2).
So, with the means and standard deviations from each seed, you can kind of create a percentile score, which is the probability of teams of equal seeds that perform worse; basically, how the particular year's team fared relative to other teams of the same seed from the past 28 years. This has nothing to do with whether you underachieve during the regular season.
Coach K has had 20 one/two seeds out of 27 tournaments compared to Boeheim's 7 one/two seeds out of 23 tournaments; of course K is expected to achieve more in the tournament and he has. You can argue all day long about player talent, regular season success, overall legacy, etc. but this is irrelevant if you want to have an unbiased comparison of tournament success (i.e you need to account for seeding, which is just a proxy for how good a team has been up to the tournament and, most likely, during the tournament).
For each tournament, you can get a score based on the number of wins and the seed. For instance, if you win two games as a one seed, you'd get a score of .19 but as a seven seed you'd get a score of .91. There are problems with this simplified approach (especially with non-normal distributions of discrete counts) but, whatever. It's better than some talking heads say how crappy JB has been while giving minimal evidence. Granted, he is lower than many "great" coaches but he higher than some, too. Preparing for the tournament and positioning yourself for higher seeds during the regular season is also part of it, which is why what Coach K has done at Duke is pretty amazing.
Here's the complete list (since 1985, minimum of ten tournaments):
Coach/average tournament percentile
Izzo 0.626
Smith
0.6
Crum 0.595
Pitino 0.578
Calhoun 0.573
Donovan 0.56
Tubby 0.554
Richardson 0.551
Calipari 0.548
K 0.545
Self 0.543
Roy 0.54
Thompson 0.532
Matta 0.521
Sutton 0.502
Boeheim 0.497
Fisher 0.473
Knight 0.409
Olsen 0.395
Barnes 0.388