You Are Here > SR/College Basketball > CBB Blog > College Basketball Analysis

Which Coaches’ Teams Underperform Their Seeds?

Posted by Neil Paine on March 21, 2011

Watching Texas and Pitt destroy my bracket for what seems like the fifth or sixth time in the last 10 years, I was compelled to ask: is it just perception, or do Rick Barnes’ and Jamie Dixon’s teams always significantly underachieve in the NCAA Tournament?

Luckily, I can answer that question two ways. The first is to look at every NCAA Tourney game since the field expanded to 64 teams in 1985, and measure the probability of any team winning any game based on the seeds of the two teams involved. The logistic regression formula, based on 1,686 games (including Sunday’s results), is this:

Expected W% ~ =1 / (1 + EXP(0.1738176 * Seed Diff))

Where Seed Diff is simply the team’s seed # minus the opponent’s seed #. For instance, when a 4-seed plays a 5-seed, as Texas did Sunday, their seed difference is (4 – 5) = -1, which yields an expected win % of 54.3%. And when a 1-seed (like Pitt) plays an 8-seed (like Butler), the seed difference is -7, giving an expected W% of 77.1%.

Anyway, add all up of these expected wins for every coach’s NCAA career, compare to his actual wins, and you can see which coaches have disappointed the most over their post-1985 careers:

Coach Games Exp. Wins Wins Diff
Rick Barnes 39 24.31 20 -4.31
Norm Stewart 17 10.65 7 -3.65
Oliver Purnell 6 3.24 0 -3.24
Kelvin Sampson 25 15.14 12 -3.14
Gene Keady 33 21.07 18 -3.07
Lou Henson 19 12.00 9 -3.00
Eddie Fogler 8 4.92 2 -2.92
Danny Nee 6 2.92 0 -2.92
Bob Huggins 46 29.72 27 -2.72
Jamie Dixon 19 13.60 11 -2.60
Skip Prosser 15 8.55 6 -2.55
Bob Knight 42 26.53 24 -2.53
Jerry Green 8 5.46 3 -2.46
Fran Dunphy 15 4.40 2 -2.40
Steve Lappas 6 4.28 2 -2.28
Al Skinner 16 9.28 7 -2.28
Dave Odom 19 12.16 10 -2.16
Mike Brey 15 8.02 6 -2.02
Billy Tubbs 22 15.96 14 -1.96
Dana Altman 10 3.90 2 -1.90
Dave Bliss 15 7.87 6 -1.87
George Raveling 5 2.82 1 -1.82
Bobby Lutz 7 3.77 2 -1.77
Mike Montgomery 31 18.67 17 -1.67
Jud Heathcote 14 8.54 7 -1.54
Hugh Durham 5 2.54 1 -1.54
Rick Stansbury 10 5.48 4 -1.48
Jeff Mullins 3 1.43 0 -1.43
Tim Welsh 3 1.37 0 -1.37
Lute Olson 61 40.37 39 -1.37

Well, what do you know? Barnes and Dixon are actually among the worst offenders when it comes to underperforming their seeds in the NCAA Tournament; in fact, Barnes is the worst of any coach since 1985. And at the other end of the spectrum, here are the coaches who overperformed their teams’ seeds by the greatest number of wins:

Coach Games Exp. Wins Wins Diff
Mike Krzyzewski 100 71.09 79 7.91
Roy Williams 75 50.49 57 6.51
Tom Izzo 48 28.86 35 6.14
Rollie Massimino 15 5.06 11 5.94
Rick Pitino 52 33.72 38 4.28
Jim Calhoun 60 40.03 44 3.97
Thomas Penders 23 8.12 12 3.88
Dean Smith 49 33.49 37 3.51
Steve Fisher 31 18.63 22 3.37
Gary Williams 43 24.76 28 3.24
Sonny Smith 12 3.81 7 3.19
John Chaney 38 19.03 22 2.97
Brad Stevens 11 5.26 8 2.74
Larry Brown 16 10.27 13 2.73
Tubby Smith 44 26.31 29 2.69
Billy Donovan 32 21.32 24 2.68
Jim Larranaga 10 2.43 5 2.57
Denny Crum 32 18.58 21 2.42
John Beilein 14 5.61 8 2.39
Dale Brown 18 6.77 9 2.23
Paul Westhead 7 1.78 4 2.22
Nolan Richardson 39 23.90 26 2.10
Rick Majerus 29 15.93 18 2.07
Quin Snyder 9 3.10 5 1.90
Todd Lickliter 6 2.12 4 1.88
Mike Anderson 13 5.14 7 1.86
Pete Gillen 17 6.20 8 1.80
Bill Guthridge 11 6.26 8 1.74
Dick Tarrant 7 1.30 3 1.70
Jerry Tarkanian 30 20.35 22 1.65

OK, so we’ve seen how badly some coaches underperform compared to the way their teams are seeded in the games they’ve coached. However, that’s only part of the damage done by underachieving coaches — it captures the loss in their final game of the tourney, but it doesn’t detect the missing future wins expected of a top seed going forward. Not only should a top-seeded team like Pitt have won on Saturday, but they also would have been favored in at least the next two rounds, plus could expect no worse than .500 odds in the two games after that.

From 1985-2010, a #1 seed should expect to win 3.42 games on average in any given tournament, so Pitt’s 1-win performance is 2.42 wins worse than you would expect from a team with their seeding. Throwing out the 2011 results for any team still active in the 2011 Tourney, here are the coaches whose teams underperformed the most when accounting for lost wins incurred after the team was initially upset:

Coach Tourneys Avg Seed Exp. Wins Wins Diff
Gene Keady 15 4.9 25.78 18 -7.78
Norm Stewart 10 5.9 13.41 7 -6.41
Mike Montgomery 14 5.4 22.95 17 -5.95
Eddie Fogler 6 6.7 7.70 2 -5.70
Lute Olson 23 4.1 44.52 39 -5.52
Kelvin Sampson 13 6.3 17.20 12 -5.20
Lou Henson 10 5.8 14.12 9 -5.12
Danny Nee 6 8.7 5.03 0 -5.03
Billy Tubbs 8 2.9 19.02 14 -5.02
Oliver Purnell 6 8.2 4.83 0 -4.83
Bob Knight 19 5.3 28.76 24 -4.76
Rick Barnes 19 5.8 24.66 20 -4.66
Jamie Dixon 8 3.6 15.59 11 -4.59
Dana Altman 8 9.3 6.23 2 -4.23
Skip Prosser 9 7.2 10.22 6 -4.22
Fran Dunphy 13 11.2 6.21 2 -4.21
Steve Lappas 4 4.5 5.79 2 -3.79
Mike Brey 9 7.3 9.40 6 -3.40
Dave Odom 9 5.2 13.31 10 -3.31
Jud Heathcote 7 5.1 10.29 7 -3.29
George Raveling 4 7.8 4.23 1 -3.23
Rick Stansbury 6 6.7 6.94 4 -2.94
Leonard Hamilton 5 6.6 5.91 3 -2.91
Dave Bliss 9 7.1 8.73 6 -2.73
Kevin Stallings 7 6.4 7.63 5 -2.63
Bob Huggins 19 4.9 29.60 27 -2.60
J.D. Barnett 3 7.7 3.58 1 -2.58
Jerry Green 5 6.0 5.45 3 -2.45
Ralph Miller 3 9.3 2.39 0 -2.39
Al Skinner 9 6.6 9.37 7 -2.37

Barnes and Dixon (and, oddly, Bob Knight) still look bad by this metric, although Barnes is no longer the absolute worst since 1985. Meanwhile, here were the best coaches at exceeding expected wins based on their teams’ seeds:

Coach Tourneys Avg Seed Exp. Wins Wins Diff
Tom Izzo 14 5.2 22.48 35 12.52
Mike Krzyzewski 25 2.2 65.09 77 11.91
Rollie Massimino 5 9.0 3.63 11 7.37
Steve Fisher 10 6.7 12.80 20 7.20
Rick Pitino 15 3.7 31.29 38 6.71
Denny Crum 12 5.6 15.08 21 5.92
Billy Donovan 10 4.4 16.41 22 5.59
Larry Brown 4 3.8 7.66 13 5.34
Gary Williams 16 5.4 22.83 28 5.17
Dean Smith 13 2.6 32.02 37 4.98
Jim Calhoun 18 4.3 37.04 42 4.96
Roy Williams 20 2.6 50.23 55 4.77
Thomas Penders 11 9.6 7.23 12 4.77
Jerry Tarkanian 9 4.1 17.29 22 4.71
Nolan Richardson 14 5.5 21.45 26 4.55
John Beilein 6 9.7 3.78 8 4.22
Sonny Smith 5 9.4 2.97 7 4.03
John Chaney 16 7.2 18.06 22 3.94
Rick Majerus 11 5.9 14.09 18 3.91
Ben Howland 9 5.4 15.13 19 3.87
Tubby Smith 16 5.4 25.31 29 3.69
Brad Stevens 3 7.0 2.60 6 3.40
P.J. Carlesimo 6 5.0 8.88 12 3.13
Jim Larranaga 5 11.8 2.01 5 2.99
Jim Boeheim 22 4.0 37.04 40 2.96
Jim O’Brien 7 5.9 8.20 11 2.80
Pete Gillen 9 10.4 5.25 8 2.75
Clem Haskins 7 6.7 8.49 11 2.51
Richard Williams 3 5.0 3.55 6 2.45
Mike Anderson 6 8.8 4.64 7 2.36

Texas fans will cringe over this, but note the name “Thomas Penders”… That’s Tom Penders, the former Longhorns coach who resigned from the school under pressure in 1998. I should note that Penders’ departure from UT had more to do with allegations of leaked player grades and verbal abuse of players than Texas’ 14-17 record that season, but it still has to be a bitter pill to see Penders rank among the most overachieving coaches, while his successor, Barnes, ranks among the biggest underachievers — especially in the wake of another disappointing Longhorn loss this weekend,

The only question that remains is, why do I (and countless others across the country) continue to pick teams that are coached by these guys?

22 Responses to “Which Coaches’ Teams Underperform Their Seeds?”

  1. false~cog Says:

    My only issue is that you accept the seeding of the committee at face value. I think it’d be neat to use the same method but based on oddsmakers values or other sources’ rankings @ season’s end.

  2. reservoirgod Says:

    What about using Ken Pomeroy ratings to eliminate the effect of teams being poorly seeded by the selection committee?

  3. Neil Paine Says:

    Good points, I might be able to do that at some point. It would probably have to be with our Simple Rating System (SRS), though, since Kenpom doesn’t go back past 2003. Also, the big reason to use seeds is to capture public perception — this may be unfair, but the public’s expectations are based largely on the committee’s seeds, so this does measure how well the coaches have done relative to those expectations.

  4. Toby H Says:

    My biggest problem with this is that coaches are ranked better for getting to the same round when they are coaching a lower seeded team. This means a coach is getting rewarded for underperforming during the regular season and receiving a worse seed, and a coach gets hurt by overachieving during the regular season and getting a high seed.

  5. Jahiegel Says:

    Pete Tiernan undertakes annually to evaluate coaches and conferences by performance against seed expectation (an objection to which endeavor I share with Reservoirgod, although I am inclined to think that neither performance against seed expectation nor performance against objective [i.e., metric-based] expectation is sufficiently revelatory on its own), and the second set here, as it should, tracks closely with his most recent rankings when given on a per-tournament basis. I wonder whether some combination of the two methods you propose isn’t better, though; I consider that we should conclude that a coach whose team, seeded first, loses in a regional semifinal to a twelve has oftener than not underperformed more severely than a coach whose team, seeded first, loses in a regional semifinal to a four, even as the second method would detect no difference.

  6. Matt Says:

    Very interesting. What happens to the list when you also factor into the equation the number of games played?

    Meaning, it took Barnes 39 games to sink to -4.31 while it only took Purnell 6 games to get to -3.24. Likewise Coach K is at +7.91 over 100 games while Rollie’s 5.94 was in just 15 games.

  7. Jahiegel Says:

    @Matt Tiernan’s rankings are given as PASE by tournament, which is similarly useful. Purnell, for instance, is 0.804 games below seed expectation per tournament, whilst Izzo, first on the list of active coaches, is (through last year) 1.013 games above seed expectation (amongst the retired coaches whom Tiernan examines, Rollie is well first [1.473 games ASE], whilst Eddie Fogler [0.950 games BSE] is last).

  8. Wally Says:

    Looking to see what Bo Ryans record is; since he isn’t listed he must be mediocre – meaning not in the best 30 and not in the worst 30.

  9. Scott Says:

    The connotation here is that under performing coaches are bad. Shouldn’t titles won carry some serious weight? Bob Knight is under performing? Three titles and only one of them was as a number one seed (1976). Too many variables to hold any credibility for most on this “study” (number of games played, subjectiveness of the assigned seed, time frame of reference, i.e. only doing this since 1985, and more). Most of Bob Knight’s best coaching was done prior, as well as others. There are several listed on the “under performing” list I would choose any day to coach my school.

  10. PB Says:

    Please don’t try to get fancy with data if you don’t know how to use numbers.

    Your silly model is a counting stat that punishes coaches for having greater opportunity. Rick Barnes has a -4.31 differential? Well, yeah, he’s had 24 expected wins.

    If you want to be fair about it, you should be using a rate stat. That is, for every X expected wins, the coach wins Y percent of them. By that measure, Barnes wins 82.3% of the games he should. By contrast, Kelvin Sampson only wins 79.2% of the games he should.

    Yet another lazy, dumb attempt to make Barnes out to be worse than he is.

  11. Brian Says:

    Very interesting. I would agree with some of the posters that it needs to be adjusted for a number of games. For instance, Mike Krzyzewski has won 7.91 games more than would be expected. On a percentage basis, that’s 11% more wins than would be expected. Tom Izzo has 20.8% more wins than would have been expected. Some of the others, such as Rollie Massimino, have 100% or more wins than would be expected. I think you need to use a minimum number of games as a cut off. Perhaps 20 tournament games overall. Anything less than that and the numbers lose meaning.

  12. Alan Says:

    Did you include Bob McKillop of Davidson in you analysis? In 2008, #10 Davidson beat #7 Gonzaga, #2 Georgetown, and #3 Wisconsin before losing to #1 Kansas.

  13. Ali Says:

    Curious why there was no Bill Self stats in this analysis?

  14. giantslor Says:

    I, too, was surprised not to see Bill Self in the list of underperforming coaches. What gives? Has he overperformed in other years?

  15. Sicajwax Says:

    Which university are you at? hailey little sex 47267

  16. אימון אישי Says:

    Thank you for your data!
    I loved!

  17. Mjvkayzf Says:

    I’m unemployed hot porn teen video 8389

  18. Ynpdizbm Says:

    I like it a lot Sex Models Upgrading
    >:-)))

  19. Filiberto Roseth Says:

    Magnificent site. Lots of useful information here. I am sending it to several pals ans additionally sharing in delicious. And naturally, thank you to your effort!

  20. Berenice Dolinger Says:

    We stumbled over here different website and thought I should check things out. I like what I see so now i am following you. Look forward to looking into your web page again.

  21. Maurice Yarish Says:

    Unemployment will improve once we have a new administration.

  22. brindes Says:

    I really delighted to find this internet site on bing, just what I was searching for :D too bookmarked.