Thursday, August 14, 2008

Best Performances by NCAA Coaches in the Post-Season

I'm going to replicate the work of an interesting blog called, "Yet Another Basketball Blog " and examine the success rates of women's basketball coaches. Like Dan Hanner, I'll look at the current leaders in appearances plus wins in the NCAA, but I'm changing the time frame to five years instead of ten years. The current school of each coach is listed, with annotations if the coach has changed schools

NCAA Appearances - Wins - Coach - School

5-24 Pat Summitt Tennessee
5-18 Geno Auriemma Connecticut
5-15 Tara VanDerveer Stanford
5-14 Gail Goestenkors (Duke 4/Texas 1) Texas
5-14 Sylvia Hatchell North Carolina
5-13 C. Vivian Stringer Rutgers
5-12 Brenda Frese Maryland
5-11 Kim Mulkey Baylor
5-10 Andy Landers Georgia
5-8 Melanie Balcomb Vanderbilt
5-7 Charli Turner Thorne Arizona State
5-6 Muffet McGraw Notre Dame
5-6 Sherri Coale Oklahoma
4-6 Pam Borton Minnesota
5-4 Doug Bruno DePaul (IL)
5-4 Jim Foster Ohio State
3-5 Kristy Curry (Purdue 3/Texas Tech 2) Texas Tech
5-3 Dawn Staley (Temple 5) South Carolina
5-2 Wendy Larry Old Dominion (VA)
3-4 Gary Blair Texas A&M
3-3 Sharon Versyp (Maine 3/Purdue 2) Purdue
3-3 Elaine Elliott Utah
4-2 Jeff Mittie TCU
4-2 Bill Fennelly Iowa State
5-1 Don Flanagan New Mexico
3-3 Deb Patterson Kansas State
4-1 Joanne Boyle (Richmond 2/Cal 3) California

We now want to develop an idea called "PASE", which is used both by Henner in the blog and by ESPN. PASE stands for "performance above seed expectations".

There aren't many benchmarks we can use to compare college coaches. Win-loss record certainly won't do it. A team's win-loss record depends seemingly on two things:

a) strength of other teams in conference, and
b) strength of non-conference schedule.

However, since 1994, the women's NCAA tournament has had 64 teams, with teams receiving a rank from "1" to "16" dependent on subjective assessment of strength (and later, Sagarin RPI). What we can do is compare a team's performance based on their seed with the performance of all the other teams that had that same seed in past tournaments.

We come up with an "expected win" count per seed - looking back over performance, we can say that a "6" seed is expected to win 1.03 games. Your average six seed makes it to the second round. If you're a college coach and can take your sixth-seeded team into the Sweet Sixteen, that's a plus. If you lose your first round game, that's a minus.

First, here are the expected wins associated with NCAA seed in the women's tournament:

1 - 3.73
2 - 2.65
3 - 2.38
4 - 1.78
5 - 1.15
6 - 1.03
7 - 0.85
8 - 0.48
9 - 0.60
10 - 0.35
11 - 0.37
12 - 0.23
13 - 0.13
14 - 0
15 - 0
16 - 0.02

Why isn't the #1 seed worth more wins? Because if you make it to the Final Four, you'll probably bump against the other #1 seeds.

Note that the #8 seed has had particularly bad luck. History shows that the #9 seed does better than the #8 seed.

The #14 and #15 seeds have never won a first round game in the history of the 64-team tournament. As for #16 being 0.02, I refer you to March 14, 1998, a dark day in Stanford history as they became the only #1 seed - in women's or men's play - to lose to a #16 seed as Harvard beat a Stanford team with two injured leading scorers by the score of 71-67 in the first round.

So given their assigned seeds, which coaches have done the best? We will look over the last five years only and determine how well, per year, each coach does above the expected number of wins.

Best Coaches vs. PASE (minimum 3 appearances in five years)

Pat Summitt 1.29
Tara VanDerveer 0.60
C. Vivian Stringer 0.59
Geno Auriemma 0.57
Brenda Frese 0.38
Pam Borton 0.34
Andy Landers 0.27
Muffet McGraw 0.14
Kim Mulkey 0.13
Charli Turner Thorne 0.13

This is pretty much a list of the big names in women's college ball - the best coaches are the ones with longevity, and they have that longevity because they can keep coming to the NCAA tournament and performing beyond expectations. Pat Summitt has won a few championships recently. That helps, and we see that on the average Coach Summitt gets one game beyond where she's expected to finish.

Worst Coaches vs. PACE (minimum 3 appearances in five years)

Don Flanagan -0.16
Wendy Larry -0.24
Tom Collen -0.25
Gary Blair -0.29
Joanne Boyle -0.40
Deb Patterson -0.52
Kay Yow -0.53
Sylvia Hatchell -0.54
Sherri Coale -0.73
Jim Foster -1.24

Flanagan is a good coach at a weak school. Tom Collen wasn't so weak that Arkansas didn't run off and hire him, and the same case goes for Joanne Boyle at California. Surprisingly, Sylvia Hatchell of North Carolina and Jim Foster have had some lousy post-season results. I wonder how well Sherri Coale of Oklahoma will do once Courtney Paris leaves the Sooners.

My next goal is to examine the relationship between a team's success and the number of McDonald's All-Americans it is able to sign. That should be fun.


Q said...

I've been thinking about a way to look at coach of the year and this makes me wonder if wins above Pythagorean rating (which I know you do) could be used for that.

There are obviously a number of other factors that go into coaching, but this seems like a pretty good model to start with...good work.

pt said...

I don't think that you should use Pythagorean rating for evaluating coaches. If this were true, a good manager should be able to consistently outperform his or her Pythagorean rating. In baseball, this doesn't happen; relationship between a particular manager and team Pythagorean rating shows no correlation in one director or another. A coach can have his team "perform" above Pythagorean rating one year, and then the team will perform below it by the same amount the following year with seemingly no relation.

Of course, basketball statistics have not been examined. Maybe it *does* work in basketball. At least, it's worth taking a look at.