Tourney Time: A Look Back at PASE

petrel made this post before the NCAA tournament, however it is interesting to look at now: to what extent do first round losses by Texas, Georgia Tech, or Rutgers reflect on coaching?

About a year and a half ago, I posted about PACE in the Pleasant Dreams blog. The idea of PACE was borrowed from the ESPN's "Performance Above Seed Expectations"

So what does PACE mean? The idea behind PACE is that if you examine the history of the NCAA tournament, you can make an estimate of how many games a #X seed is expected to win in the tournament. Let's go over the numbers from women's basketball since 1994 - the first year the women's tournament expanded - and explain each one.

3.73 - 1
2.63 - 2
2.36 - 3
1.81 - 4
1.11 - 5
1.08 - 6
0.84 - 7
0.47 - 8
0.63 - 9
0.36 - 10
0.36 - 11
0.25 - 12
0.13 - 13
0.00 - 14
0.00 - 15
0.02 - 16

We'll take a look at the first number, the 3.73. In general, since 1994 a #1 seed can be expected to win 3.73 games in the tournament. Before you question that number - doesn't a team win it all if they win six games? - remember that there are four #1 seeds in a tournament. If one of the #1 seeds wins, then three lose and we obtain this number by counting all wins and losses by any of the #1 seeded teams.

The number makes sense. It's almost 4 wins. Winning the first four games in the NCAA women's tournament takes you to the national semifinals - where you're likely to bump into another #1 seeded team, leading to an even split between those #1 teams in the national semifinal - one wins, one loses.

Also note the zeros next to the 14th and 15th numbers on the list. A #14 seed or a #15 seed has never won a game in the women's basketball tournament since the tournament expanded to 64 teams. In sixteen years, the combined record of those #14 and #15 seeds is 0-128.

Not that there haven't been a few close calls. Occasionally enough a #14 or #15 seed can get within single digits of a #3/#2 seed in the first round. The closest a #15 team has come to knocking off a #2 was last year when Texas-San Antonio took Baylor to overtime before falling 87-82 in South Bend, Indiana. This was the game where coach Kim Mulkey of Baylor was hospitalized earlier on game day due to a medication reaction. The Roadrunners actually led 71-69 in overtime before falling.

As for the #3 vs. #14 matches, the #14 team has finished within two points of the higher seeded team twice - North Carolina beat Austin Peay 72-70 in 2003 and Boston College squeezed by Eastern Michigan 58-56 in 2004. So we can conclude that sooner or later, a #14 or #15 team is going to get lucky. Will it happen this year?

Note that the #16 team has an expected wins value of 0.02...not zero. In 1998, #16 Harvard managed to beat an injury depleted #1 seed Stanford 71-67. The first-round defeat of Stanford helped #9 seed Arkansas make it to the Final Four - the lowest-ranked seed ever to make a Final Four since 1994.

An interesting pair of numbers are those associated with the #8 and #9 seeds. As a rule, a #9 is more likely to go deeper in the tournament than a #8 seed is. It might be something to think about when you're filling out your first round brackets.

The real power of PACE is that one can look at individual basketball coaches and determine how well they perform against expected value. For example, note that a #4 seeded team is expected to win 1.81 games in the tournament. If Coach X coaches the #4 seed to two wins, you can say that Coach X has earned + 0.19 wins above expected value. If Coach X wins the first game but loses the second-round game, Coach X has performed -0.81 wins over expected value. You can then follow Coach X over the years across the various ways his or her team is seeded and evaluate the coach based on PASE - how well a coach performs above and beyond his or her seed expectations.

I did this in 2008, and it was a pain in the ass to track coaches over NCAA seeds, total wins, and total NCAA appearances. I had to limit it to the previous five tournaments. Over 2004 to 2008, Pat Summitt had the best PACE with a +1.29 record - her teems could be expected to perform 1.29 games beyond the seed expectation. (Geno Auriemma was 0.57 over that time period, behind Tara VanDerveer with +0.59 and C. Vivian Stringer with +0.57.) The most overrated coach? Jim Foster of Ohio State with a PACE of -1.24 from 2004 to 2008. In general, his teams could be expected to perform 1.29 games under their seed expectation.

So what does the above tell us?

* That # 9 seeds beating #8 seeds might be good bets.
* That #2 and #3 seeds are usually sure bets in the first round.
* That Pat Summitt will probably take Tennessee to the Final Four at the least.
* That Ohio State might have an earlier-than-expected exit.

Hope you enjoy the tourney. I know I will.