Sunday, March 14, 2010

The Hummel Effect

I promised some comments about which region is the toughest ect, but first I have to tackle another question. What does it mean for Purdue that Robbie Hummel is out? If you watched on Saturday, it doesn't look good.

I attempted to recreate Ken Pomeroy's adjusted offense and defense with and without Hummel. (One minor issue that I’m going to ignore. Ken puts some additional weight on recent games and I don’t know the exact methodology.) But weighting all games equally, here is how Purdue has performed with and without Hummel:

With Hummel (27 games)
Adj Off Eff 114.3
Adj Def Eff 86.4
Pythag Winning Perc = .9615

Without Hummel (5 games)
Adj Off Eff 94.7
Adj Def Eff 85.6
Pythag Winning Perc = .7622

Odds of beating Siena with Hummel: 82%
Odds of beating Siena without Hummel: 37%
Current odds on kenpom.com: 74%

Now, you can argue that the Minnesota game was a bit of an outlier, and that there isn't enough of a sample size to say how Purdue will truly perform in the tournament. But when I comment on the toughest regions, I will be using these numbers.

Statistically, What is the Toughest Region?

Basketball Prospectus will soon have your full log5 analysis with the Final Four odds in each region. But when you want to compare teams, sometimes it is hard to visualize how much is team quality and how much is seeding. In particular, Duke’s odds of reaching the Final Four are more than twice as high as West Virginia’s odds. But does West Virginia really have a gripe about seeding? Or do the odds simply reflect Duke’s better margin-of-victory numbers? Today I sort out the impact of seeding by taking the exact same team and putting them in multiple places in the bracket.

Consider the Final Four odds for Kansas if they had been the 1 seed in each region:

Kansas’ Odds
47.2% – #1 Midwest
44.5% – #1 East
51.0% – #1 West
52.5% – #1 South (Purdue w/out Hummel)

-The truth is that placement doesn’t make a whole lot of difference this year on the top line. No matter where Kansas had been a 1 seed, their odds would only have swung about 8%.

-Seth Davis and Greg Anthony argued that the committee didn’t do Kansas and Kentucky any favors and the numbers suggest that the Midwest and East are in fact the toughest roads for a 1 seed.

-The East is toughest because it includes two of the best per-possession margin-of-victory teams in the nation in Wisconsin and West Virginia. This is actually the worst of all worlds for Kentucky. Most of the time when a team gets two tough teams in their region, it will be in the 2/3 slot, but Kentucky might have to face both #4 Wisconsin and #2 West Virginia to get to the Final Four.

-Note that I’m counting all games as neutral site games, so I’m not counting any extra advantage for Kansas playing in the Midwest.

-For those of you who are curious, if I don't make my Hummel adjustment
#1 South – 50.3% (using current kenpom.com odds)

Does seed matter?

But forget Kansas. The seed doesn’t matter much for them because they are clearly one of the best teams in the country. What about a good, but not great team? In this case, I’m thinking about an imaginary team with a Pythag Winning percentage of .9550. For this type of team, what would the Final Four odds look like in each slot? (I use the without Robbie Hummel figures in all tables below.)

25.1% - 1 South - Duke
23.6% - 1 West - Syracuse
20.7% - 1 Midwest - Kansas
19.9% - 2 East - West Virginia
19.3% - 2 West - Kansas St.
18.7% - 1 East - Kentucky

-As mentioned above, Duke has the easiest path of any team in the tournament.
-On paper, West Virginia actually has an easier road than the 1 seed Kentucky because Wisconsin is on the other side of the bracket.

18.3% - 4 East - Wisconsin
17.3% - 3 West - Pittsburgh
17.0% - 2 South - Villanova
16.3% - 3 East - New Mexico
16.1% - 3 South - Baylor
15.3% - 2 Midwest - Ohio St.
15.0% - 4 West - Vanderbilt

-Avoiding Kansas and Duke is a big deal according to the numbers. That’s why Wisconsin got arguably an easier Final Four path than Ohio St despite a worse seed. Honestly, if you are the Buckeyes, how sad are you to see Kansas and Georgetown in the way of a Final Four?

14.7% - 5 West - Butler
14.6% - 7 West - Brigham Young
14.1% - 6 East - Marquette
13.1% - 6 West - Xavier
13.0% - 3 Midwest - Georgetown

-Same terrible situation for Georgetown. How sad are Hoya fans to see Ohio St. and Kansas in the way of a Final Four bid?

-BYU actually has an easier path, facing one of the weakest at-large teams in the field in Florida, and then a potential Sweet Sixteen date with a Pitt team that has not been dominant.

-It should be apparent by now that earning a better seed isn't everything in the NCAA tournament. A bigger factor is what opponent's you draw.

12.4% - 5 East - Temple
11.5% - 8 West - Gonzaga
11.4% - 8 East - Texas
11.3% - 7 East - Clemson

-Gonzaga won’t be favored in their opening round game. But this is all about the weakness of the West region as a whole. Really, there’s no question the toughest regions are the Midwest and South as we’ll continue to see:

11.2% - 4 Midwest - Maryland
10.8% - 5 South - Texas A&M
10.6% - 7 South - Richmond
10.5% - 4 South - Purdue
10.1% - 6 South - Notre Dame
9.8% - 5 Midwest - Michigan St.

-I shed a serious tear for Michigan St. They potentially have to face co-ACC champs Maryland, Big 12 champ Kansas, and co-Big 10 champ Ohio St. to get to the Final Four. If Tom Izzo wants to live up to his reputation as a tournament coach this year, he’ll have to earn it.

-You’d think Texas A&M would benefit with Purdue missing Hummel, but they actually get one of the strongest lower seeds in the field in #12 Utah St. Texas A&M won’t be favored in the opening round game, and they especially won’t be favored when you remember the game is in Spokane.

8.5% - 8 South - California
8.4% - 6 Midwest - Tennessee
7.4% - 7 Midwest - Oklahoma St.
6.3% - 8 Midwest - Nevada Las Vegas

-Yes UNLV, you won’t even be favored in your first round game. You got a terrible draw.