# Possession-based data say Pitt’s a champ, but Louisville isn’t

Ken Pomeroy has been publishing tempo-free statistics for nine years now and basketball analysts are relying more and more on them to gain insights into the college game. If you follow this blog , you know that KenPom stats are among several factors I use to identify the types of teams that overachieve and underachieve in the tourney.

For those of you who aren’t clued in to possession-based statistics, you owe it to yourself to visit www.kenpom.com. The basic idea is that raw numbers like points scored and allowed are only meaningful in the context of the number of a times a team possesses the ball or defends against a possession. In other words, the most accurate way to gauge a team’s offensive or defensive ability is to analyze its efficiency in scoring or preventing scores. Consider this: which team is better offensively—a grind-it-out team that has 60 possessions in a game and scores 72 points, or a greyhound squad that has 80 possessions and gets 80 points? Sure, the greyhounds score more points, but they get an average of only one point per possession. Meanwhile, the grinders score an average of 1.2 points.

These stats are called “tempo-free” because they take the bias of game pace out of the equation. Tempo-free statisticians have devised formulas to calculate a whole set of numbers. The three key stats, however, are:

• The number of points a given team scores per 100 possessions (called “offensive efficiency”)
• The number of points a given team allows per 100 possessions (called “defensive efficiency”)
• The predicted winning percentage of a team based on its offensive and defensive efficiency (called “Pythagorean winning percentage”)

I could go into great detail on how these numbers are calculated, but it would take a lot of explaining. Besides, Ken has already done it, and much better than I ever could. They’ve even adjusted the stats for the quality of opponent and home court advantage.

I’ve been reluctant to draw definitive conclusions from possession-based stats because I was waiting until I had a larger sample size. But a reader asked if the stats could be used to identify potential champions. I’d did this once in a December 31 post and the results were surprising.

So I’m doing it again. I examined the possession-based data for the last nine dances. The ultimate champion went into the tourney with no lower than a 115.1 offensive efficiency (OE) value and no higher than a 92.2 defensive efficiency (DE) value. Kemba Walker’s third-seeded UConn squad set both these upper and lower limits in 2011. The 2009 North Carolina squad was equally soft defensively, but their OE was 123.8.

Let’s assume that this year’s champion will fall into this range, with an OE no lower than 115.1 and a DE no higher than 92.2. Based on Pomeroy’s calculations, which teams are the best candidates to win the tourney? Take a look at this chart of the top 20 teams based on KenPom’s Pythag rating:

Only six teams are efficient enough on both ends of the court to match the qualities of the last nine champions. See the names listed in red and all caps along the bottom of the chart? Florida, Duke, Indiana, Michigan, Minnesota and Pittsburgh rate out as potential champs.

Interestingly, all but 18 of the top 20 teams met the defense criteria, but 12 fell short of the offense limit. I’ll bet that offensive efficiency is down overall this year while defense is up. If that’s the case, we’re better off filtering the teams by their OE and DE ranking.

So let’s do another analysis. The last nine champions have owned an offensive efficiency rank no lower than 17 and a defense rank no lower than 25. Here’s what the same teams we just studied rank for offense and defense possession-based efficiency (number in red fall outside our limits):

• TEAM (OE, DE)
• FLORIDA  (2, 2)
• LOUISVILLE  (15, 1)
• DUKE  (7, 5)
• INDIANA (5, 14)
• Michigan  (1, 45)
• KANSAS  (16, 4)
• MINNESOTA  (8, 16)
• Syracuse  (21, 3)
• PITTSBURGH  (11, 17)
• Ohio State  (19, 10)
• VCU  (29, 6)
• Wisconsin  (23, 13)
• Creighton  (3, 69)
• Gonzaga  (4, 75)
• Arizona  (12, 31)
• Cincinnati  (58, 9)
• Kentucky  (30, 24)
• Michigan St  (34, 25)
• Miami FL  (57, 15)
• LIMITS  (17, 25)

By this analysis, two teams that didn’t make the grade with raw numbers are now on the list: Louisville and Kansas. Conversely, one team fell off: Michigan.

That means only five teams satisfied both the raw efficiency numbers and the rankings we employed: Florida, Duke, Indiana, Minnesota and Pitt. The Blue Devils and Hoosiers aren’t much of a surprise; they’re among the top five schools in the AP rankings. Minnesota (#9 AP) and Florida (#10 AP) might raise a few eyebrows, since they’re ranked below teams like Louisville, Kansas, Syracuse and Arizona. The real shocker here is Pitt. The Panthers aren’t anywhere near the AP Top 25. But if you go by possession-based data, they’re the ninth most efficient team in the land.

So would you put a bet down right now that Florida, Duke, Indiana, Minnesota or Pitt would be the 2013 champion? I wouldn’t either. I need to see more conference play before I get comfortable with the idea that this year’s champ will follow in the footsteps of the last nine. But one thing’s for sure: I will be paying attention to the teams that make the KenPom grade come Selection Sunday.

This entry was posted in Basic Concepts, Champ Credentials. Bookmark the permalink.

### 8 Responses to Possession-based data say Pitt’s a champ, but Louisville isn’t

1. Rob says:

• ptiernan says:

Thanks Rob! Hope you’re doing well.

2. Matt says:

Good stuff. Glad to see I could prod you further in this area.

• ptiernan says:

Thanks for the nudge, Matt. Between this method and the traditional champ check, it should be interesting to see if we can zero in more finely on the set of potential champions.

3. Matt says:

Luke Winn over at Sports Illustrated did a similar-type post last Friday at si.com. Interesting comparsions: http://sportsillustrated.cnn.com/college-basketball/news/20130118/college-basketball-early-warnings/

• ptiernan says:

There’s one gigantic difference here, Matt. Luke is looking at data from AFTER the tourney. I only use pre-tourney data. After all, that’s the only data that matters when you’re trying to fill out your bracket. Numbers that bake in the results of the tourney after the fact have no predictive value.

• ptiernan says:

Just as an example…before the 2011 tourney, UConn was ranked 17 in OE and 25 in DE. It was only after their success in the dance that their rankings when down to 16 and 14. That’s a big difference, especially on the defensive side of the ball. We can’t use data that’s the result of success to predict success.

• Dan says:

Pete, You are 100% correct. One of the biggest flaws in the Kenpom historical data is the inclusion of the Tournament #’s, I could never figure out how to back out these #’s to get accurate Pre-Tourny OE and DE #’s and I appreciate you doing this..