Quantifying the Eye Test: Comparing the Best College Basketball Teams

By  John Ezekowitz

College basketball players and teams are endlessly compared to one another. Putting context to our observations is a vital part of what we do not only as hoops fans, but also in our everyday lives. Of course, our eyes often deceive us: Catch Texas at any time in January, and they look like the best defensive team since the mid-decade Memphis outfits. Catch them last Saturday against the Colorado Buffalos, and they look like the Texas team that fell apart last year. Which vision is closer to the truth?

Luckily, we have a statistical technique that can begin to answer those questions. The concept of Similarity Scores was first proposed by Bill James, and adopted to basketball in the last few years. At its most basic level, a similarity score takes in the relevant stats of the two teams one wishes to compare and creates a value that represents the difference between the two teams. By repeating this process for all teams, we can see which teams are the most similar.

I decided to run the similarity scores for the seven teams identified as possible one seeds in the tournament. I compared Ohio State, Duke, Kansas, Texas, Pittsburgh, San Diego State, and BYU to every college team from 2004 through this year. For the statheads, I adopted Neil Paine’s NBA methodology, as described here, with one major change: I controlled for strength of schedule by dividing the similarity score (as described in the link) by a factor of (1- (SOS Team A- SOS Team B)^2). I used Ken Pomeroy’s Pythagorean SOS numbers.

The results often yielded just as many questions as answers. These are not intended as predictive comparisons, but rather descriptive. Of course seeding in the Tournament matters when comparing teams and results, but the comparison itself is often enlightening.

First up, Duke:

Duke has some very close comparisons, all of whom shoot the ball very well, did not turn it over, and defend from the field well. 2009 Gonzaga was a very good team that had the misfortune of running into the UNC juggernaut in the Sweet 16. One more note: stop sleeping on George Mason this year. They are really, really, really good.

Kansas:



The first two comparisons should be heartening to Kansas fans. Kansas was the only team to have its best comparison be a National Champion—let alone its best two comps. 2005 Utah State is not an expected comparison. Those Aggies, led by the uber-efficient duo of Spencer Nelson and Jaycee Carroll, fell to Arizona as a 14 seed in the NCAA Tournament. They were woefully underseeded, however, and the Jayhawks will not have that problem. Kansas has shown the highest ceiling of any team in the nation this year, but can they put it together for six straight games in March like their best comparisons? We shall see.

Ohio State:

The Buckeyes are by far the most unique of the seven potential one seeds I examined. This is principally caused by their amazing ability to simultaneously force a lot of turnovers on defense while not committing fouls (number one in the nation in free throw rate). Interestingly, most of Ohio State’s best comparisons were guided by elite lead guards: Arron Afflalo, Deron Williams, Evan Turner (in college), Adam Emmenecker (in college) and Darren Collison. Do the Bucks have an elite PG? Only Aaron Craft has an assist rate over 20 percent. This underscores the uniqueness of the 2011 Buckeyes.

BYU:

BYU does not come out looking especially good in this analysis. The most accomplished team here is probability the 2008 Washington State squad that lost to a very good North Carolina team in the Sweet Sixteen. Let us take a moment to muse on how circumstances change over the course of a year: the stats say that the 2010 and 2011 editions of BYU are very similar, yet the Cougars will likely get a top 2 seed this year, whereas they were a 7 seed last year. Is the singular talent of Jimmer Fredette covering up a somewhat weaker team, or is that very talent enough to lead the Cougars on a March run?

Texas:

The Longhorns appear to be in a bit of a free fall right now after back-to-back losses to Colorado and Kansas State. Unsurprisingly, their best comparisons are teams that held their opponents to very low percentages from the field. The three comparisons to John Calipari teams are interesting: Rick Barnes and Calipari are not usually coaches associated with each other.

Pittsburgh:

File this under: Jamie Dixon Consistency Syndrome. It might be the highest compliment you can pay a coach. It is truly amazing that Dixon’s 2005, 2006, 2009, and 2011 teams are all so similar. All five of these teams are extremely good on the glass. Here the vagaries of March are apparent: the 2006 and 2009 editions of the Panthers were statistically almost identical, but in 2006 Pitt ran into Patrick O’Bryant and Bradley who successfully kept Pitt off the glass and pulled the upset.

San Diego State:


These are probably the worst teams of any of the best comparisons listed here. They are all squads that combine average shooting, below average free throw rates, and stalwart defenses. As I mentioned in my Free Throw Plus post, an inability to knock down shots consistently and to get to line consistently does not seem like a recipe to succeed in the long run in March. SDSU’s comparisons seem to bear that theory out. Again, it must be noted that SDSU will have a significantly higher seed than these teams (excluding Pitt), and thus may advance further based on a weaker draw.

About these ads
This entry was posted in NCAA Basketball. Bookmark the permalink.

4 Responses to Quantifying the Eye Test: Comparing the Best College Basketball Teams

  1. Brian says:

    Can you provide a CSV of the data you used? I’d like to get it into an R database

  2. Harmy G says:

    Illinois lost in the finals to UNC in 2005.

  3. Pingback: Close But No Cigar « Market Research Application

  4. Mike says:

    Great post! Thanks for this. I love thinking about the old teams. There’s no way these current teams could hang with some of those older teams. I found a pretty extensive 64-team bracket on a site. Good stuff. I found your article searching for more opinions on the best team. Nice!

    If you want to check out the 64-team bracket. It’s enjoyable if nothing else: http://www.tchuddle.com/2011/04/the-greatest-ncaa-division-i-men%e2%80%99s-basketball-tournament-teams-ever-champion/

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s