clock menu more-arrow no yes mobile

Filed under:

Better living through play-by-play data: Five statistical tidbits from the 2011-2012 college basketball season

The 2011-2012 college basketball season ended months ago, and the NBA is done. The next season is months away. Even the NBA draft, which isn't actually basketball but at least relates to it, is also finished. We are in the deep dark off season, with nothing to tide us over, except for a handful of Olympic games.

With no basketball, I have been spending some of my free time writing Python programs. This is how I fill the time not taken up by work, removing poison ivy from my wife's garden, or rebuilding the band saw I bought for $35 (still in progress -- I am somewhat fearful that it will cut me in half) . For the last day and a half, I have been on a heavy programming bender. The result is something I am proud of; all of the team pages at hoop-math now contain information about the initial shots of possessions. These tables break down shooting results based on how the possession started (defensive rebound, steal, etc.), how long after taking possession a shot occurred, and the distribution of shot types. For example, using these data we can tell that Texas had an effective field goal percentage of 50% when they shot within the first 10 seconds after getting a defensive rebound, and that 36% of their shot attempts in these situations came at the rim. I also have data for each team's defense. I have been using these data for a while, but now I am able to share them with everyone in a format that is easy to use.

Now I want to play with my new toy. Follow along after the jump, where I will present five statistical tidbits derived from these data.

1. Louisville's transition defense was really good. Louisville made the Final Four more or less on the strength of their defense. The Pomeroy ratings had Louisville's defense as #1 in all of Division 1, while their offense was kind of crappy. Louisville held opponents to an effective field goal percentage (eFG%) of 43%, which ranked #4 in the country.

Most teams give up a higher effective field goal percentage in transition than they do in the half court. For example, the average eFG% on initial shots in the first 10 seconds after a defensive rebound is 53%, whereas initial shots after a rebound that occur after 10 seconds have an eFG% of 47%. Transition shots are more likely to occur at the rim, which is a big part of the reason why they are more likely to go in. Additionally, transition shots at the rim are less likely to be contested, and go in a higher percentage of the time early in the shot clock than they do as the possession drags on.

Things were different against Louisville. In the first 10 seconds after a defensive rebound, opponents only shot an eFG% of 43% on their initial field goal attempts. In the first 10 seconds after a Louisville made basket, the eFG% of opponents was also just 43% on initial shots. Louisville's opponents did get to the basket in these non-steal transition opportunities (37% of initial shots in these situations were taken at the rim), but the opponents only made 53% of these attempts. This is a really low conversion rate for shots at the rim in transition -- the average shooting percentage for these sorts of shots is about 64%. While I haven't sliced the data up enough to quantify by how much, Gorgui Dieng was surely involved.

2. Florida likes to shoot threes in transition. Florida took a lot of three point shots. Last season, 45% of their field goal attempts were from three point range. Florida particularly liked to shoot threes in transition. After a rebound, 50% of their initial shot attempts in the first 10 seconds of a possession were threes. 61% of their initial shots in the first 10 seconds after a score by an opponent were three point attempts.

Florida even liked to take threes after stealing the ball. On average, 63% of initial shots within 10 seconds of stealing the ball are at the rim, and only 22% are from three point range. This makes sense; after stealing the ball, most teams take it to the rack. But for Florida, 37% of their initial shots in transition after stealing the ball were from three point range.

3. Counter punching quickly with the three point shot seldom works out. For whatever reason, shooting percentages on quick three point attempts after a score by an opponent aren't very high. Three point attempts taken within 10 seconds of an opponent score go in 32% of the time. Really fast three point shots taken within 5 seconds of a made basket by an opponent go in only 30% of the time. Many of these shots are probably ill-advised, leading to lower shooting percentages.

When looking at individual teams in this way, sample size is always a concern. But let's set that aside, and look for a team that really shot poorly on quick three point shots after its opponent scored. One team that leaps out is Xavier, who made 6% of their three point attempts taken within 10 seconds of an opponent basket.

4. The one weakness of Kansas' defense. Kansas won the Big XII last year in what many were predicting to be a down year. After winning the conference, the Jayhawks made it all the way to the title game. The moral of the story: anyone who uses the words "Kansas" and "down year" in the same sentence together is probably full of crap.

Kansas' success was largely due to having a fairly good offense and a great defense. But there was one weakness in that defense -- Kansas could be beat in transition. After pulling down a defensive rebound, Kansas' opponents were able to manage an eFG% of 53% on shots taken within the first 10 seconds of the possession. Kansas' ability to protect the rim was part of what made it's defense great. Kansas did a really good job of making it hard to get to the rim, allowing opponents to take only 28% of their attempts at the rim. And if opponents did get to the rim, Jeff Withey was waiting there. Withey and friends blocked 17% of attempts taken at the rim, leaving opponents with a 54% shooting percentage on dunks and layups.

But in transition, things were different. After pulling down a rebound, opponents were able to convert 71% of their attempts at the rim in transition against Kansas. Teams also got off 40% of their transition attempts from three point range after rebounding the ball. If teams didn't score quickly against Kansas, it became ugly, with field goal percentages at the rim of less than 50%.

It probably isn't fair to pick on Kansas here -- this is a weakness that many teams have. Even Kentucky could be hurt in transition.

5. Texas' problems with transition defense. The 2011-2012 Texas defense really struggled with three things. I have beaten to death their problems rebounding, and we all know about their tendency to foul. Additionally, Texas' defense didn't do very well in transition. After opponents rebounded the ball, they managed an eFG% in transition of 56%. After a made basket by Texas, opponent transition shots yielded a 60% eFG%. The transition shots off of missed baskets by Texas were at the rim 37% of the time, which is about average for college basketball. Teams converted these shots 69% of the time, a bit better than average.

Texas was hurt by three point shots in transition. After a Texas miss, 39% of transition shots were from three point range, and the FG% on these shots was 40%. After a Texas score, teams were able to counter punch with quick three point shots. 48% of transition shots after a Texas score were three point attempts, and these shots went in 53% of the time.

Once settled in the half court, Texas' defense became very good. Opponents' eFG% on initial shots in non-transition situations was about 44%.

Summary and a word of caution

I am still not entirely sure all of the different ways that I will use these data going forward. I am just feeling my way through the data, as I have not previously had such a simple and easy way to browse.

We also need to remember the problems associated with small sample sizes. In many cases, we have to slice the data really finely to get this sort of information. The data may not be all that predictive as a result.

But retroactively, the sample size problem isn't that big of a deal. These data show us what happened, and help us to understand why some teams did better than others.