More of this would help the Texas offense.
This was a trying season for Longhorns basketball fans. Texas struggled through an up and down season with a young team. Nowhere did this seem more apparent than when Texas was on offense.
At least that is the story. I have been writing all year about how the Texas offense was actually not too bad, and that the team was hurt more by their defense (particularly the rebounding). I admit up front that this is probably not the popular narrative. The problem is that the data do not really agree with the popular narrative. Why is that?
I think the answer to this question is pretty simple. The Texas offense didn't look very good at times this season, but despite looking bad scored a pretty decent number of points per possession. On the season, the Texas offense scored 108 points per 100 possessions, which was good enough to be the 50th highest total in the nation. Compare that with the number of points allowed by the defense, which was 99 points per 100 possessions, which was Texas at #152 in division one. These are just the raw totals. Kenpom.com adjusts for the strength of the opposing defenses and offenses. Kenpom.com rates the Texas offense at the #31 offense in division one, and the defense as #45.
I don't think that Texas fans felt like they were watching the #31 rated offense in the country this season, judging by the comments at this blog. And I think I know why. The Texas offense was not particularly good at a few things that really stick out. The offense at times appeared to stagnate, and Texas had a lot of trouble making shots from the field. Particularly the initial shots of a possession, which are the shots that are generated by the structured offense.
First off, we need to tackle the things that the Texas offense did well this season. Texas crashed the offensive glass. This has been a signature of Rick Barnes' teams. Texas rebounded 38% of their own missed shots, which is the 17th highest total in the country (Kenpom.com, $). Texas also got to the free throw line a lot (#34 in division one for free throw rate), and made their foul shots (FT%=73%, which is the 47th highest total in division one). While this team turned the ball over more than Barnes' teams usually do, they still did a decent job of protecting the rock, turning the ball over in a tad under 19% of possessions (#80 in division one).
But when you come home from work, crack open a beer, and watch a game, you seldom say to yourself that your team is not turning the ball over, or is getting to the foul line a lot. Maybe you notice the second chance shots, or maybe you don't. But you do notice if a bunch of guys are struggling to get open on offense. And you really notice if they are missing shots. In other words, you are probably missing a lot while the game is going on. I miss it, too. But I try to make up for it by not drinking beer while watching basketball (I stick to whiskey), and by studying the box score and play-by-play data after the game.
That said, what you saw was real. Texas did struggle to shoot from the field, and did struggle to get a lot of value from their standard half court sets. The data show it pretty clearly. If you go by the boxscore data, you find that Texas' effective field goal percentage for the season was 48.5%, which is a bit lower than the NCAA division one team median. In case you haven't been reading my columns all season long (or have been, but have been very confused) effective field goal percentage (eFG%) is a lot like regular field goal percentage, except that it weighs three point shots higher than two point shots, and thus better describes the value a team gets from shots from the field.
Better living through play-by-play data
We can go beyond the box score, and dive into the realm of play-by-play data to get a better sense of what Texas gets out of their transition offense, and their half-court offensive sets. There is an incredible amount of information that can be harvested from play-by-play data. The problem is, play-by-play data are only widely available in a raw unprocessed form, and you have to work at it to pull out useful information. Fortunately, the statgeeks here at BON do the work for you. If you haven't read pleasepaykindle's recent posts that use play-by-play data to dig into the 2011 Texas football season, you really should. And if you have been following the basketball Inside the Numbers columns, you know that I have been using a lot of play-by-play data to understand what has been happening this year. My focus has really been on shot distributions, and I have used these data to build a pretty comprehensive website of college basketball shot location statistics. But I have recently been trying to take things a step further.
All of the data below only contain information about the initial shot of a possession. This is because I really want to isolate the data that describe the initial field goal attempts that Texas got this year out of their transition offense and offensive sets. In other words, I want to put numbers to go along with what you saw while you sipped (or pounded) your beer (or beers). To help put the numbers into context, I have chosen to compare Texas' results with those of Missouri's and Kansas'. These are two of the best offenses in the conference. In Missouri's case, they were one of the very best offenses of the last decade. Additionally, I have included data that reflect the average values for all of the college basketball games in my database. While these data aren't based on every single division one game, there are over 4400 games in my database from the 2011-2012 season. Nearly every game from major and top tier mid-major division one basketball is in there. The database is just under 75 MB in size, which makes it hell to process with my VIC-20, even with the memory expansion cartridge. (I have a whole stack of cassette tapes where I store the data.)
To set the foundation for what will follow, let's first talk a bit about the statistics that describe how quickly Texas took their first shot. After taking possession of the basketball, I view a typical possession as having up to three phases. The first 10 seconds of a possession is the transition phase. I have a post that should soon go up at TeamRankings.com that discusses why I refer to the first 10 seconds of a possession as the transition phase. In brief, what happens in the first 10 seconds of a possession depends substantially on how a team took possession of the ball. For example, there is usually a pretty big difference in how things unfold in the first 10 seconds after a team rebounds their opponent's missed basket when we compare it with what happens after a team takes possession after a basket is made by the opponent. After the first 10 seconds of a possession, things seem to settle into a standard half-court game, where the manner in which a team takes possession no longer really matters on average. At the end of a possession, things change, as teams start to get more desperate to beat the shot clock. This results in lower percentage shots.
The histogram below shows the distribution of initial shots taken in the transition phase, middle phase, and end phase for Texas, Missouri, and Kansas. I have color coded things in a sensible manner. Also shown is the average distribution for the entire database. Texas, Kansas, and Missouri all get more shots than average in transition. Missouri is only slightly ahead of Texas and Kansas. Kansas does the best job of avoiding having to take initial shots late in the shot clock. Missouri and Texas are both slightly better than average at this.
The next graph shows the effective field goal percentage (eFG%) on shots taken at various points in the shot clock. You can see across the board, each of the three teams tend to get worse as possessions become longer, as does the database average. Missouri was better than 60% in transition, and just under 60% in the 11-25 second range. This is fantastic. Kansas was a bit better than average in transition, and remains above 50% in the 11-25 second range, indicating that they were doing pretty well in their half-court sets. Texas more or less matches the database average in all three time phases. Texas had an effective field goal percentage of 53% in transition, 47% between 11-25 seconds, and 43% in the last 10 seconds of the shot clock.
Digging a little deeper: What sort of shots did the Texas offense get?
Play-by-play data allow us to break down shots into three categories. There are shots at the rim (layups, dunks, and tip ins), two point jump shots, and three point jump shots. A typical field goal percentage for shots at the rim is between 60% and 65%. Two point jump shots are usually made 35% of the time. Three point jump shots go in about 34% of the time (eFG%=51%). So generally, shots at the rim are the best, followed by three point shots. Two point jump shots are the least efficient shots of all. I have made a graph to represent the types of initial shots Texas, Kansas, and Missouri get at various points in the possession. The average values for my database are also included. The graph is a bit complicated, so let's work through it slowly. The vertical axis of the graph shows the percentage of field goal attempts that were three point attempts, and the horizontal axis shows the percentage of field goal attempts at the rim. Each team is color coded with the same colors that I have used above. There are three data points for each team. One point represents the types of shots a team took in the first 10 seconds of a possession, one point the types of shots between 11 and 25 seconds, and one point the types of shots taken after 25 seconds. For example, for the shots Missouri took in the first 10 seconds of their possession, 46% were at the rim, and 35% were from three point range. In general, being in the upper right hand corner of this plot is better than being in the lower left hand corner.
If you study this plot, a couple things will jump out at you. First, Texas' shot distribution isn't that different from the database average. Both the Kansas' and Missouri's shot distributions are more favorable that the Texas distribution. Missouri and Kansas both were getting to the rim in transition at a higher rate than Texas. Missouri also got to the rim at the end of possessions more than Texas did, while Kansas got to the rim better than Texas throughout the possession. From three point range, Kansas took fewer transition threes than Texas, but their half-court offense three point shooting rates are pretty similar. Missouri took a lot of three point shots.
It is also important to look at how Texas did in terms of effective field goal percentages for different shot types at different points in the possession. These are recorded in the table below (for initial shots). I also have shown the data for Missouri and Kansas. I have omitted the database averages here, although overall they are actually pretty close to the Texas numbers. We can see that in transition, Texas' shooting numbers are not as good as Missouri's, but aren't all that different from Kansas. Kansas was better overall in transition because they got to the rim more often than Texas did. Missouri was great in transition, in large part because they shot so well on jump shots. Once the transition period ended, Texas' shooting percentages dropped off more than Missouri's and Kansas' shooting percentages. Missouri's performance in the 11-25 second range really stands out, and Kansas maintained their high field goal percentage at the rim throughout the possession (size really helps).
The final piece of the puzzle is to look a bit more at transition offense. In general, teams get their best transition opportunities off of steals. Generally, steals result in a lot of quick shots at the rim, with fairly high field goal percentages. 47% of all shots that come after steals occur at the rim within 10 seconds, and the field goal percentage on these shots is 72%. Looking at individual teams is a little more difficult, as sample sizes when we slice the data this finely become small, but generally most teams manage to convert steals into quick shots at about this rate. Texas, Kansas, and Missouri are all pretty similar in terms of getting quick and easy buckets off of steals. The main difference between these teams is that Missouri shot a higher fraction of their total shots off of steals. About 11% of Missouri's initial shots were taken in transition off of steals, whereas Texas and Kansas took about 8% of their initial shots in transition off of steals. This is right at the average for the database. Missouri turned turnovers into quick offense at enough of a rate that it helped out their offense substantially. An otherwise average shooting team with Missouri's rate of converting steals would have raised their effective field goal percentage by around 0.4%. Not a huge improvement, but still helpful.
There are two other opportunities where teams can look to score in transition. You can run off of a rebound of a miss by the opponent, and you can run after your opponent makes a basket. Generally, it is far easier to get a quick shot after an opponent miss than an opponent make. The database average is that 41% of initial shots after an opponent miss come within 10 seconds, whereas only about 14% of the initial shots after a made basket by the opponent come within the first 10 seconds of a possession. The shots after an opponent miss are generally better as well. They are far more likely to be at the rim. The average effective field goal percentage of transition shots after an opponent miss is 53%, compared with 49% after a made basket by an opponent.
44% of Texas' initial shots after an opponent miss were in transition. These shots only had an effective field goal percentage of 50%, which was primarily due to a fluke. Texas happened to shoot a really low field goal percentage on shots at the rim in this situation (60%). I think this was a fluke because Texas shot over 75% on shots at the rim in transition on steals and after made baskets by the opponent. Sample size is an issue here.
Setting that aside, the best non-steal transition opportunities generally should come off of missed shots by the opponent. This has a large effect on shooting percentages. Using the entire database, the initial shots taken after a miss by an opponent have an effective field goal percentage of 49%. This is not just the shots in transition, but includes all initial shots. After a made basket by the opponent the effective field goal percentage is 47%, as there are fewer easy opportunities in transition to raise the average. In Texas' case, their effective field goal percentage on the first shot after getting a stop was 50%, and after a made basket by the opponent was 45%. Getting stops is an area where the defense can help the offense. In Kansas' case it did. Kansas took 1 shot after recovering a miss by an opponent for every shot they took after a made basket by an opponent. Texas took 0.8 shots after an opponent miss for every shot they took after an opponent make. Texas' ratio is right about what the average is for the database. So Texas' poor defensive rebounding percentage also hurt their offense, as it reduced the number of easy transition shots that they were able to get.
Wrapping it up
Texas' offense in the 2011-2012 season had an effective field goal percentage that is close to the division one average. This is largely the result of getting about average performance out of their initial shots. Overall, Texas still had a very good offense. They managed to get to the free throw line, avoid turnovers, and create extra opportunities with offensive rebounds.
If Texas wants to open things up and get better in transition next season, they should start by getting better on defense. In particular, the defensive rebounding troubles that they had probably cost them a number of opportunities to get out in transition.