In the final minute of the first half of Sunday's game between North Carolina and Oklahoma, Clark Kellogg noted that despite the Sooners being dominated on both ends of the floor, the game was being played at the Sooners' preferred pace. Kellogg made the seemingly logical connection that other analysts before him have made throughout Roy Williams' tenure at UNC: the Tar Heels are very good and play at a very fast pace. The conclusion then must be that an opponent should not play at a fast pace against them.
As noted in the Sweet 16 log5 analysis, UNC is 29-1 in games played with 71 or fewer possessions in the Tyler Hansbrough/Ty Lawson era. Make that 30-1 after the victory over the Sooners. The lone defeat was to Florida State in this season's ACC Tournament, a game in which Ty Lawson did not play. Carolina is actually 26-0 in such games with Lawson in the lineup. For full disclosure, UNC has played in at least one overtime game that deserves to fall into this category--their Elite Eight loss to Georgetown in 2007--and it's possible that overtime losses to Maryland this season and Florida State last season do, too. Regardless, the point still stands. UNC is at least as likely to win a slow game as a fast one, perhaps more likely.
To better understand why the Tar Heels' performance is essentially immune to the pace of the game, I gathered data on every Carolina possession against Division I competition this season. Additionally, I've collected data on nearly a million possessions of college hoops action over the past five seasons. In all cases, I have weeded out garbage-time possessions in order to get a true measure of tendencies when the game is being contested and to avoid the effects of strategic fouling late in the game.
This provides a rich data set from which once can uncover some interesting trends in the game. For our discussion here, I'm only concerned with efficiency as a function of time elapsed in a possession. The season wouldn't be complete without at least one thought-provoking graph, so here's what that concept looks like in picture form...
The horizontal axis represents the time elapsed until what I call "the first action," which is either a turnover or a shot attempt by the offense. Therefore, when a team misses a shot five seconds into the possession, gets its own rebound and scores 25 seconds later, its score is placed in the five-second bin and not the one for 25 or 30. While these might not seem like fast-break points at first, they result from a possession that started with a fast-break attempt, thus it's only fair to include them when calculating efficiency from those attempts.
The blue line is the average efficiency for the time of first action during the possession. The red line is the percentage of possessions that contain a first action at the given time. The data here isn't particularly exciting. There's a big advantage for the offense early in the shot clock and a disadvantage late.
You might wonder why the plot extends beyond 35 seconds. After made shots, it is often impossible to discern from play-by-play data when the subsequent possession actually begins. I'm assuming it's immediately after the shot is made, but often that's not the case. Thus, you can have possessions that appear to go on longer than 35 seconds. In addition, while the entries in play-by-play data imply accuracy to the second, the folks at the scorer's table are busy people, and this kind of precision is not always achieved.
Let's return to the graph itself and what we can take away from it. Teams need to focus on taking shot attempts in the first three or four seconds of a possession! Well, no, of course not. Good shooting opportunities rarely present themselves that quickly. The issue is that the graph represents a composite of different types of possessions. For instance, we can separate out possessions that occur after a steal, and make a similar plot.
The efficiency spike in the first plot is largely due to the spike in this plot, and this is where we can start to tie things back to UNC. This data here tells you what you should already know--the best way to convert defense to offense is after a steal. Based on plots like this, I've determined that the best cutoff for defining a fast-break possession is eight seconds.
The average D-I team scores a whopping 1.33 points per possession on fast-break possessions after a steal. That's .31 points per possession more than the average on all possessions during the study period. Not surprisingly, the Tar Heels are one of the best in the country at converting steals to transition points, scoring 1.46 points per possession on those occasions. It's not just that UNC is so effective, it's also that they do it so often. The average team runs about 62% of the time after a steal; Carolina does it 80% of the time.
OK, we've established that if you're playing UNC, you don't want to hand the ball off to Ty Lawson at the mid-court stripe and watch him go in for layup after layup. If that's how you arrive at a fast-paced game, you'll end up seeing J.B. Tanner come off the Carolina bench before the final media timeout. I'll concede that point, but nearly every other piece of data we can find on possession length supports not fearing an up-and-down game against the Heels.
First, you should understand that steals are not a significant part of the Heels' offensive diet. They forced a steal on fewer than one in every 11 possessions in ACC play, the second-lowest rate in the league. Of course, steals are not the only events that present running opportunities. UNC likes to run after they grab a defensive rebound as well, and these cases are much more frequent than steals. While other college hoops teams run about 34% of the time after a missed two-point shot, UNC does it 51% of the time.
However, Carolina scores a pedestrian 1.14 points per possession in these cases, which is barely above the national average of 1.11. More importantly, this figure is lower than their efficiency in possessions that start on a dead-ball situation. The Tar Heels rack up 1.17 points per possession after a made shot or some other stoppage. This is outstanding--the national average is exactly a point per possession. Shorter possessions for Carolina drive up the possession count (which has a marginal benefit for a superior team) but not the team's offensive efficiency in a significant way.
Applying this same method to the UNC defense doesn't reveal the obvious incentives that the offensive analysis does. I'll be honest, when I last looked at this data about two months ago, UNC's transition defense didn't look good. It's improved, and now they give up 1.08 points per possession on transition opportunities after a two-point miss, which is just a bit worse than their half-court defense (1.04).
The thing is, there are plenty of opportunities to run if necessary. Their opponents' running percentage is right around the national average. But keep in mind that the easiest way to avoid live-ball turnovers is to avoid long possessions. There's not an obvious benefit to passing up good shots in the hopes of getting a great one later in the possession, and there is a major penalty to this behavior: the risk of Carolina taking the ball away and converting in three or four seconds.
Look, Nantz and Kellogg have to fill two hours of game time, and the crew in New York has to fill 20 minutes of halftime, so I can't hammer them for focusing on the pace of the game now and then. But as it has seemed to be throughout Ty Lawson's stay in Chapel Hill, tempo has little to do with Carolina's chances of winning. They are as good at scoring against a set defense as any team in the country, and longer defensive possessions increase the chance that they can turn defense into quick points.
Ken Pomeroy is an author of Basketball Prospectus.
You can contact Ken by clicking here or click here to see Ken's other articles.