Picking up the Pace
It has been noted in a few recent local broadcasts that the Rockets believe Kyle Lowry to be the best in the league at picking up the pace when he enters the game. And if the Rockets believe it, there’s a good chance that there are some hard numbers backing it up. In this post, I am not going to attempt to validate this claim (I assume is it is based on an On/Off pace differential for every league player who gets rotation minutes). Instead, I want to examine a little more closely how the pace fluctuates over the course of the game, and how that corresponds to particular Rockets players on the floor. In the comments section of my previous entry, where I presented miscellaneous stats for each Rocket within the first 7 seconds of offensive possessions, one of the questions was whether an increase in transition opportunities coincided with the Rockets second unit (in particular: Lowry, Landry, Budinger) entering the game. Hopefully I can help address that question here, again using the Rockets play-by-play data from Basketball Geek.
The first thing I want to look at is how the Rockets pace changes over the course of the game. What do I mean by “pace”? A conventional definition is “possessions per 48 minutes”. This would work fine, but here I will use something slightly different — average time (in seconds) per offensive possession. For the Rockets, an average possession lasts about 15 seconds. But because the Rockets are so aggressive at going after offensive rebounds, looking only at time per possession could give a skewed reading on the extent to which the Rockets are looking to run. So, I also consider another metric which is the average number of seconds per possession before the Rockets take their first shot, draw their first foul, or commit a turnover.
The graph below depicts how both of these metrics change over the course of the game. I’ve segmented each quarter into 3 minute chunks, though the last “chunk” actually represents the final 3 minutes of regulation and any overtime minutes (the Rockets have played only 5 minutes of overtime so far this season). The bars represent total number of Rockets possessions in each segment this season. The red line (denoted “poss_time”) is average time per possession, and the green line (denoted “poss_first”) is average time to first shot, drawn foul, or turnover per possession.
A couple things stood out to me when I saw this. There is a considerable gap between the red and green lines, which I imagine corresponds to the Rockets’ superior offensive rebounding. The Rockets tend to slow down, overall, in the second halves, and particularly in the middle of the fourth quarters. And, finally, it was interesting to me how the number of possessions within each 3-minute chunk appears to go up over the course of every quarter. In particular, there is a prominent spike in number of possessions at the end of each quarter. Curious, I decided to look closer at the time per possession for the Rockets versus their opponent.
From this, an interesting pattern emerges. In the last 3 minutes of each quarter, Rockets opponents spend on average about 1 second less with the ball per possession. Is this a sign that the Rockets transition defense slips to close out quarters? Is it that the Rockets pick up their defensive rebounding? Are they better at forcing turnovers? I’d like to look into this further in a future post. But for now, I will proceed by considering how the pace changes based on who is on the floor. To illustrate this, I took a page from 82games.com and their “player pair” +/- chart.
The table below shows change in average time-per-possession with the player who’s name is on the very top row, versus without that player. Consider the cell for the ‘Brooks’ row, and the ‘Lowry’ column. This means that Rockets possessions are 1.6 seconds shorter when Brooks is paired with Lowry versus when he’s on the floor without Lowry. The cells are color coded so that the more red they are, the more the player at the top of the column is “picking up the pace”. Conversely, the more blue they are, the more the player at the top of the column is slowing the team down (pace-wise — not necessarily saying that’s a bad thing).
I also created the same table for the poss_first metric described above. I expect the results to be similar, and they are:
The bottom row serves as a summary of each players On/Off impact on the the pace. As we’ve been hearing, Lowry does indeed appear to have a significant impact on increasing his team’s pace. Battier, on the other hand, is at the opposite end of this spectrum. Pretty much across the board, with whichever players he’s sharing the floor with, the pace will slow down with him. Earlier results indicated that Battier does not produce a lot, in terms of individual “box score” stats, in the first 7 seconds of possessions. Couple that with these findings and it should be fairly clear — Battier is not an ideal fit for a fast-paced offense.
Returning to Lowry, observe that in his row, Brooks has the biggest impact in picking up the pace when joining him on the floor, just as its true going the other way (Lowry picks up the pace more than any other player when joining Brooks on the floor). This, of course, makes a lot of sense. Both PGs are quick, and are potent in their own way in transition. Lowry is great at pushing the ball down the floor, drawing fouls, and finding players at the basket or on the 3-point line. Brooks is a potent shooter in transition and has the greater ability in penetrating defenses that are more or less set. Put them both on the floor, and the pace should go up dramatically. Consider the following graphs:
With Brooks and no Lowry, possessions take a little longer. Adding Lowry, especially in the second halves, the possessions seemingly shorten by 2-3 seconds on average — a dramatic pace increase. But its perhaps worthwhile to look still deeper, considering also how many possessions each combination gets together over the course of the game:
The above line graphs show the pace fluctuations with Brooks, Lowry, and both together over the course of the game on average, where gaps in the lines correspond to data points with 0 possessions. The bars indicate the number of possessions within each time segment (relevant, because not all the data points for the lines are based on a significant number of possessions). From this, we see that Lowry gets the bulk of his minutes and possessions in the second and fourth quarters. In those time periods, generally speaking, it doesn’t appear to be so much Lowry by himself but rather the combination of Brooks and Lowry that contributes significantly to the pace increase. So, if the Rockets really want to increase their pace, I think I can safely suggest one of two things: (1) sub in a more transition-oriented player (Budinger!) for Battier, and (2) pair up their PGs.