We talked in Part 5 about the apparent paradox of the Rockets’ offense: despite our most prominent recollections, the Rockets were a consistently good offensive team in 2016 in all of the circumstances in which we possessed measurable data. I just couldn’t help myself though. After watching the team lose what seemed like countless games by missing open 3 after open 3, after hearing commentators repeatedly say that the roster didn’t fit the scheme, and even after quipping myself that “the Rockets [were] a team of bad 3 point shooters playing in a system designed to shoot 3’s,” I had to keep digging.
I compiled the table below via NBA.com’s opponent splits. Simply put, the stats you see below are how the team fared against each opponent. ‘3p% against’ is the percentage at which the Rockets shot 3’s against each particular opponent and OffRtg against is the Rockets’ offensive rating against each particular opponent. Understand that as far as analytics go, this is as rudimentary as it gets – in a data set with just two or three items, one random hot shooting night could completely skew the end result. But I’m an attorney, not a statistician. While I’m literate, this is as far as my data manipulation abilities go.
[table id=7 /]
Recall from Part 1 that the Rockets’ offense improved this year from 12th to 7th. Their shooting improved from 20th to 14th, but their three-point shooting fell from 14th to 19th. They shot 30.9 3’s a game (#2 in the league) compared to 32.7 a game last year (#1 by a wide margin). So essentially, the Rockets shot even more 3’s, shot worse, but their offense overall improved. Were it the case that the roster didn’t fit the scheme, wouldn’t the offensive rating have finished much lower in the rankings?
What I’ve been wanting to see is if the overall efficiency and shooting numbers were inflated by good performances against bad teams. Was I, or were we, just remembering the cold shooting nights as part of a confirmation bias regarding the team?
The Rockets shot 3’s at 34.7 overall on the year. Their offensive rating was 105.5 on the year. Now naturally, you’d expect the numbers to come down against the best teams. I guess I’m interested in seeing if the shooting consistently regressed.
[table id=8 /]
The above shows the Rockets’ performance against the other fifteen playoff teams. They smoked the Pistons and Raptors. And they shot above their average of 34.7 against the Grizzlies, and, ironically, the Spurs and Warriors, the two teams which you’d probably call the best in the league. But while shooting above their average against those two teams, their offensive rating was lower than the 105.5 figure the team posted for the year.
[table id=10 /]
Above were the Rockets’ best performances. Some of those opponents probably do not come as a surprise.
Ultimately, I don’t know if any of this means anything. Everything is pretty scattered. The Rockets shot above their average against the two best teams and had some of their best shooting performances against some other playoff teams. From my point of understanding, for our purposes here, my hypothesis was false: while the Rockets did smoke some of the league’s worst teams, as everyone does, they didn’t inflate their numbers just by beating up on bad teams. Their performance was pretty scattered across the spectrum.
Here’s Richard Li, via e-mail, with some more thoughts on this topic:
I think this is also a question of vocabulary. I don’t like calling the free throws and 3s/efficient shot selection/analytics trend a “scheme.” I prefer the word philosophy. It might as well just be evolution at this point, technically a theory but one that is de facto true.
Point is, every team should be trying to maximize their number of efficient offensive possessions. Hilariously, some still don’t (Lakers), but that’s another conversation. How teams maximize those possessions, though, differs wildly. And that’s what I call a “scheme.” The Warriors fast break like crazy. The Spurs operate out of the high post. Both teams are good at creating efficient possessions, just with different schemes.
Back to the Rockets. They obviously believe in evolution (efficient possessions as defined by data). They don’t have a scheme at all, so it doesn’t matter what players they have to fit a scheme because no scheme exists. But what the Rockets offense shows, I think, is that trying to achieve efficiency while not having a scheme to attain it is STILL BETTER than having the world’s greatest scheme that tries to achieve inefficient possessions. So, basically, all else being equal, shooting threes poorly is still a better option than not shooting threes at all because your coach doesn’t believe in threes or some other dinosaur-like thinking.
Memphis is a pretty good example. They have a scheme. They have players that fit their scheme. But the end result of their scheme just isn’t efficient. So their offense is consistently worse than the Rockets’ despite them having a better scheme and better players for their scheme.
In a previous post I likened the Rockets’ lack of a system to having good ingredients but just throwing them in a bowl instead of actually cooking them. I’ll continue the food metaphor, mainly because I’m hungry. Some food is just better than others. Let’s say I’m cooking a fancy five-course meal. You’re making plain toast. If I just make an average five-course fancy meal, it’s still better than the best plain toast you could make. Hell, a below average five-course fancy meal is still preferable to the best plain toast in the world. I think this is what’s happening. The Rockets understand that a five-course fancy restaurant meal is inherently better, even if they’re not great at making it. The result, however, is still preferable to the teams that insist on cooking plain toast.
Rahat again – bottom line, to close out this topic, for this series, with the data we have available, the Rockets were a consistently good offensive team. This is basically all I can conclude with what we have publicly available and with my conceptual understanding. My curiosity on this topic still has not been satiated, but I’ve exhausted the extent of my resources and capabilities. If I had the methods at my disposal, I’d seek to pinpoint shooting game by game against good teams during certain targeted stretches, rather than grabbing the aggregate data. That would give you an answer as to whether having bad shooters just randomly chuck up shots during pressure sequences is a good strategy and if any games got out of reach during these times. It would separate the data in the event they hit a few at other points when the game didn’t matter, beefing up the percentages. I suppose you could go back through the play-by-play data in the game logs and chart each of those sequences, but I don’t have that kind of free time. Maybe the Rockets can put an intern to that task.