Mark Twain wrote in his biography, “Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to (Benjamin) Disraeli would often apply with justice and force: “There are three kinds of lies: lies, damned lies and statistics.”
It’s a quote with a bit of irony, if for no other reason that Disraeli never actually said that, which would make the attribution of him saying that either a lie or a damned lie. That’s OK, though, as long as it isn’t a statistic.
The second piece of irony is in the less famous portion of the quote: “Particularly when I have the arranging of them myself.” That suggests two possibilities. Either Twain is arranging the numbers to convey the lie, or in arranging the numbers, he sees the lie. Either way, it demonstrates the real problem. It’s not in the numbers themselves; the deceit is in how the numbers are presented.
And such is the NBA. We often assign a meaning to the number that isn’t there. When that happens, the statistics aren’t off. Don’t blame them. The understanding, or perhaps even the intent of the presenter of the stats, can be off. But the numbers are not “lying.” They are real. They just might not mean what they’re being said to mean. When numbers don’t make sense, though, we need to understand how they can be what they are — not just dismiss what they are.
This is the first article in a series that will look at some deceptive numbers and the truth behind them. It’s not a stat a lot of people are aware of (much less quote), but the idea behind it gets bandied about quite a bit, which is why I want to use it as a “baseline.”
The Houston Rockets‘ net rating with Ryan Anderson on the court is plus-7.9. That’s the highest of any player on the team. Also, their net rating is plus-2.5 when he’s on the bench. That’s the lowest of anyone on the team. And that 5.4 point differential (obviously), also leads the team.
If you’re using net rating as a measure of how much impact a player has on the team, it would seem that Anderson is the most important player Clutch City has. And there literally is not a human being alive who would argue that case. Yet, people use the “net rating differential” as a basis for such things as the MVP award. It’s not a meaningless number, but it’s one that contains a lot of pitfalls, which one needs to be aware of.
There are three things we want to look at when viewing these kinds of stats. First, what is the position depth like? Second, what is the minute distribution like? Third, is there some substance behind the stat?
Consider a very basic and fundamental fact: When a player sits, someone else is playing in his place. This should go in the “No Duh” Hall of Fame, but this is an overlooked reality that sits at the heart of on/off stats: Very often they are more of a measure of the difference between the starter and his backup than how much value a player brings to his team (unless you’re defining that “value” by the difference between a player and his backup, i.e., “where would they be without him?” arguments.
Here is a look at the on/off numbers for the Rockets who played more than 1,000 minutes and how many minutes they played:
The two players who primarily served as Anderson’s backup, Montrezl Harrell and Sam Dekker, also had the worst net rating on the team. And while Anderson is slightly above the rest of the cluster, Harrell and Dekker are way down there. This indicates that Anderson’s big net rating boost might be more of an indication of how bad the backups were more than how good Anderson is. While Dekker and Harrell both have plenty of potential and played hard, they are both very young and have a long way to go.
This dynamic was also at play in the MVP award with Russell Westbrook. His on/off numbers were inflated by the fact that his primary backup for the first half of the season was so bad that it literally put the camera in pain (think about it, you’ll get it.)
The second aspect of this that is a bit more irritating is that just finding the difference between the net rating has a hidden fallacy in it, because it gives equal value to the player’s “off” numbers and his “on” numbers, even though they are not equal. How many players have the same number of possessions on and off the court?
This sort of thinking has a double problem in it. First, it understates what the player does when he is on the court; second, it overstates the negative aspect of his being on the court.
James Harden played roughly 6,323 possessions last season. Anderson played 4,488. That means the extra minutes that Anderson didn’t play actually worked in his favor, since the Rockets were worse without him. If I put real numbers on this, it might help.
The Rockets outscored their opponents by 425 points with Harden on the court and 48 with him off it. They also outscored their opponents by 407 with Anderson on the court and 66 with him off it. Ergo, Harden’s difference of 377 was bigger than Anderson’s 341.
Sometimes, prorated stats are helpful. But we have to be careful that we’re not equating what happens without a player and what happens with a player. What he does on the court, both statistically and logically, should account for more than what he’s not doing when he’s not on the court.
The Substance behind the stat
All that said, it’s worth noting: The Rockets had a 59.4 true shooting percentage with Anderson on the court, and a 57.7 true shooting percentage when he wasn’t. That’s a significant difference, and more importantly, the 59.4 percent is a good enough number that it merits something to look at.
Anderson is the quintessential stretch 4. He is the stretchiest 4 of all the 4s in the league. He doesn’t just make deep shots; he makes deeeeeep shots. And his ability to make them well behind the line makes the court a whole lot bigger.
There’s a difference between a corner 3 and an above-the-break 3. There’s a difference between a 3 with your toe just behind the line and one with your feet three feet away from the line. The further out you go, the more area the defense has to cover. In fact, for every foot a player can increase his 3-point threat, that adds another 22 square feet of area the remaining four defenders have to cover (provided the shooter is single covered).
When you have multiple players who can do that, the court gets very big, and defenses get very thin. That, in turn, makes life easier for your cutters and rim-runners.
Here is your top 10 in made shots at least 25-feet from the basket (ergo at least one foot behind the line):
Look at that list and see how many other power forwards are on it. You’ll notice there aren’t any. In fact, the next-most belonged to Serge Ibaka, who made 80 and after that, it was Marcus Morris at 78. So Anderson made almost as many any as the second and third-best 4s combined and he did that missing 14 games on the season.
There’s enough space on the court here to fit the Star Trek Enterprise.
Look at this shot:
And look at it the moment before the release. Look where Anderson is standing. He’s out in Curry-land.
So while there might be some deception behind the net rating differential, there is a modicum of truth there, too. Anderson’s unique ability to connect it from well behind the line makes the Rockets a better offense.