NBA Stats Inflation, and why The Tick Cartoon is like Michael Jordan
Revisiting a topic of why "Advanced Stats" are difficult, and comparing it to a great cartoon.
One of my other creative endeavors is the NERD! Podcast with Jeff May. Both being children of the 90s, we’ve been discussing nostalgic properties we grew up with, particularly how, for a variety of reasons, some age poorly. On our most recent episode on the Ben Edlund indie comic “The Tick,” a funny line came out - “Basically, the Tick, the Animated Television show, is the Michael Jordan of cartoons.” This was uttered right after Jeff May made a controversial take - Batman the Animated Series, the award-winning show widely regarded as one of the greatest ever? It’s got more bad episodes than people remember. I’d advise you to listen to NERD! if you’re a fan of 80s and 90s nostalgic properties. However, I want to use this newsletter to expand upon a tough stats issue: why comparing players in the NBA is hard.
Also, if you enjoy the NERD! podcast and want to help it out? Jeff’s Patreon is here.
Also, here is a shameless plug of the Tick episode here:
The Wins Produced Formula
Let’s break down the “glasses” I use for viewing NBA players, which, for any cross-generational comparison, is a must, as we’ll see shortly. The Wins Produced formula uses some very straightforward math and Econometrics, which are lacking from many “Advanced Stats” in basketball. The three significant steps are as follows.
Translate all boxscore statistics into a “common currency” and use that to get a raw value. For example, a 2-point shot is worth … a point because a shot costs … a point (2-1). Turnovers are worth -1. All this works because all boxscore stats can be based on points or possessions.
Ensure that players are classified by their positions, such that there are an equal number of minutes in each position (PG, SG, SF, PF, C). Many players have multiple positions. Newer play-by-play data on sites like basketball-reference can make this breakdown incredibly granular (example here).
Compare a player’s raw value from 1. with the average performance at their positions from 2. This tells you how much better (or worse) than average a player is, and this can easily translate into point-margin (think, the Vegas Spread, this player is worth +3 points to the line) or wins, e.g., Michael Jordan was worth 26 wins in 1989.
Joel Embiid and Kareem Abdul-Jabbar
One of the most vexing players to me is Joel Embiid, a player I loved in the draft and applauded Sam Hinkie for selecting at #3 when he fell from his top spot due to injury concerns. Of course, he missed his first two seasons due to injury, only played half of his third season, and as of this writing, has only crossed 2,000 minutes played twice in his career. So, the injury concerns had some merit.
Regardless, Embiid’s name routinely shows up in the MVP discussion. With apologies, that is absurd due to time played and performance relative to other big men in the NBA, such as reigning MVPs Nikola Jokic and Giannis Antetokounmpo.
But, let’s talk about that 1. from the Wins Produced section. On Boxscore Geeks on a player’s page, you can see this as the ADJP48 number. (Adjusted Production per 48 Minutes), which tabulates a player’s contributions to winning before looking at their position or the league they play in.
At the All-Star Break this season, Joel Embiid has a per-36 stat line of:
34.1 Points (on 64.6% True Shooting)
10.5 Rebounds
4.2 Assists
1.2 Steals
1.6 blocks
3.6 Turnovers
3.3 Personal Fouls
Joel Embiid’s Boxscore Geeks page here
This has netted him an ADJP48 of 0.484 or, in 2022-2023, NBA Center Wins? Roughly five wins, which doesn’t even place him top ten in Centers, let alone all NBA players. For perspective, Nikola Jokic, the reigning MVP, has an ADJP48 of 0.697.
But let’s go way back to 1977-1978, when Kareem Abdul-Jabbar placed 4th in MVP voting and, according to the Wins Produced formula, was the top Center in the league. Here’s his per-36 stat line:
32.0 Points (on 61.2% True Shooting)
12.7 Rebounds
4.3 Assists
1.5 Steals
2.9 Blocks
3.3 Turnovers
2.9 Personal Fouls
Kareem Abdul-Jabbar’s Boxscore Geeks page here.
If you thought, perhaps, the difference in points relative to rebounds and blocks evened out? You’d be right. Kareem’s ADJP48 was 0.486, a rounding error better than Embiid’s. Yet, when the dust settled, Kareem was worth over 15 wins to his team. Even if Embiid stays healthy, he’ll be lucky to sniff 8, as the season is already 70% over. So, what gives? Well, the league got better. In Kareem’s day, the league average True Shooting was 51.5%. This season? It’s 58.0%. Ted Williams is famous in baseball for hitting 0.400, a number we’ll likely never see again. In the NBA? It’s the reverse. Players score better, thanks largely to the three-point line and better shot selection. Numbers that were MVP-worthy in the old NBA are pedestrian now. Embiid even looks excellent compared to players like Shaquille O’Neal in his 2001 MVP season. But the issue is, he plays in a league where a lot of the players, thanks in large part to players like Shaq, both increasing the popularity and science of the sport, are now better!
That said, let’s give MJ his due as a result.
MJ and Luka Doncic
Let’s do a similar experiment on two players. In 1989, Michael Jordan had a per-36 stat line of:
29.1 Points (on 61.4% True Shooting)
7.2 Rebounds
7.2 Assists
2.6 Steals
0.7 Blocks
3.2 Turnovers
2.7 Personal Fouls
Michael Jordan’s Boxscore Geeks page here.
Good enough for an ADJP48 of 0.457 (Wait, you say, that’s worse than Embiid’s … we’ll get there)
Luka Donic, who is an MVP-worthy candidate this season?
32.8 Points (on 61.4 True Shooting% wow!)
8.7 Rebounds
8.0 Assists
1.5 Steals
0.5 Blocks
3.6 Turnovers
2.6 Personal Fouls
Luka Doncic’s Boxscore Geeks page here
If you think the steals and fouls are almost a wash for the points? You’d be right. Doncic’s ADJP48 is 0.462, a thin margin in front of MJ.
Comparing is Hard!
If you took Kareem Adbul-Jabbar in the middle of his prime when he was the best Center and a perennial MVP candidate? He wouldn’t even be a top-10 player at his position now. But MJ? If you took 1989 MJ and put him in the modern NBA? His numbers would still be MVP-worthy. Sadly, both 1989 MJ and 2023 Luka get less credit MVP-wise than they deserve due to weaker squads. I digress.
Now, a funny aspect of basketball is the love of the meme of “positionless basketball.” This isn’t a thing. The rules limit how many players are allowed close to the hoop. Players assigned as the primary ball handlers have been and are still a thing. The three-point line ensures the need for players that can shoot three-pointers and guard players that shoot three-pointers. It is not shocking that the players tasked with staying by the hoop have higher production in a vacuum, as they are more likely to get higher-percentage shots and more rebounds — the Wins Produced formula does adjust for team impact of rebounds and assists. But, at a more meta-level, it respects that basketball is a team game, with players having different responsibilities!
This experiment, though, shows why player comparison gets so tricky. If I compare Michael Jordan’s best season to a modern MVP-worthy season from a top guard in the league? He fits right in. But the same numbers would be bad for a center in 2022-2023 but good for a center in 1977-1978. Knowing what a player’s numbers mean without knowing the context of those numbers or the role they have been tasked with? It is impossible without some governing math.
Regardless some fundamental lessons hold. Production in the NBA has improved dramatically, much like the quality of comic television. In both cases, an all-time great like Batman the Animated Series, or Kareem Abdul-Jabbar, might not hold up to today’s standard. On the same front, there are diamonds like The Tick the Animated Series and 1989 Michael Jordan. And in both cases, there are enthusiastic nerds to overanalyze them! We’ll see if the next episode of NERD! can elicit such a long math post! Well, you can be assured of long basketball math posts regardless.
Until then!
-Dre