As you might expect from a basketball writer, I consume an absurd amount of NBA content even in the offseason. Games, film, blogs, websites, videos, podcasts… I’ve excised non-basketball media from my life. It’s a necessity when you try to know all things about all players, which turns out to be kind of hard.
One of the many, many podcasts I subscribe to is No Dunks, from The Athletic; it’s a breezy listen that has given me plenty of inspiration over the years. Recently, the No Dunks ensemble has become fond of saying that Khris Middleton is like a Ferris wheel: you never know whether you’re going to catch him soaring high or sinking low on any given day.
My recent Middleton piece focused on his insanely consistent yearly statistics, so the podcast got me wondering. Was it true that he was inconsistent game-to-game? How did he rank compared to his peers? So I decided to see if the No Dunks crew were correct, and if they weren’t, who was the biggest Ferris wheel? Which player is the NBA’s High Roller?
Warning: boring math talk ahead, but only for a paragraph. Bear with me.
I took every NBA player’s stats from every game last season and whittled it down to the 90 guys scoring at least 15 points per game and who played at least 20 minutes per game (individual games with <20 mins were excluded). I then calculated “Game Score,” an all-in-one metric that uses every part of a box score to create an individual rating for that player’s game1 (points are the biggest driver). Finally, I took the standard deviations of each player’s Game Scores and normalized them by dividing them by the player’s mean; this helps stabilize comparisons between players whose average Game Score is 15 and guys whose average Game Score is 25 (and who likely have higher variance because they have further to drop).
The TL;DR: Game Score is an all-in-one metric that summarizes a player’s box score contributions in a single game. Points/rebounds/assists/steals/etcetera good, missed shots/turnovers/fouls/etcetera bad. I looked at variations in the Game Scores of the 90 players scoring at least 15 points per game to see who was the most and least consistent.
It’s important to note that the standard deviation divided by the average doesn’t mean anything tangible in and of itself, but it is useful for rank-ordering players. So here we go!
The Most Consistent Players in the NBA
The most consistent players in the NBA, in order (with points per game for reference):
There are some exciting things here! For one, the best players in the league are generally the most consistent: Jokic, Embiid, Durant, Antetokounmpo, LeBron, Doncic (some of the other top guys just missed the cut; more on them in a second).
Clearly, being big is helpful: The top seven most consistent players are power forwards or centers, and then there’s triple-double machine Doncic, who’s 6’7” and routinely racks up double-digit assists and rebounds. Ja Morant led the league in points in the paint, so statistically, he almost acts like a big.
Zach LaVine, however, was a shock. The 6’5” shooting guard is a high-volume jump shooter, and even though he’s very effective, I never would have guessed he’d be here. Jump shooting, by its nature, tends to be higher variance, and LaVine isn’t a tremendous rebounder or assister compared to the other names on this list, which would’ve helped to stabilize his Game Scores. He just rarely has an off night, something you can independently verify if you look at his game logs. I view LaVine with new respect after this exercise.
The Least Consistent Players in the NBA
And now, the least consistent players:
Congrats to Marcus Morris Sr. for being the wildest Ferris wheel in the NBA!
Some broad conclusions jump out. Unlike the previous list, most of these guys are smaller guards (which makes sense since Game Score tends to slightly overvalue efficient big men — rebounds are easier to come by than assists, and FG% is important). Most of the guys on this list are players whose role changed a lot game-to-game (like Marcus Morris or Anfernee Simons) or young guys still learning how to bring it every night (Bey, Green, and Porter Jr.). The Rockets starting backcourt being on here is predictable but still interesting.
Jackson and Dort were somewhat high-volume and low-efficiency shooters last year, so it makes sense they’d be this low. Gary Trent is a shoot-first, -second, and -third kind of guy whose Game Score was almost entirely based on his points. On a crowded Toronto team, if his shot wasn’t on, he might not contribute much else to bolster his Score.
Buddy Hield is similar (although one thing to watch — Hield had a massive surge in assists last year after he was traded to Indiana. He’s never been much of a passer, so I’m very curious to see if this boost maintains into this coming season).
Hayward was a surprise since he’s usually portrayed as the steady veteran on a young Charlotte team. His battles with injuries have clearly sapped his athleticism and effectiveness, however, so, in hindsight, it makes sense he’s on here.
Overall, there was a clear correlation between stardom and consistency, however. That shouldn’t be a surprise — guys become superstars precisely because they can do their job at a high level night in and night out.
Here are all 90 players together on one chart. An oversimplified way of reading it:
Top: Inconsistent
Bottom: Consistent
Right: Superstars
Left: Smaller stars
Some other notable names: Kyrie Irving is the most inconsistent high-volume scorer in the league, which makes a lot of sense. His strange season that saw him floating in and out of the lineup must have contributed to the variance (although Durant, who had to deal with Kyrie’s absences and returns, was one of the most consistent players).
Tatum is rightfully knocked for his inconsistency, and it shows here. Steph had a somewhat up-and-down year, too: people have already forgotten that he had a bad second half of the year (by Steph standards) before he turned it on in the playoffs after his return from injury.
And our man Middleton? Quite consistent for his peer group (you can see him in the bottom-middle of the graph, at approximately [15, 0.4]). Overall, he was the 25th-most consistent player out of the 90-person sample. Sorry, No Dunks crew, but no Ferris wheel here!
The full list of 90 players and their numbers, for those curious:
Folks, if you like OR don’t like this kind of analysis, please let me know either in the comments or by emailing me at mike@basketballpoetry.com. I want to make sure the things I’m writing provide value to you as readers; if this is too esoteric/boring, I don’t want to waste your time or mine. And by the same token, if this is the kind of statistical analysis you enjoy, I’d like to hear that too!
At the end of the day, I can only be successful if you all enjoy my work, and I want to ensure that I’m writing for you to the best of my abilities. Thank you for making it this far (and I promise the next couple of posts will be a little less number-y).
-Mike
John Hollinger invented Game Score. The formula is: (Points) + 0.4*(FGM) + 0.7*(Off Reb) + 0.3*(Def Reb) + (Stls) + 0.7*(Assts) + 0.7*(Blks)- 0.7*(FGA) - 0.4*(FT Missed) – 0.4*(Fouls)-(TOs). It has flaws (mostly that it can’t account for defense beyond steals/blocks and tends to slightly overvalue efficient big men), but it’s easily accessible and a reasonable indicator of how a player played offensively in a given game.