The Celtics beat the Cavs on Wednesday night 102-88, finishing the first half of the regular season with a record of 31-10. It’s been a roller coaster of a season so far, from the Hayward injury, to the 16-game win streak, to the mid-December slump, to the 26 point comeback against Houston punctuated by Marcus Smart adopting James Harden as his son. Before the Celtics kick off the second half of their season against a much improved Minnesota team, let’s look back at how each player performed through the first 41 games.
One way to help gauge player performance is to compare a player’s stats to other players’ stats from previous years to find comparable seasons. Basketball-reference provides similarity scores on each player page but it is cumulative for a career and appears to only be based on Win Shares. Since I couldn’t find a good source for similarity scores for individual seasons, I decided to create my own. I’m going to briefly walk through how I created these scores. If you’d like to skip straight to the good stuff, knock yourself out.
To build similarity scores, I calculated how each player’s per game, per 36 and advanced stats compared to the league average in that season. I then compared how each player’s stats compared to league average compared to every other player’s stats compared to league average, weighting each statistic by its importance relative to determining similarity. We care more about two players averaging the same amount of points than turnovers so we assign points with a higher weighted average. Although I tried to weight each statistic by importance, it is somewhat arbitrary. Who says what specific stat makes a player more similar to another player? In this case, I do.
It’s important to remember that we are looking at each player’s stats vs. the league average in that season. For example, Larry Bird averaged 1.7 3PA per game in 1979-80, slightly more than Jaylen Brown averaged last season. Last year Isaiah Thomas attempted 8.5 3PA per game, 5x more than Bird averaged. However, teams only shot 2.8 3PA per game in 1979-80 compared to 27 3PA per game in 2016-17 so Larry Bird actually rates as a more active 3-point shooter than Isaiah Thomas for those seasons.
Why the need to compare against league averages? Well the game has changed so much between rule changes (e.g. illegal defense, hand checking) and strategy (e.g. Moreyball) that historical comparisons wouldn’t come up nearly as often if we didn’t compare to league averages. I mean do we really think that Jaylen Brown is as prolific of a 3-point shooter as Larry Legend? Plus who doesn’t want to see names like Rory Sparrow or Winston Garland pop up as comparisons? You can’t convince me that Rory Sparrow wasn’t an original member of the Rat Pack.
Here’s a few things I hope we can gain from looking at these comparisons:
- Describe current performance. Provide a fuller, more complete picture of the value a player is providing by comparing them to who their game most resembles.
- Find unique players. Players with low similarity scores show that they provide value in ways that few other players have been able to emulate.
- Project future performance. Maybe, just maybe we can use these scores to project future breakout or bust players.
To provide more insightful comparisons we will only look at players who are either one year younger, the same age or one year older than the player we are comparing them to. We will also only be looking at Celtics who have played at least 500 minutes this season. For each player I will provide best and worst case long-term outlooks as well as a brief blurb to describe the comparisons. Without further ado here are the top 10 similarity scores for Celtics’ players this season:
Best: Stephen Curry/Kobe Bryant
Worst: Damian Lillard/Gilbert Arenas
Kyrie is already an established All-Star and I don’t see much of a downside to him, barring injury. I don’t think he’ll ever reach the heights Curry has but then again no one was predicting Curry to go on to have one of the greatest offensive seasons in NBA history back in 2014.
He’s not as good of an all-around player as Kobe but the Mamba might be his best comparison. While they have their fair share of differences, they both are known for one thing – putting the ball in the basket. Each can score in a variety of ways – crafty finishes at the rim, midrange pull ups, catch and shoot, late clock situations. They both love taking difficult shots and they both excel at making them. So while Kobe was a high flying 6’6 shooting guard and Kyrie is a below the rim 6’3 point guard, I think he can provide similar offensive value to Kobe.
For Kyrie to take that next step in stardom it will have to come on the defensive end where he rates as average at best. If his defense never improves he may top out as a high volume scoring point guard without much other value added. Think Gilbert Arenas, who comes in as his 6th highest comp. Obviously a large part of his career was ruined by injuries but we’ve seen Kyrie struggle with staying healthy as well (knock on wood).
Best: Al Horford
Worst: Al Horford
Al Horford is Al Horford is Al Horford. The man is 31, he’s figured out what kind of player he is and that player is absolutely worth a max contract. All of his top comparisons are high IQ big men who are exceptional passers and most can spread the floor as well. He’s putting together one of his finest all-around seasons, setting career highs in APG, 3P% (43%!!) and BPM. While his defensive counting stats may be down from previous years he has absolutely been the anchor for the #1 defense in the entire NBA.
That he has no player with a similarity score over 90% shows both his unique attributes as a player and also the shifting nature of the center and power forward spots in the NBA.
Best: All of Australia
Worst: Timofey Mozgov
Aron Baynes is having a better season at age 31 than Mozgov had last season at age 30. Aron Baynes is making $4.3M this year while Mozgov made $16M last year. Aron Baynes did not make this commercial. Enough said. Long live Baynes.
Best: Luol Deng
Worst: Jeff Green
Brown was the guy I was most interested to run this analysis for and honestly the results are less than encouraging. The names I feared would show up are staring me in the face. Guys with loads of talent that could never quite put it all together. At worst I think Jaylen will end up as a high-level role player à la Marvin Williams but there’s a deeper, darker side of me that sees Jeff Green. However, if he continues to improve his outside shooting, defense and playmaking I think he could be a better version of Luol Deng.
A lot of his highest sim scores are sharp shooters (Ray Allen, Rashard Lewis, Sean Elliot, Gallinari) which doesn’t feel quite right. Jaylen is more of a slashing, cutting, transition menace with All-Defensive potential. On the other hand he’s shooting 40% from 3 on 4.5 attempts per game. I’m not sure Brown will ever have a season where he shoots >40% from 3 but if he can shoot between 36-40% on 5+ attempts per game he is going to be a problem and may end up blowing all of these comparisons out of the water.
Best: Kawhi Leonard
Worst: Luol Deng
His top 13 similarity scores are all 20-year-olds which speaks to the maturity in his game. Deng is quite a safe floor and that still feels like I’m selling Tatum short (and yes I think Tatum’s floor is about equal to Jaylen’s ceiling). Tatum is oozing with star potential. The pre-draft concerns that he was an iso-heavy, midrange chucking, defensive liability couldn’t have been more wrong. He’s led the NBA in 3P% for most of the season and his length allows him to alter passing lanes, block shots and finish at the rim with ease.
When you extend the age range for his similarity scores to 18-21 his top 2 comps are Rashard Lewis and Kawhi Leonard. If you combine parts of Rashard and Kawhi together I think that’s about where Tatum will end up. Not quite the defender as Kawhi, not quite the shooter as Rashard but put them both together and you’ve got one hell of a player.
Best: Byron Scott, Aaron McKie
Worst: Luther Head
My sim scores are not sure what to do with Rozier. The majority of his comparisons are to big wings (Majerle, Turkoglu, Mashburn) due to his rebounding prowess. According to Basketball-Reference he currently has the highest TRB% of any player 6’2” or shorter in NBA history (min. 1000 MP) and there’s no one else close. This recent surge in 3-point shooting makes it hard to peg Rozier’s future. He’s already a good free throw shooter so if he can develop into something like a 44/38/80 guy he’s got a bright future. If his shooting regresses back to his first two years I think Aaron McKie is a pretty accurate comparison. A gritty do-it-all player who is not a great shooter but finds ways to impact the game through his rebounding and defense. I could really see Rozier’s career going one of a thousand different ways. He’s the toughest player on the Celtics to project.
If we only look at 23-year-olds most similar to Rozier, guess who comes up as his 5th highest sim score? That’s right Mr. GM himself, Danny Ainge. Now we know why Ainge refuses to trade Rozier.
Best: Tough to say
Worst: Take your pick
Semi at least deserves a full season before we label him as the next Yakhouba Diawara or Jeryl Sasser. The good news is he’s got relatively low similarity scores across the board so he’s not all that comparable to any of these guys. He’s proven that he can play NBA level defense but he will have to shoot better than 32/29/62 to remain in the NBA. He has the 4th highest 3PAr in the league this season despite shooting just 29% from deep. You try telling those biceps to stop shooting…
Best: Dennis Rodman!?
Worst: Keon Clark
Theis will not be the next Rodman, I just couldn’t help myself. Rodman or not, I really like Theis and as our very own Ryan Mahanna has outlined, his game is well-equipped for the modern NBA. I think Theis will be a solid backup center for years to come. His #1 comp, Kyle O’Quinn, seems like a likely outcome for Theis but really if he follows the trajectory of almost any of these guys it would be a successful career. Even though Theis is already 25, he’s still only a rookie. It takes time to adjust to the NBA so I think he has a lot of room for improvement.
Best: Gary Payton/Mookie Blaylock
Worst: Lindsey Hunter/Brian Shaw
Marcus Smart = Gary Payton. If you come away from this article with one thing let it be that. Realistically, Smart will not become the next Gary Payton but I could see him being Mookie Blaylock with a little less offensive punch. Okay, he probably won’t become the next Blaylock either, but I believe in you Marcus! I love how the majority of Smart’s highest comps are to hardnosed 80s/90s players…and Delonte West, Sebastian Telfair and Alexey Shved. In a way that sums up Smart’s career so far – if you mix Alexey Shved with Gary Payton, you get Marcus Smart.
Horford and Tatum have unique skill sets, Jaylen Brown’s future might not be quite as bright as we’d like to believe, Theis is the second coming of Rodman and Marcus Smart=Gary Payton+Alexey Shved.
Is this formula for similarity scores perfect? No. Is comparing against league averages the best method? Maybe. Can we gain some insight from this analysis? I’d like to think so!
Hit me up on Twitter if you have any specific requests for a player’s similarity scores for any season.
Latest posts by Graham Allen (see all)
- Kobe Boston to German Rodman: The Celtics Similarity Scores - January 5, 2018
- A Smart Bet: Marcus’s Shot Will Improve with Experience - August 21, 2017
- The Celtics Quarter Life Crisis - May 10, 2017