Similarity Scores

On the eve of the 2007 NBA Draft, I have elected to reveal the beginnings of my Similarity Scores project. Admittedly, this is not an entirely new development. In fact, I have taken cues from previous articles such as this one. However, a new year brings new information and as far as I know, no one has projected similarity scores for this year's draft.

A little background: Similarity Scores are a method introduced by Bill James in baseball to get an overall sense of where a player's development is heading. The theory behind it is that similar types of players will develop in similar types of ways. In basketball, we are much more excited about a shot-blocking center than a short point guard who can't create his own shot. Similarity scores compare draft prospects to past draftees in order to see which players they statistically most resemble.

I have intentionally limited the categories across which I make comparisons. I have tried numerous combinations, and different weightings, and have been pleased to see that generally similar results are yielded as long as height, weight, and age provide the baseline. My categories are height, weight, age, John Hollinger's pure point rating (an upgrade over assist/turnover ratio), Rebounds/40 minutes, steals/personal foul, blocks/personal foul, and three-point efficiency. The reason these categories have been chosen over more obvious ones like scoring rate and shooting percentages is because except in extreme cases, scoring rate and shooting percentages generally show very little correlation from the college to the NBA level.

Now, the results! For each player, I list his number of comparable players over 800 - this number is very small in some cases, but close to 40 in other cases. I chose to weight it this way rather than have 5 comps for some players and 150 for others. I then list the average rating of the player's top 10 comps, or all comps over 800 for that player, whichever is greater. NBA players are rated on a simple 1-5 scale: 1 is a bust, 3 a starter, 5 a superstar. (NBA players drafted between 1991 and 2003 are included as 'comps' in this project)

First, potential stars:

Greg Oden
Comps: 3
Avg. Rating: 2.4
Similar to: Tim Duncan, Chris Mihm, Joel Przybilla

Oden's average comp rating is only 2.4, halfway between a bench player and a starter. Although this rating is good by normal standards, I fully expect Oden to be a star in this league and a possible championship centerpiece. I don't put a lot of stock in these results because there simply aren't many good centers who have been drafted in the last 15 years. Tim Duncan shows up as his 5th best comp, which is promising, but with only 3 comps over 800, there aren't a lot of conclusions to draw anyways.

Kevin Durant
Comps: 0
Avg. Rating: 3.9
Similar to: Chris Bosh, Chris Webber, Carmelo Anthony

At least since 1991, there hasn't really been anyone like Durant. No freshman has dominated the NCAA like he did in quite a long time. The system comes up with a lot of young forwards who went on to be stars in the NBA, which is a good sign - but Durant may surpass them all.

Mike Conley, Jr.
Comps: 0
Avg. Rating: 3.1
Similar to: Jason Kidd, Mike Bibby, Allen Iverson

Again, there aren't many players truly similar to Conley, who combined an excellent pure point rating with a fantastic steals/pf ratio, all as a true freshman PG. His closest comparable players are strong ones, and like Durant, it is his uniqueness that suggests future stardom.

Brandan Wright
Comps: 0
Avg. Rating: 3.2
Similar to: Chris Webber, Joe Smith, Rasheed Wallace

Another strong freshman talent, his comps all grew up to be gifted offensive forwards in the NBA. He may not have the desire to be transcendent, but he has all the talent to average 20+ points within 3 years.

Al Horford
Comps: 3
Avg. Rating: 2.6
Similar to: Carlos Boozer, Rasheed Wallace, Lorenzen Wright

Horford is where the system stops projecting stardom for players, but I still think he'll be fine. Although it would be a disappointment if he turned into Lorenzen Wright, his comps mostly include solid NBA forwards, and I suspect he'll fall somewhere in between Wright and Boozer in terms of NBA success.

Corey Brewer
Comps: 19
Avg. Rating: 2.26
Similar to: Michael Finley, Latrell Sprewell, Bob Sura

Although I initially was sour on Brewer using this system, I now think he might have a lot of potential. It is probably underrating him, if anything, because it doesn't really pick up on his defensive potential. Moreover, despite my questions about his offense, players like Rip Hamilton and Jim Jackson also show up on his comp list. This suggests it is not unreasonable to expect him to develop an offensive game that will complement his defensive strengths and turn him into a possible All-Star.

Question Marks:

Here are some players who my system is not as high on. A lot of these players have tons of good comps, which suggests their skills are hardly unique, suggesting they are more fit to be bench players in the NBA.

Al Thornton
Comps: 14
Avg. Rating: 1.86
Similar to: Chris Mills, Ruben Patterson, Bryon Russell

Thornton is 23 and better be ready to make an immediate impact. There are zero top-level starters and stars among his 14 comps, suggesting that at best, he will be a solid starter, and possibly just a reliable bench player. This isn't all bad, but it means he lacks the clear upside of a Thaddeus Young.

Nick Young
Comps: 24
Avg. Rating: 1.8
Similar to: Steve Smith, Jalen Rose, Bryant Stith

Although his top 3 comps include Calbert Cheaney and Steve Smith, suggesting he does have some potential to stick in the league, the majority of his comparable players were wash-outs. Dahntay Jones, Lavor Postell, Michael Dickerson, Ed Gray, etc etc. Young is going to have to be a dead-eye shooter because his game doesn't bring much else to the table.

Jason Smith
Comps: 37
Avg. Rating: 1.89
Similar to: Bison Dele, Mark Blount, Francisco Elson

Smith, besides meeting the dreaded stereotype of a big white stiff playing in a small conference, also does not yield a list of promising comps. Many players were similar to him, and the majority of them simply weren't any good. I don't really see Smith bucking the trend.

Warren Carter
Comps: 31
Avg. Rating: 1.55
Similar to: Brian Cook, Roshown McLeod, Brian Scalabrine

Sorry, Warren.

Acie Law IV
Comps: 9
Avg. Rating: 1.7
Similar to: Jason Hart, Kirk Hinrich, Alvin Williams

Although I admire Law's hart, he's a tweener, and most of his comps turned out pretty badly. However, it is worth remembering that similarity scores provide a picture, not the whole story. Although these numbers suggest Law has a lot of bust potential, if some scouts are correct, he could also become a tougher version of Kirk Hinrich, and that wouldn't be all bad.

Solid players:

Joakim Noah
Comps: 14
Avg. Rating: 2.36
Similar to: P.J. Brown, Kenyon Martin, David West

Rodney Stuckey
Comps: 22
Avg. Rating: 2.18
Similar to: Steve Francis, Aaron McKie, Latrell Sprewell

Jeff Green
Comps: 5
Avg. Rating: 2.40
Similar to: Richard Jefferson, Lamar Odom, Jason Richardson

Julian Wright
Comps: 3
Avg. Rating: 2.60
Similar to: Antawn Jamison, Jerry Stackhouse, Rasheed Wallace

Spencer Hawes
Comps: 3
Avg. Rating: 2.30
Similar to: Tony Battie, Rasheed Wallace, Lorenzen Wright