by Ken Pomeroy on Thursday, September 1, 2011
When the player history pages were introduced a few weeks ago, my hope was that they would become something more than just a listing of a player’s advanced stats during his career. Ideally, it would provide you with a complete description of the player’s role and abilities. Today represents a small step towards that ideal as I’ve added player comparables to those pages. This is a tool I’ve had in spreadsheet form for a few years and was previously described here.
Basically, the tool finds the five players with the stats most similar to the player being examined. There’s a more complete description at basketball-reference.com although my formulation differs slightly from theirs. I’m including the following 13 stats using the relative influence of each listed.
Team Rating (4x)
Off Rating (3x)
Assist rate (2x)
TO rate (2x)
Block % (2x)
Steal % (1x)
Shot % (1x)
Team rating is thrown into the mix as one of the comparable stats to minimize the chance of comparing Big Ten players with NEC players, for example.
One thing I have changed from my original implementation is to only compare players from the same eligibility class. Freshmen only get compared to other freshmen, etc. Also, I’ve conformed to the James/Hollinger/Pelton precedent of using higher scores for better matches where 1000 indicates an identical match. (This never happens.)
As mentioned in the referenced Basketball Prospectus piece above, there’s nothing more grating that the white-guy-compared-to-another-white-guy syndrome that pervades modern analysis. Almost everybody does this and the few that complain about it suffer from a similar bias. After all, it’s just as bad to consciously avoid comparing a white guy to a white guy as it is to do it.
This is at least an objective way to make comparisons. (Objective in so far as that right now, you are bound by the weights I chose the weights for each stat.) It’s not going to be useful for every player, or perhaps even most players, but I think it’s a worthy reality check on a player’s production by asking: What other players had similar production? With that in mind, is Jared Sullinger/Kevin Love a bad comparison? I don’t think so if you step out of the racial box.
This doesn’t mean that Sullinger’s future will mimic Love’s. Using comparables to predict a player’s future stat line is probably a useful exercise, but it should involve using a player’s top 20 (or more) comps to find a range of outcomes. And maybe you’ll see that someday as development continues on the player pages.
Something else to consider when using this tool is the level of similarity in the list of comparables. A score over 900 is a great match. Anything above 850 is still useful. Once you get below 850, then any comparisons are somewhat weak. A good example here is Buffalo’s Javon McCrea, who’s getting deserved offseason press in anticipation of what should be a monster sophomore season. There is no freshman since 2006 that produced a similarity score of better than 800. There are very few people who, like McCrea, do everything well as a freshman. It’s still encouraging that JaMychal Green and B.J./Byron Mullens are on his list of comparables, but this where caution should be used in terms of projecting a professional future. McCrea is significantly shorter than either of those guys. Nonetheless, he’s clearly your typical undersized/overtalented mid-major player.
Another example of uniqueness is Marreese Speights’ freshman season at Florida, for which only one player cracks a 700. There’s just nobody who was forced to watch so much action from the bench who was as productive as Speights was that season.
Let’s use similarity scores to examine Jeff Goodman’s recent blurb about Alec Brown, a seven-foot sophomore at Green Bay, who is apparently on the NBA radar. Goodman recounts predicting to Brad Stevens that then-freshman Gordon Hayward would eventually leave Butler early. It was a bold prediction to be sure, but it’s not like Hayward was a slouch as a freshman with comps of Robbie Hummel and Patrick Beverley on his list. Brown’s comparables are generally not as famous, but being closely matched with Andrew Nicholson indicates there’s some evidence Goodman isn’t too far off in highlighting Brown’s prospects.
By the way, if you screw around with this enough, I promise you’ll stumble across some interesting things. My nomination for “the anonymous player most likely to be a household name someday” is Saint Mary’s Tim Williams. Williams only played one in every four minutes last season, but his season-long stat-line was impressive. Three of his comps are Jordan Hill (#8 pick), Tyler Zeller (possible first-round pick in ’12), and Dwayne Collins (2nd round pick in ’10). A fourth comp, former Marshall player Tyler Wilkerson, received a semi-serious look from the Spurs last summer.
Williams is in a difficult position to get more playing time this season, but keep an eye on him when he does see action. Along those lines, let’s hope that new UNLV head coach Dave Rice has the good sense to free Quintrell Thomas. At least in Williams’ case, most of his stats were put up against weak non-conference competition. The same explanation can’t be made for Thomas’s lack of minutes last season.
One should use this tool with care when it comes to predicting the future, either in college or the NBA. You need only compare the comps for Elias Harris as a freshman and sophomore to see how a promising rookie season is no guarantee for future stardom. Like everything on this site, this is another lure for your analytical tackle box. Use it wisely.