Feature Watch: Player Splits
Posted by Neil Paine on September 10, 2009
After guiding Baseball-Reference's users through the player splits yesterday, we thought it would be a good idea to apply the same treatment to the player splits here at BBR today. Granted, our basketball splits are not as intensive as their baseball counterparts (yet), but they still can really help you dig deeper into a player's stats for a particular season or his entire career.
For example, let's take Dwight Howard's splits for 2008-09. On that page you can see not only his overall performance, but also his production at home vs. on the road, his value in wins vs. losses (surprisingly, not all that different), and two splits which I find very useful — his pre- and post-All-Star stats, and as a subset his performance breakdown by month. I like to use these numbers to get a real sense for how a player progressed (or fell off) as the season went on. Full-season stats are obviously the best predictor of future performance in a general sense, but in-season trends for young players who suddenly "got it" during the year can be informative, as well as the inverse: late-season collapses for older players. And on the career splits pages, you can identify players that are habitual hot or cold starters to begin the season, which should come in handy for all you fantasy basketballers out there.
But perhaps the best feature of the splits page are the breakdowns vs. each opponent. These are divided up by conference, division, and of course, the individual teams themselves. Since matchups are of so much importance in the NBA, this feature can be really great in helping you identify players who excel against a particular team, coach, or opposing playing style (although the usual caveats about sample size certainly apply). We've even discussed the possibility of adding new opponent categories in the future based on SRS, W-L%, and Offensive/Defensive Rating. This would effectively allow you to do your own "who rules the top defenses?"-style studies with any players you want.
And, lest I forget, the research potential of these splits has come into play with the recent allegations of scorekeeper fraud in the awarding of blocks and assists. It was said that someone like Dikembe Mutombo (and many other Denver bigs, for that matter, so not just Deke) were getting huge benefits of the doubt when it came to blocked shots, and if you look at some of the more glaring examples of the phenomenon, our splits clearly show a ridiculous disparity between home and road blocked shots. The same goes for assists, where home-team scorekeepers are known to dole out assists to "their" point guards far more liberally than they do for opposing PGs. Using our player splits pages, these trends are easy to find and ready to be dropped on somebody's argument like an atom bomb.
So check out our splits, and don't hesitate to drop us a line if you have any questions, comments, or suggestions about how we can make them better.
September 10th, 2009 at 5:07 pm
Another nice thing about player splits is to see what months a player excelled in and when he took it slow or if the team circumstances changed in some way. For instance if you look at Jordan's 1989 campaign, you can see pretty clearly the period where Doug Collins swung him to the PG position.
September 11th, 2009 at 3:40 pm
I love using player splits for looking at up and coming players who maybe only started for a third or half of the season. Ramon sessions is a good example. As Jason said, checking out month by month averages is useful as well. As a Knicks fan, looking at Duhon's splits clearly showed how he broke down the last 2 months of the season due to playing too many minutes.
September 16th, 2009 at 12:12 pm
"...the research potential of these splits has come into play with the recent allegations of scorekeeper fraud in the awarding of blocks and assists.."
The linked discussion at this point in the post does not allege any "scorekeeper fraud".
Is there some perceived need for 'sensationalism' up in here?
September 16th, 2009 at 12:18 pm
There, the link has been changed to this post. The guy basically asserts that scorekeeping fraud is routine, at best condoned, and at worst outright encouraged. I don't think that's being overly sensational -- in essence, he claims that deliberate book-cooking is a way of life for scorekeepers.
September 16th, 2009 at 2:12 pm
OK, I do not wish to detract from the attention we should be giving to the Player Splits pages. That is, after all, where this all started, after I reported some of these home/away anomalies.
But the APBRMetrics poster known as "IrishHand", who kept stats for the Grizzlies in their first 3 seasons, wasn't 'defrauding' anyone, in the traditional sense. Giving a player extra assists, or blocks, or a cute nickname, doesn't overtly take anything from anyone else.
I'll admit to using terms such as 'bogus' to describe extra assists and blocks, which in some cases have probably been sufficient to give league Ast/G and Blk/G 'titles' to the wrong guys. And so, perhaps a case could be made that New Orleans' scorekeepers have 'defrauded' Steve Nash of another couple of Assist titles.
However, these are just arguable possibilities, not allegations of outright fraud. We just try to interpret the numbers. There's no real chance of proof, only of probability.
September 16th, 2009 at 3:26 pm
I don't want to belabor the point, either, but "fraud" is defined as: "deceit, trickery, sharp practice, or breach of confidence, perpetrated for profit or to gain some unfair or dishonest advantage."
They either knowingly recorded assists and blocks that did not happen or deliberately assigned them to the wrong player, in order to increase that player's profile and in the process create more publicity & make more money for the team. They breached the confidence that ostensibly the league (and certainly the fans) placed in them to record the events on the court as faithfully as possible. If that isn't fraud, I don't know what is. Just because it happened regarding something as silly in the long run as an assist or a block here or there doesn't mean we shouldn't call a spade a spade.
September 17th, 2009 at 8:32 am
The scorekeeper is the employee of the team, just as the GM and the coach are. The GM feels he's doing his job to say publicly something like, "Antoine is the perfect complement to our offense", and the coach may play Antoine with him as the offensive focus, because he "has confidence in his ability". This may all be contrived to drive up Antoine's market value.
The lowly scorekeeper would be undermining the efforts of his superiors, if he doesn't take the not-so-subtle hint to make sure Antoine gets credit for rebounds, assists, whatevers, to boost his marketability. There may well be nothing in his job description specifically "to record the events on the court as faithfully as possible." It's a team effort.
If you are the GM who takes Antoine off the payroll of his previous team, can you allege 'fraud' regarding the statements and statistics that accompanied him? Or should you have been a better researcher?
Consider The House that Ruth Built. Babe Ruth was left-handed and drove most balls to right field. Yankee Stadium was built for him (and lefty Lou Gehrig), with a short right field and a ridiculously long left field wall. Are their homerun records 'fraudulent'? I think not. Can a 'home field bias' be estimated to standardize their homerun records? I should think so.
Almost every NBA team wins more games at home than away, yet there's no fraud involved. Refs call more fouls on the visitors, yet they aren't normally investigated for it. In general, more people are more happy this way.
We can make adjustments to statistics, based on homecourt bias. They may never become 'official' stats, but they certainly may improve our judgments of players. If these 'adjusted' stats receive enough attention, then scorekeeping 'liberties' might get more scrutiny.