20 Second Timeout is the place to find the best analysis and commentary about the NBA.

Friday, November 14, 2014

LeBron James' Rescinded Phony Triple Double is a Symptom of a Larger Problem

Intelligent NBA observers know that there are serious problems with the accuracy/meaningfulness of box score numbers, which by definition means that the so-called "advanced basketball statistics" that are cherished by "stat gurus" are also deeply flawed. Box score numbers are not only devoid of context--points scored by a bench player during garbage time of a blowout game count the same as points scored in the fourth quarter of a close game that will decide which team wins a division title--but they often are just flat out wrong. Faulty scorekeeping is a serious NBA problem; when I have charted Chris Paul's assists I have consistently found that he is credited with more assists than he deserves--and it is perhaps most telling that I have never found an instance when a player was not credited with an assist that should have been recorded. The scorekeeping errors only happen in one direction (i.e., padding totals as opposed to depriving players of assists that they rightfully earned) and I am not convinced that this is primarily "home cooking," though that may play a part. NBA scorekeeping is either inherently sloppy or else there is a bias toward artificially inflating the numbers of certain players, perhaps with the intention of comparing today's stars favorably with stars from previous eras by making it seem as if today's stars are setting records that they are not really setting.

The latest publicly acknowledged example of NBA scorekeeping gone bad took place on Monday, when LeBron James supposedly posted 32 points, 12 rebounds and 10 assists while leading his Cleveland Cavaliers to a 118-111 victory over the New Orleans Pelicans. Upon further review, the NBA removed one rebound and one assist from James' totals. Both scorekeeping errors occurred during one continuous sequence: the incorrectly tallied rebound happened when a missed shot bounced off of James' hands into the hands of his teammate Tristan Thompson, who controlled the ball and should have been credited with the rebound. Thompson then made an outlet pass that resulted--after many dribbles--in a Kyrie Irving fastbreak layup. No assist should have been awarded on that play but instead the scorekeeper gave James an assist, which makes about as much sense as giving Mark Price an assist on the play.

One can dismiss this as "home cooking" or say that Cleveland's official scorekeeper just had a bad night but the reality is that egregious errors regarding LeBron James' box score numbers do not only happen in Cleveland. This is the second phony LeBron James triple double that the NBA has corrected, with the first one happening in Madison Square Garden on February 4, 2009. Scorekeepers would not be making such obvious mistakes unless this kind of box score padding is an accepted practice. The NBA claims that it regularly checks game film to make sure that the box score statistics are accurate but--in light of the numerous uncorrected box score mistakes that I have found in just a handful of Chris Paul's games--it seems more likely that the only reason LeBron James' phony triple doubles were rescinded is that observers outside of the NBA publicly called attention to these outlandish mistakes. Meanwhile, routine errors that are not publicized become part of the sport's historical and statistical record.

Labels: , , , , ,

posted by David Friedman @ 6:13 PM

12 comments

12 Comments:

At Saturday, November 15, 2014 1:21:00 PM, Anonymous Anonymous said...

I don't think it's homecooking either. It's interesting that the nba says they do this every game, which they obviously don't. But, at least they'll do it with triple double games. Some plays are hard to determine who gets the deserved stats, though. They give rebounds to guys who tip the ball often, probably shouldn't

However, how do you know they accurately kept stats 30-40 years ago compared to today? Our technology is so much better today, with HD replays available every play, and they still can't get it exactly correct. I seriously doubt stat keeping in the past was any better, if not probably worse. The biggest problem I see is how statkeepers record assists. Giving James an assist on the overturned play was ridiculous, but giving Thompson one there is done all the time. I wonder if that's more an interpretation of the rules. And if in doubt, maybe the nba tells them to give the assist. Other plays like steals/rebounds primarily that are sometimes hard to determine who gets it will always be hard to determine depending on the play and regardless of how many replays you see. There are some awful stats kept, but I seriously doubt there wasn't as many in the past. Maybe 1-2% of stats are wrong at most, but most of those are up for interpretation. The nba should review every game, and make adjustments regardless if it's a triple double or not, that's the bottom line, and shouldn't be too hard to do.

Other stats that might be kept correctly, but doesn't seem right, are some turnover plays. If a guy makes a good pass, but the recipient doesn't catch it cleanly resulting in a turnover, the passer gets the turnover, not the recipient. Maybe that's how the stat should be kept, but when the stat gurus look at it, they don't take this into account.

Also, if a player is driving and never gets into his shooting motion or is barely starting to attempt to shoot, and the defender strips the ball, it usually counts as a missed FG and a block. If the ball goes out of bounds, the offense retains it, should be no stats. If the defense gets it, then it's a steal/turnover. Maybe this is what the rule says, but it doesn't seem right.

 
At Saturday, November 15, 2014 6:26:00 PM, Anonymous Anonymous said...

I remember someone pointing this out a while ago: https://www.youtube.com/watch?v=UdGhN_I5yH0
LeBron is somehow credited with a defensive rebound and an assist to Mo Williams on this play. The play-by-play on Basketball Reference shows it too. http://www.basketball-reference.com/boxscores/pbp/201004250CHI.html

 
At Saturday, November 15, 2014 6:59:00 PM, Blogger David Friedman said...

Anonymous:

Assists are awarded on a higher percentage of made field goals now than they were in the past even though the modern game is more isolation-based, so I think that it is reasonable to assume that assists are currently overcounted.

Based on my own observations and my conversations with people who are old enough to remember previous basketball eras, I believe that modern scorekeeping is less accurate, at least in terms of assists being awarded more liberally. Steals and blocked shots did not become official stats until the early 1970s and I am not sure if the standards for those categories have changed over time. I am a bit suspicious of Chris Paul's record for most consecutive games with at least one steal but I did not chart every one of those games to find out if something fishy happened with the scorekeeping.

You are correct that in general the passer and not the recipient is charged with a turnover and I agree with you that there are many instances when this is not fair.

The stripped ball scenario is inherently subjective regarding whether the play is a steal/turnover or a block/missed shot, assuming that there is a change of possession. If the offense retains possession, then neither a steal nor turnover can be recorded and the play is either a block/missed FG or just a deflection (which is not an official stat, though some teams record that data unofficially to keep track of "active hands" defensively).

 
At Saturday, November 15, 2014 7:12:00 PM, Blogger David Friedman said...

Anonymous:

I think that the Mo Williams play is a lot more common than many people realize. LeBron James is a great player and he deserves every MVP that he has won but I don't trust the accuracy of his statistics nor do I trust the accuracy of the "advanced basketball statistics" that "stat gurus" create out of thin air based on these faulty box score numbers.

 
At Sunday, November 16, 2014 5:52:00 PM, Blogger Nick said...

Assists are kind of an especially problematic case in terms of old vs. new, as the definition has changed; it used to only be an assist if the player didn't dribble after, while now it's much more liberal; so long as the player scores shortly after the pass without more than one or two moves (already silly) to get free, it's usually an assist.

However, I agree in general with anonymous' point that old stats are probably on balance less reliable than new ones. Fewer people employed to keep them and doing so without the level of either technology or, presumably, oversight they have today.

Out of curiosity, how much faith do you put in John Stockton's assist totals? On the one hand I've seen him accused of home cooking multiple times, but on the other hand many- if not the most- of his assists were fairly basic pick-and-roll drop offs to Malone, which is certainly an assist.

Personally, I've been largely ignoring assists as a meaningful statistic since I noticed Rondo- a slightly better than average passer surrounded by mediocre teammates- routinely near the top of the league. Assists already were flawed by not accounting for created free throws (Nash is likely the all-time leader in these) or the hard pass that leads to the easy past that leads to the assist (Nash, Larry Bird, Shaq, and Hakeem are four examples of guys who had several of these per game almost every game). It's a shame we don't have a better statistical passing metric.

Heck, a case could be made that either Pistol Pete or Larry Bird is the greatest passer ever, but neither had especially gaudy assist numbers.

Lastly, yes, turnovers should be attributed to whomever's fault they were whenever applicable. Suffering through this early Suns season watching Dragic get dinged with a turnover when Bledsoe passes him the ball with .5 left on the shot clock, or when Marcus Morris is LOOKING AWAY FROM THE BALL WHILE CUTTING TO THE RIM has been killing me.

Kobe's been a victim of a lot of those over the years as well, although perhaps only about half as many as Kobe himself probably thinks he has, based on his response when one of his high degree of difficulty rocket-passes isn't received.

 
At Monday, November 17, 2014 8:01:00 AM, Blogger David Friedman said...

Nick F:

The official definition of an assist has not been changed by the NBA but the quality and consistency of NBA scorekeeping of assists has markedly declined. An assist has always only supposed to be a pass that leads directly to a made field goal and if the shooter did more than half of the work then the passer is not supposed to receive credit. The number of dribbles is not necessarily the critical factor; an outlet pass to a player for an UNCONTESTED (note emphasis) fast break layup can be correctly scored as an assist even if the recipient takes several dribbles provided that the resulting shot is indeed UNCONTESTED--but an outlet pass to a player who must evade good defense by faking or contorting should not be scored as an assist even if the recipient did not take any dribbles.

In the old days, there were fewer stats to track (no steals, blocks, turnovers, offensive/defensive rebounds--only total rebs were tracked--three point FG) and there is good reason to believe that overall the stats were more accurate, even if there was some occasional home cooking.

Assists have always been somewhat subjective but now they are often just ludicrous. Another problem with assists--which has less to do with how they are scored than just the inherent flimsiness of the stat--is that some players only make passes that are very likely to lead to an official assist, even if this results in them not making the right basketball play. Stephon Marbury was a horrible, selfish teammate whose teams did worse when he arrived on the team and better as soon as he left (i.e., the anti-Jason Kidd) but he posted gaudy assist numbers because he would only pass the ball if his teammate had an easy shot. That may sound good superficially but what it meant in practice was he would hog the ball otherwise, reluctant to make a pass that would lead to someone else's assist. This led to a lot of overdribbling and taking bad shots.

I think that Stockton's totals were padded, though I think at that time the problem was more related to home cooking as opposed to general laxity (I don't have the time to check out his home/road splits, if indeed they have been posted for his entire career).

Kobe has been the victim of many unjustifiably scored turnovers and also many of what I term "hand grenades" (FG attempts that would have "exploded" in his hands if he had not shot the ball just before the shot clock expired after receiving the ball late in a possession). Kobe has always been more concerned with winning than with preserving his stats/"efficiency," which is not necessarily the case with many other players (LeBron wants no part of "hand grenades" that will diminish his FG%).

 
At Monday, November 17, 2014 11:35:00 AM, Anonymous Anonymous said...

I've heard that they didn't allow dribbles after a pass at all for assists in the past, regardless if it should've been an assist or not, but not sure if that's true or not and if so, for how long they recorded it like that. But, it seemed like there was much less dribbling in the halfcourt before, so I'd still think the # of assists given out wasn't much less, if any.

It does seem like that scorekeepers are more liberal with assists now. I remember Rondo got an assist once by passing to Ray Allen, who then passed to Perkins, who then scored. Actually, Allen shouldn't have gotten an assist either, if I'm remembering correctly. However, I still seriously doubt statkeeping 30-40 years ago was setup to be any better than they do today. They're recording things like dunks, FGA from certain distances, etc. So much more going on today. Maybe that's one reason why certain stats could be off a little today, because there's so much more to be recorded. I don't see it as that big of a problem, though.

But, Rondo is a great passer and facilitator. He's a great athlete, not looking to score. But, he's like Nash was with the Suns. They each have the ball 90% of the possession, and then either shoot or most likely pass to the guy who scores.

Marbury isn't as good as his stats indicate, but he was a very good player, who made 2 AS and 3 3rd team all-nba. Iverson also was similar to Marbury, in that he'd usually only pass if it'd lead directly to an assist. Kobe would have so many more assists if he didn't play in the triangle. He'd have so many hockey assists being double teamed, passing to Pau, and then Pau gets the assist. And Goldsberry came up with the 'kobe assist', offensive rebounding putbacks. There's a lot more going on than just looking at assist totals.

How does Dragic get a TO when he gets the ball with 0.5 seconds on shot clock? Shot-clock violation TO? What I've noticed in the box score, is that the team gets a TO for shot-clock violations, not any one individual.

I often hear people say it's every team's go-to guy's responsible to take that last shot or end of shot clock shot. That might be true in principle. But in reality, this doesn't happen the same for each go-to guy. It's obvious Kobe is stuck with these shots a lot more and has no qualms about shooting them, whereas James/Durant have actually admitted to not shooting them just because it'd hurt their FG pct., which is sad.

 
At Monday, November 17, 2014 3:15:00 PM, Blogger Nick said...

I must be mistaken, then. I could have sworn I'd read many times that in Oscar's era, assists were only attributed if the player did not dribble after receiving the pass.

 
At Monday, November 17, 2014 7:35:00 PM, Blogger David Friedman said...

Nick:

Oscar has certainly said that in many interviews and he may be right but my understanding is that if the pass led directly to an unimpeded shot then an assist would/could be awarded. Regardless, there is no doubt that the interpretations have been liberalized in recent years even if the black letter rules are the same. I don't have the time right now to hunt down the official NBA scorekeeping rules from the 1960s, if I even have them in my possession. I know that for at least the past 10 years or so the official rules have permitted more than one dribble but only if the shot is uncontested. The guiding principle has always been that if the shooter did more than half of the work then the passer should not receive credit.

Regarding Bird, he was a great passer but I recall him getting some assists after McHale went through his low post repertoire of fakes. I thought that assists started to be padded in the mid to late 80s but things have just gone nuts in the past decade or so.

Rick Barry told me (and he has told others as well) that the only number he trusts is FT%. Everything else can be manipulated or padded either by a player choosing not to shoot "hand grenades" or a friendly stat keeper, etc.

 
At Monday, December 01, 2014 12:22:00 PM, Anonymous Matt DiFilippo said...

The one that always seemed fishy to me was Elmore Smith's block record. In a span of five games, during the first month blocks were recorded as an official statistic, Smith had games of 17, 14, and 14 blocks -- and no one else reached 14 blocks in a game until 1985!
I always suspected it was just someone who didn't know how to award blocks or wanted to pad the stats.

 
At Tuesday, December 02, 2014 1:33:00 AM, Blogger David Friedman said...

Matt:

Smith was an excellent shotblocker but you are right that those numbers look odd/inflated.

 
At Wednesday, December 03, 2014 9:51:00 PM, Anonymous Nikola said...

Matt/David:

A season before, in 1972-73, when blocked shots were recorded unofficially, Smith was credited with 300 rejections for a 3.95 bpg average, compared to his 4.85 average in 1973-74 when he played one less minute per game, the first NBA season blocks were recorded as an official statistic.

David:

The NBA rescinded another LeBron's triple-double back in 2006.

http://scores.espn.go.com/nba/recap?gameId=260401005

Scorekeeping in the 1950s:

Mar 11, 1958 Daily Boston Globe:

"How do you feel about the system throughout the league of scoring assists, if there is one set system?"

Cousy: "That's the trouble. I think there are eight systems. I think as I tell the boys in my Summer camp when they're scoring a game. Give an assist always to the last player to pass the ball to the man that scores, no matter how that final player scores."

For Cousy's and the Boston fans' benefit Dinny Whitmarsh, official scorer at the Garden, who, along with his WHDH partner, Sid Pike and Norman Altman, handle the "assists", describes just what is an assist.

Whitmarsh: "An assist is strictly a judgment call. You can't go by the book. A few years ago, Haskell Cohen, publicist of the N.B.A. told us scorers that we were giving out too many assists and suggested changes.

"Now, we do not give an assist if the player who last receives the ball, MAKES THE BASKET THROUGH HIS OWN EFFORTS AND NOT BECAUSE OF THE PASS. In other words, if Cousy passes to Bill Sharman, and Bill has to dribble a couple of times and go behind a "pick" and scores...no assist. No assist is given if the receiver fails in his initial goal try, but taps in a rebound. "In the case of Tommy Heinsohn, it's a bit different. Heinsohn may take a Cousy pass and then he usually makes a couple of moves or fakes, gets position for his hook and scores unaided...we give Cousy an assist."

 

Post a Comment

<< Home