the advanced stats narrative

One of the goals of my last abandoned blog project was to challenge the so-called advanced stats movement in sports. The self-identified “geeks” of this movement tend to imagine a noble battle between barbarians armed with silly sentiments and dumb box score stats (bla bla who I generally appreciate loves using the terms smart and dumb) and themselves – the noble champions of smartness who embrace the methods of civilized scientific (analytic) thought. My problems have never had anything to do with advanced stats themselves. If you are going to use stats, why not spend time trying to find/produce the best stats possible? That is a no brainer. My real issue is the hubris, self-image, and framing sometimes found among so-called stat geeks.

A paper by Gibson and Poposki  [1] found that individuals who attempt to project a certain quality in themselves tend to underestimate this quality in others. Individuals who want to seem smart tend to underestimate the intelligence of others. In this sense there is a vast difference between the attempt to be smart about using data (in itself) and the attempt to appear smart about using data. In the first case, there is no tendency to diminish the intelligence of contributions of others. If the goal is being as “smart” as possible one has an incentives to seek out the best ideas in others. Ok, this view has critical limitations but it also has some insights. If the goal is to seem smart, the incentive is to paint everyone as a complete moron which would often require rejecting interesting and useful ideas. Of course, in practice the distinction between trying to be and trying to appear is difficult to distinguish, but my main concern is that the narrative stat-geeks provide tends to produce at least some of the latter. Reducing the analysis of sports to smart objective people who go to Harvard/MIT versus emotional village idiots makes us all a bit more idiotic.

This noble story, casting the geek as hero, is also counterproductive. Henry Abbot of TrueHoop often laments resistance to advanced stats without seeming to realize its packaging motivates rejection. (I like TrueHoop. I think it is the best part of ESPN. I think its influence has also made the non-TrueHoop aspects of ESPN’s NBA coverage better as well. But…)”Dear sports fan who uses dumb box score data, some nerdy white guys from top colleges have proved just how stoopid you are! Applaud their genius and abandon your stoopid ways.” Now, I’m not going to say this offer is completely unproductive. Of course it will attract some people who want to be part of the new cool crowd, but it also creates lots of resentment. Let me also say it creates some eyeball-rolling. In 2011, saying things like, “oh I’m such a geek” or “step back while I get wonkish,” are similar to statements like, “oh, I’m so attractive and popular.” I’m not going to go into a fit about geek/nerd becoming mainstream, but lets just accept it is mainstream, stop boasting behind the guise of self-deprecation and leave it at that.

Now, the stat geek may respond that traditional views of sports are so backwards that they must be offensive and confrontational, but I think at least some of this resentment is completely unnecessary. Lots of advanced stuff is relatively in line with “unadvanced” stuff. Abbot recently made the point that D. Rose’s new success at getting to the free throw line is a win for the stat geeks. But isn’t it an example that the war between dumb and smart stats is overplayed? Pre-Hollinger did people think free throws were unimportant? Of course not. This isn’t to say Hollinger doesn’t contribute anything, but what he contributes is a refinement and “advancement” of old-school thinking – not its refutation. We now have a better grasp of how many free throw attempts a player gets per minute, possession, shot, or shot at the rim. We also can do a better job estimating the (expected) value of different shots, but we’ve known free throws are good to get for some time now.

Aside from being bad diplomacy, the stat geek narrative inflates (or perhaps misspecifies) the importance of advanced statistics. Again, Abbot has used to the term smart and dumb stats. I think this is a bad distinction. Whether a statistic is dumb or not, depends on whether it is useful or not. Usefulness depends on what we are using it for. We can not categorically dismiss a statistic as dumb without assuming there is only one use of data. The stat geek dismissal of per game statistics, as one example, is based on the completely unnecessary imposition of the job of a general manager maximizing win probability as the only way to view, analyze, or appreciate sports.

I’ve used the qualifier unnecessary a few times now but it is unavoidable. It is completely possible to say that it is dumb for general managers to only evaluate talent on ppg without declaring war on everyone who is interested in league leaders in ppg. I haven’t done the polling on this (but no one has as far as I know) but it seems that pre-advanced stats most people thought highly of league leaders in ppg, but also recognized the statistic had serious limitations. Crude unadvanced stats fans are not morons. They do not think ppg is the be- all end-all in terms of best players, but are nonetheless interested in the data itself. If I want to predict which player will help my teams chances of winning a game the most, ppg might be dumb. If I’m interested in who scored the most points per game during a season, ppg is actually kinda useful. Of course, here is where the stat geek rages – “but why care about such crude data?!” Because sports is not just about the analysis of likelihood and probabilities, it is also the ex-post appreciation of the crude what actually happened.

It is perfectly legitimate to discount low probability events (just don’t over discount) in making forecasts. It is also perfectly legitimate to appreciate low probability events when they happen.

It is also interesting (and troubling) to note who advanced stats proponents see as their other when they identify themselves. From TrueHoop :

Animals putting the ball in the hoop. Monkeys, dogs, elephants, birds … they all get the concept. And no joke: We wonder sometimes why everybody values offense more than they should. Why defense is left for coaches and stat geeks to worry about. I suspect it’s because the basic concept of “put this ball in that hoop” has a profound and basic appeal that extends even beyond our species. Steering the ball-handler to the help defense … that’s more of a refined taste.

Lets just ignore the millions of people who are neither coaches nor stat geeks who nonetheless worry about defense for now (is Rodman a coach or stat geek?). Monkeys, dogs, elephants, bird, and some group of people called “everybody” operate on an analytical level that is only able to understand offense (ball go thru hoop! yay!). Defense, well that is left for coaches and stat geeks! Luckily for us non-human animals and (presumably unwashed) everybodies there are those with a more refined taste! Clearly, we must submit to these coaches and Harvard/MIT people! Help us upper-class white people! (Again, we can see how polarizing this framing is. It will attract somebodies who don’t want to be just an everybody, but it is off-putting to many others.)

Now, why the hell did I have to mention race?! That is a really good question. I, myself, am asking it to myself. Or, if not to myself primarily, then to chief popularizer of advanced stats (Moneyball) himself, Michael Lewis. Who cares that Shane Battier has a white mother? Lewis wrote a very popular article on Battier in the NYTimes that devoted over three paragraphs to his racial identity. From a strictly sports-analytical perspective what is this about? Isn’t the whole point of the article to champion Battier as an excellent basketball player when looked at from a purely objective scientific point of view. When we abstract from everything that doesn’t really matter in deciding a basketball game (“dumb” stats, attitude, dunks, flashiness, culture, etc), Battier is a great player. Unfortunately, stupid fans (non-human animals and everybody) appreciate style over substance. Fortunately, the advanced stats movement will save us, restoring substance to its rightful place as all that really matters. Isn’t bring race into this discussion at odds with Lewis’s main point? Battier and Lewis have the following exchange on his reputation:

Chris Webber won three state championships, the Mr. Basketball Award and the Naismith Award. I won three state championships, Mr. Basketball and the Naismith Awards. All the things they said I wasn’t able to do, when I was in the eighth grade.”

“Who’s they?” I asked.

“Pretty much everyone,” he said.

“White people?”

“No,” he said. “The street.”

Everybody, meet everyone. Presumably, after a few paragraphs on Battier’s race we are supposed to read “the street” as black. This is of course interesting, and I do not doubt that race and identity are super important in the experience of playing basketball. I know that  when/where I grew up, whenever I played with black players I’d usually get a white nickname. This was never fun. For a kid who looked up to Jordan, Webber, Olajuwan, Dominique, etc. you don’t like kids shouting “Bill Purdue” whenever you touch the ball. It was, however, preferable to the identity-based trash talk of other white kids (“hey homo! we are going to kick your ass!”). It was a meritocracy. I can only speak from my experiences here, but you could start out a Purdue and graduate to a Kukoc. For me, Kukoc was always the pinnacle. Kukoc was the best you could do. Of course, I’d prefer Pippen or Jordan but that was essentially unattainable. Bird and McHale always felt like underhanded compliments. Anyhow, I know, racial identity matters. I’m sure Battier has awesome things to say about it, but again, what role does it play in the article? I think the role here is clear. It racializes the distinction between an advanced, analytical, team-based view of sports and the unadvanced, pre-analytical, individual-based view. White and black. Substance and style. The streets may not have respected him but Daryl Morey does! Maybe I’m being unfair to Lewis but what else is this passage doing in his article? Indeed, he explicitly racializes this opposition in the process of attempting to “defend” black sports:

For instance, is it a coincidence that many of the things a player does in white basketball to prove his character — take a charge, scramble for a loose ball — are more pleasantly done on a polished wooden floor than they are on inner-city asphalt?

It is hard to have thought about race much without getting nervous here. It is not that blacks are inferior when it comes to playing white basketball, it is just their geography! If it is not the tropical weather it is the hard ashphalt! One really wants to challenge Lewis on how much “black basketball” he has played/watched. One respects Lewis too much to think that he would assert black’s don’t associate character with good defense and scrambling for the ball based on watching so-called streetball on ESPN, but really, what is he basing this on? Now, I agree this might be what white people think about white basketball, but isn’t advanced stats supposed to be about abstracting from the fuzziness of our biases and prejudices in order to see the objective reality of sports?

Now, let me back up a huge step. I’m not saying Lewis or other proponents of advanced stats are racist. Could some of them be? Sure, but thats true for any group of people. And, let me be super clear that some proponents, such as Abbot who I have quoted somewhat critically here, and other TrueHoop contributors do an admirable job on race and NBA players. The point is that when it comes to attacking fuzzy unadvanced views of basketball the racialized conservative old white guy worrying about the decline of civilization/basketball is hardly ever targetted. Indeed, the usual targets – selfishness, bad shot selection, style over substance streetball players, per game stat chasers, etc. – also happen to be the targets of the conservative old white guy contingent. In most cases I doubt it is conscious, but in practice it is as if proponents of advanced stats attempt to appeal to this contingent in how they frame the debate over sports. In Lewis’s example it seems conscious. Not that Lewis is thinking , “maybe I can make my piece more popular by appealing to a racialized views of basketball.” I’m not saying this at all, but I do think it is very plausible that in attempting to tell his story about something new and different (advanced stats) he relies upon some very old and familiar tropes and frames. In doing so he undermines one supposedly-existing view of basketball (the view he implies is “black”) but reaffrims another (his white basketball).  Turns out some smart white kids from Harvard/MIT proved your white n-word using grandfather was right about today’s thuggish players!

Let me just keep repeating that I’m not saying all advanced stats people would be comfortable with that. I don’t think they want to prove such a granddad right. But I really wonder why the matter is often framed this way. Why were efficiency-based ratings used to attack Iverson’s reputation, but not to defend Zach Randolph? Maybe it is just an issue about perceived threats. I do not think that there are many people who slavishly follow per game scoring stats as if they were the ultimate metric and don’t give a shit about defense. To the extent they exist they are like 12 years old and will probably come to appreciate advanced stats when they get to high school. Are per game scoring stats used as a heuristic in the absence of better information? Yes. Does this lead to biases and sometimes bad evaluations? Sure. But the simple solution to this is providing better stats that make the heuristic unnecessary. At the same time, the conservative “sports is a morality play pitting men of character and substance versus stylistic thugs” seems to be an incredibly popular view of sports. And not just with tweens, but with fully grown adults – fully grown adults who watch almost no basketball but speak with certainty about players based on stereotypes, length of shorts, or the droopiness of a players eyes. If I wanted to attack fuzziness, this is where I would target.

Could one even begin documenting the vast accumulation of complete bs created in response to Lebron’s signing with Miami. Alpha and Beta dogs? Winners? Having what it takes? Psuedo-pop-psychology based on Lebron’s tattoos? What would our great heroes of the 80s do? This is fuzziness. For every 12 year old who said, “this is awesome, my three fav superstars with high per game scoring stats on the same team!” there were 100,000 grown men having a fit about Lebron not being a “winner.” These grown men have a right to their morality play, but again, if we want to confront fuzziness with the sharp edge of analysis why not target here?

For a long time the conventional wisdom was that Zach Randolph was a bad player. If you asked the average sports fan why you’d get a bunch of fuzzy morality crap. He is not a good character guy. Well what is character and wtf does it have to do with basketball anyhow? How do we evaluate the character of people we only know through the media? He is not a team player. What are you basing that on? I saw him take a stupid shot on a youtube video linked from the Huffington Post. Well how representative was that stupid shot? He is just not a winner, his teams lose, he doesn’t have what it takes! Well now we are explaining wins by who wins, and it just doesn’t get any sloppier and useless than that. Plus, maybe he has just been on terrible teams, right? Well, I’ve heard he is a bad guy from his own team – even they blame him. So you trust the anonymous sources coming from a terrible basketball team? On what grounds are we to trust these sources as accurate? Don’t contributors to failing teams have an incentive to scapegoat others? Would we expect a coach to whisper to a reporter -“listen, it is not Zach’s fault, it is mine!”?

If I have to choose between ppg and rumors about a player’s moral character spread by a team in a downward spiral, I choose the former every single time. So the question is why we continue to let the morality crap dominate representations of sports and chase the invisible (or underage) demon of dumb per game stat everybodies. Again for every 1 twelve year old who thinks Adrian Dantley is better Magic Johnson (ppg dude!) there are a million sports fans who saying things like, “But so and so doesn’t have a ring!” It is not just that the have-a-ring argument is less analytically sound than the ppg one, although it is. It is that the have-a-ring thing is very dominant (among even adults).

In fairness, there are examples of advanced stats people who do target the ring thing. Hating Tracy McGrady is one of the cheapest and fuzziest way to appear smart in the traditional conservative sports as morality play fashion. Why doesn’t he make his teammates better?! He doesn’t care! Look at the droopiness of his eyes! Larry Bird would never have droopy eyes! Here are a few example defenses of McGrady, in part based on advanced stats. First, the McGrady Manifesto is one of my few bookmarked blog posts on the NBA. Second, see this basketball-reference blog post on McGrady’s peak, which quotes Hollinger’s 2004-2005 Basketball Prospectus:

I wrote a year ago that McGrady was a better player than Kobe Bryant and would have thought that [the 2002-03 season] proved it once and for all. That’s why I was shocked and dismayed when the MVP voting came out and Bryant ranked third while McGrady was fourth.

Look, I don’t mean to keep knocking Kobe, but the difference between style and substance here is too big to ignore. Bryant is immensely well-known, appears on TV all the time, and tends to do spectacular mid-air theatrics that you’ll talk about for the next three weeks. He’s a great player, one of the five best in the league. But he still can’t carry McGrady’s jock. Put the two side-by-side, and everything Bryant does, McGrady does as well or a little better.

The big subtle difference between the two that most talking heads can’t appreciate is that McGrady never turns the ball over. His Turnover Ratio was the lowest among small forwards and a far cry from Bryant’s. The fact that he also had the league’s top Usage Rate is doubly impressive. Basically, he created a zillion shots without any miscues, and that alone put Orlando into the top half of the league’s offenses…

McGrady does some things with flair — his penchant for throwing the ball off the backboard to himself and dunking, for instance — but a lot of his skill lies in his ability to jump straight up over a defender and nail a jump shot. It’s impressive, but it’s not as exciting as Kobe, and that creates a mistaken impression about who is really the better player. It doesn’t help that McGrady is the league’s most sleepy-eyed player since Sam Perkins…

Looking at the big picture, Tim Duncan is great, and Shaq is dominant, but McGrady was the best player in the league [in 2002-03]. McGrady’s only 24 and is getting better every year. It’s really vexing to see that the media have billed Kobe Bryant as the closest thing to the next Jordan for the past five years, and the whole time they had the wrong guy.

The basketball-reference blog also had a very nice piece challenging the view that the Lakers won (beat the Celtics) because Kobe is a winner who has what it takes and the Cavaliers didn’t because Lebron is not a winner and doesn’t have what it takes. One could disagree with the ultimate conclusions perhaps, but it is an example of using advanced stats to target what I think is the real “threat.”

But it doesn’t always work out that way. From the coverage of the recent Dorkapalooza, McGrady only came up as a talented but lazy basketball who could have been a hall of famer if he only had the work ethic of Brian Scalabrine! The analytical-empirical basis of this was Malcolm Gladwell’s people need to practice something 10,000 hours to be great story. How rigorous! Now, lets talk about McGrady the way every old-school conservative that doesn’t really follow the NBA would! Despite the fact that when he was not hurt it is hard to imagine him being much better, and when he was hurt he was hurt (nothing more bizarre than reading middle aged men blame McGrady’s back on his work ethic – do sports reporters not develop back problems? who doubts how terrible a bad back is?), lets just have fun with this really tired stereotype! Look at McGrady’s eyes! He doesn’t have what it takes to get out of the first round! Tim Duncan wouldn’t slack off! Bla bla bla. So while it would have been very unlikely for Dorkapalooza to slip into a discussion of how awesome McGrady’s dunk on Shawn Bradley was (a dunk is only worth 2 points – get analytical!) it is not surprising to find people getting misty-eyed and emotional about what Michael Lewis might uncritically call white basketball.


[1] Gibson B and Poposki EM (2010). How the adoption of impression management goals alters impression formation. Personality and social psychology bulletin, 36 (11), 1543-54

This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s