Juwan A Blog? #1: Wages of Wins

Posted on Thu 27 October 2011 in Juwan a Blog? by Alex Dewey

As a recurring feature, Alex will be reviewing and analyzing various blogs and hoops sites. No number ratings or anything silly like that, just a good overview of the sites at hand with their strengths, weaknesses, etc. To see an index of previously reviewed sites, click here.

The way many fans tell it, the field of sports statistics is a conspiracy against their favorite player (*cough* Kobe). For others, sports stats is a conspiracy against the fan experience. For many beyond that, sports stats is a useful and instructive field still in its infancy that often makes claims far above its pay grade and level of sophistication. For a fourth group, sports stats is absolutely perfect with no flaws. Now, most people are in the third camp, largely because of the way I worded that paragraph to make it seem most reasonable. Obviously you can find good examples of the first two groups on any sports comment section or any basketball forum. Of course, no one is really in the fourth group this brings us to Wages of Wins, by process of elimination.

You see, Wages of Wins is perhaps the only group of people in basketball that think that numbers speak for themselves, and that it's up to us intelligent people to give the numbers a voice against the unwashed masses. They might not say they believe this, but the evidence of their book, their head blog, and their overall perspective and approach combine to make me think this. If you want a (crude and somewhat unfair) analogy, you could say that Wages of Wins form a similar niche to Objectivists in American political culture. To make the analogy more precise (and hopefully less insulting), they take their own group's exclusive access to objective truth and intellectual courage as a collective and partially unstated given. Is either group filled with awful people or even stubborn people? No, not really. But the philosophy itself is stubborn, the heads of the two groups are frustratingly stubborn, and in general the movement represents an easy, half-assed way out of a hard problem. Let me explain:

There's a difference in any intellectual endeavor between positive and normative claims: both are subject to evidence, but are the result of fundamentally different perspectives. A positive claim goes something like this: "Increasing the minimum wage may increase the clearing price of wages but also will create labor shortages." Sure, it's not an easy empirical claim to think about or describe an experiment for, but, like you could design and describe an experiment from the nature of the claim. You know? You can use economic statistics, look at legal complications, take surveys, use research in psychology, even Platonic reasoning about an established model, and so on, to look at the historical effects of the minimum wage. You can make a good argument for or against a positive claim using both creative and established methodology.

Normative claims are a bit hairier. Whereas the positive claim above makes two casual links (Minimum wage up to wages up to supply of labor down), a normative claim makes one of its causal links a "better" or "more moral" universe. "Increasing the minimum wage will lead to a better world" Even in principle, to address whether this claim is true requires a whole lot of shared assumptions: most people might accept the claim if increasing the minimum wage increased both wages and supply of jobs for the poorest citizens. But unless you accept what someone else's version of a better world is, you are unlikely to be persuaded by their normative claims.

Now, positive and normative claims both have their purposes: positive claims help us turn our experience into sound theories about the world. Normative claims often lead to imaginative and thoughtful interpretations of the world around us. They also both have fundamental drawbacks. You can positively describe all the psychology and economic consequences of something stark in human experience like love or slavery - but at the end of the day it takes a normative claim to affirm what is right and reject what is wrong. And dually you can make loaded and rhetorical claims about the nature of things until you're blue in the face, but if you can't turn your judgments into empirical claims, you aren't going to change anything if you're right, and you're never going to find out your error if you're wrong.

Now, The Wages of Wins - to its credit - seems to steer towards and pay far more than lip service towards positive claims. One of the group's major pet peeves is the media bluster over a hyped-up, flashy player like Allen Iverson or Kobe Bryant. The group correctly notes that the media will use toy statistics and scoring numbers without context to make scorers like Iverson seem like the single most productive player on the floor at any given night, excusing shooting inefficiency with raw scoring numbers, excusing low defensive impact with the flash of steals, and so on. The numbers - as the Wages of Wins group reasons - don't lie. And just because you saw Iverson drop 60 one night doesn't excuse the shooting inefficiency he often brings to the table. Even if Kobe Bryant is legitimately great - if his stats don't match up - it's fair to call into question his sum total of accolades. I think most reasonable people can accept this.

To this end they developed on an interesting statistic (from Dean Oliver's work) - Wins Produced. I can't summarize it perfectly without getting a bit mathematical*, but to give you a toned-down version, they plot team point differential against winning percentage to calculate the marginal (economically speaking) value of points and possessions. They use these valuations more-or-less Platonically** to put box score stats like made field goals, steals, missed field goals, turnovers, and assists onto an additive scale, so that if you add together all the things in the box score given these weights, you'll come extremely close to knowing who won the game by what - and, adding up a team's stats over the course of the season - you should know more or less where they finished in the standings and/or how good they are. Then they take this (even after months of dissing them I still say) perfectly reasonable statistic and define an individual's "productivity" by their per-minute box score stats. And then they make a perfectly reasonable adjustment for position. Perfectly reasonable. Perfectly reasonable. Perfectly reasonable is their statistic.

* Like all math majors, long, fascinating, and complex chains of reasoning appear to me in my head. This is a practical joke by math to see how badly I'll butcher these chains of reasoning when communicating them to other people.

_ ** Platonically in the sense that all assists are valued the same, all missed field goals are treated the same way, all turnovers are treated the same (and the same as missed field goals), steals are treated like anti-turnovers, steals are valued identically to rebounds, offensive rebounds are valued identically to defensive rebounds. More on this soon._

Do they have a point? Sure. When a perfectly reasonable statistic fails again and again in favor of the demonstrably worse metric scoring volume, they have successfully argued that there is a problem with NBA decision-making (at least - as they empirically show - with Rookie of the Year, MVP, All-NBA, etc. selections). You see, any perfectly reasonable statistic should be at least pretty decently correlated with merit. You know what I mean? That's not to say there can't be exceptions (perfectly good reasoning often systematically ignores exceptions), but basically, if you invent a statistic where points are good, misses are bad, steals are good, and assists are also good, and so on, you're probably not going to go too far astray as long as you keep a decent sense of proportions. If your proportions are given by team point differential, you're probably on the right track. You could easily - easily - use similar chains of reasoning to argue someone is the MVP. In fact, the popular "assist = 2 points" trope is a great example of this kind of argument back in the "Kobe vs. Nash" MVP debate in 2007. So yeah, there is a point to be made here: "Maybe Kevin Garnett isn't just doing a showy and historically great job of choking in the playoffs for the Wolves, big-market announcers. Maybe he's actually the best player in the league. Here's my argument. Maybe he is a winner. Maybe he's one of the best winners in the history of the league." That's what Wins Produced is - at first glance - all about. And it's a good thing.

It's perfectly reasonable. But then they stray from solid positive claims into (as opposed to equally solid) sketchy normative claims. They take their "perfectly reasonable" statistic and make an idol of it. Now, Richard Jefferson is a perfectly reasonable, likable, and competent player, but I shudder to think that anyone out there has made an idol of Richard Jefferson. Not only would it be an allocation problem (underutilizing perfectly fine stars to idolize instead like...I don't know, Chris Paul? Gosh.), but it would - in basketball terms - represent a fundamental misapprehension of the sport. If Richard Jefferson stops your show, it's probably not a very good show in the first place.* The problem is that instead of taking this Wins Produced statistic, putting it in their back pocket to counter a specious claim like that Allen Iverson did more to make his team win than Shaq in 2001, they argue that Shaq should be the MVP because their statistic says so. And then they argue that Tim Duncan and Kevin Garnett and Chris Paul and Jason Kidd should have had the award for the next 8 years. Okay, I'm exaggerating, but the bottom line is: If he leads in Wins Produced for a year, the people at Wages of Wins will argue that he should be the MVP. Not "this should be used as a sanity check" because if a player is terribly mediocre according to this metric they're probably not an MVP. Not "this is one of many perfectly reasonable metrics, but it gives us some substantive clues into how good Kevin Love has been." Not "take this with a grain of salt, but Kevin Love comes up higher than all of Allen Iverson's career years combined." (Sorry about all these exaggerations, but it's how the Wages of Wins blog actually sounds, and it's seeping in to my own style to recall). And unfortunately while they may grasp the rules of basketball very well (and honestly do seem to be fans when all is said and done), they seem to regard any attempt to use the sport of basketball - as it empirically is, not just as the Platonic rulebook deems it to be - as a personal attack. It's hard to get them to acknowledge anything about the sport of basketball as they personally see it, opting for a bizarre and non-committal set of descriptions consistent with their intellectual views. "Kevin Love is amazing! Just look at these WP48 numbers!" Never a hint that they are watching, though for such committed people they must be watching.

* The preceding is statistical fact.

It's sad to say, but they don't have grains of salt, they don't have much context, and they don't have a multidimensional set of perfectly reasonable statistics. They're Johnny One-Note, and even though they at one point had one of the most reasonable concepts and some of the best experiments in all of basketball stats (heck, in all of sports), they just couldn't sustain one note for five years, and instead of adapting, "stretching the game out" in hip-hop lingo, and taking as an idol the braintrust which produced their fresh ideas, they Rip Van Winkled in a ranch of laurels, and curmudgeonly cut all ties with the rest of stats culture, and now subsist on incredibly lazy extensions of their original ideas, dogmatic and misguided rants about the problems in decision-making that they at one time helped to ameliorate a bit, and actually quite-good visualizations of their concepts. They idolized their original positive claims and made a normative universe around them until everyone accepts their claims. I don't know that I'm a humanist or anything, but if they're going to idolize their powerful claims, they probably could have done better idolizing the minds and processes that created those claims in the first place and then built their normative universe around these minds and processes. If they did this, we as fans - and writers, and intellectuals, and basketball enthusiasts - could at least begin to get behind them.

For a mind - infinitely more than a perfectly reasonable statistic ever could - can reason perfectly well.