The Dominance Quotient
When I was a teenager, my friends and I used to waste time at school by talking about the basketball box scores from the night before. (A box score is rows and columns of statistical information: minutes played, rebounds, assists, points scored etc. I think it started as a baseball term. Scores in a box.) We wanted to come up with a formula that measured how good a player was: the Dominance Quotient, we called it, only slightly self-mockingly.
The easiest way to judge a player’s performance is to look at how many points he scores. But a player who takes twenty shots to score twenty points is less useful than one who can score the same amount from ten shots. And what about assists, the passes that help other people score? And rebounds (corralling the ball after a miss), turnovers, fouls? How much should they count? The first thing we did was come up with a formula for what is now called ‘effective field goal percentage’, a measure of how many points, on average, players score for every shot they attempt. Then we added and subtracted lots of other statistics and multiplied them by this percentage to come up with a player’s DQ. We each picked a favourite player and compared their Dominance as the season went on. Good times.
It never really occurred to me that the DQ measured only itself – that it bore no necessary relation to anything that actually determined the outcomes of games. If we changed the formula, the numbers would change, but that’s all. Or rather, what hadn’t occurred to me is that you could come up with numbers that do bear such a relation: that a basketball game is made up of a series of small outcomes, each of which has a quantifiable effect on the likelihood of one team’s winning or losing, and that various players’ contributions to those small outcomes can be assessed and compared with each other. We were high school kids, the DQ gave us something to argue about, dayenu. But the child is father of the man. My best friends were identical twins. One of them became a maths professor, who specialises in dynamical systems. The other became an economics professor, who works on game theory. I tried (briefly) to be a basketball player.
In Moneyball: The Art of Winning an Unfair Game, which came out in 2003, Michael Lewis describes the history of the attempt to do to baseball what we had been trying to do to basketball – to quantify what wins. But baseball is a game composed of discrete moments, primarily involving two players (the pitcher and the batter), each of which results in a measurable outcome. It could have been designed by economists, for economists; it’s too easy to be really satisfying.
Football is notoriously harder to analyse. It involves complex interactions between 22 players, it isn’t broken up into discrete moments, and it produces few measurable outcomes – or at least, the outcomes you can measure (shots on goal, successful dribbles, completed crosses etc.) have a complicated relationship to winning and losing. There’s plenty of data now, computers can break down the game in lots of different ways, but it isn’t easy to make predictions on the back of this data. So conversations about football still sound more like conversations about politics – in which ideology, of one kind or another, has to fill the gaps left by an incomplete knowledge of the facts.
Basketball, meanwhile, offers an almost perfect balance of complexity and measurability. Like football, it’s a fluid sport made up of the intricate interactions of large numbers of people; but like baseball, it produces many measurable outcomes. An average NBA game involves something like 200 points, so there’s a lot of reliable data about what works and what doesn’t, with the consequence that basketball is going through something of a golden age in commentary. And the commentary is changing the way the game is played. As Brian Windhorst, a sports writer who started at the Cleveland Plain Dealer covering LeBron James, said to Zach Lowe, a journalist for ESPN and a former high school history teacher, on one of his podcasts, there’s somebody at every club in the league whose job it is to listen to what you’re saying here.
Moneyball was based on a simple insight: that baseball clubs had tended to undervalue players who got ‘walks’ (when you advance to first base without hitting the ball, because the pitcher has missed the target too often). Walks are undramatic, they look more like a fault of the pitcher than a virtue of the batter, but it transpires that not getting out has a big effect on how many runs you score, and games you win. The new NBA is based on a similarly simple premise: that three points are worth more than two (you score three points if you shoot from outside the three-point arc, roughly 24 feet from the basket), and the mid-range jumpshot, once the symbol and instrument of the canny veteran (commentators used to complain about the lost art of the mid-range jumpshot), is really the worst shot in basketball.
But that’s only the beginning of what the data can tell you. Computers can measure the success rates for open shots and contested shots, shots from the left corner and from the right, shots taken off the dribble or after a pass, and how often your team scores after you miss. So a coach might think: ‘We need that guy to shoot more, because good things happen when he misses.’ The apparent chaos of a basketball game is made up of a series of repeatable micro-actions (most of them planned), and each of them can be tracked, so you can work out on average how many points a player or combination of players generate per unit of play.
All of this may sound fairly dry, but it’s trying to make sense of the question that anybody who has ever played a team sport will remember asking after a game: what the hell happened there? (‘Football is a simple game,’ Gary Lineker once said. ‘Twenty-two men kick a ball for ninety minutes, and at the end the Germans always win.’) Professional athletes are very good at intuiting or thinking through the things they do, the choices they make in the heat of the moment, that work, that produce results; now they have tools for assessing whether those intuitions are right. And the carryover to our own lives is not merely interesting but slightly frightening. What if we had data to tell us what percentage of the time teasing works when trying to get our children to clear the table? Passive aggression? Shouting? Because the number exists.
Coaches have always preached unselfishness and teamwork, as if the good of each were best served by the good of all, but that’s a lie. Most players have always been better off chasing individual stats (points, mainly), rather than team victories, if they want to get paid when their contracts come up for renewal. To tell them otherwise is one of the lies of team games – an attempt to enforce a hierarchy that benefits only the people at the top. But now the data is refined enough to show the difference between useful and harmful selfishness.
What teams want now (apart from superstars, of which there’s a limited supply) are ‘3-and-D guys’: players who are tall enough to guard big men, quick enough to guard small men, and can hit open three-pointers, especially from the corners. They don’t have to be able to shoot off the dribble, or even drive; it doesn’t hurt if they can pass, but that isn’t the point. There’s a lot of money at stake, and part of the pressure comes from the fact that teams are strictly limited in how they can spend it. There are complicated rules, which I don’t understand, governing different kinds of contract (for rookies, free agents, long-term players, all-stars etc.), so that every possible basketball virtue, from each phase of a player’s career, has to be assessed and valued, and the teams that win have general managers who get these assessments right.
So the new breed of basketball writer has to combine many different talents. In part they’re just fans, who love talking about this shit the way I did at high school. In part they’re reporters, investigating locker-room tensions and trade rumours, medical issues and the complicated relationships between owners, coaches, players and agents. In part they’re like old-school scouts, who can break a complex game down into what goes on under the surface of our attention when we watch it, like a swan’s legs in the water. In part they’re data analysts, who can crunch the numbers to find out what’s working and what isn’t. And in part they’re almost a kind of forensic accountant, who can get underneath the rules, collective bargaining agreements and player contracts, to work out the extent to which what goes right or wrong on the court is the end-product of a series of legal and financial negotiations.
In other words, they’re great. I can’t think of a better time to be a fan. Writers like Zach Lowe, Kevin Arnovitz, Ramona Shelburne, Chris Herring, Howard Beck, on sites like ESPN, 538 and the Bleacher Report, have reminded me what talking about sport always seemed to be about when I was kid: an attempt to understand the way reality works, in a version of reality which is not only fun to play yourself, but where the stakes are low enough there’s no reason to stop arguing about it.
The NBA Finals start tonight, and for the fourth year running they feature LeBron James's Cavaliers against Steph Curry's Golden State Warriors. The sample sizes just keep growing.