Gelf Magazine points out that
Gelf Magazine points out that some statistics are truer, and thus more useless, than most:
It's a good lesson, in two parts. The first is that using a results-based statistic as a measure of results is often not-interesting; usually, you want some form of predictor instead. But the second lesson is more telling. If you have a perfect metric, that correlates exactly with your end state, you're probably measuring your end state.
For those of you too lazy to do the math, SI's True Gauge (road wins - home losses) is exactly half of the standings metric (wins - losses) whenever a team has played the same number of road and home games. There is one benefit that SI's metric has: two closely matched teams, one of which has one or two more home games, and one of which has one or two more road games, can be more readily compared. But that's a pretty fine comparison, and clearly not in the intent of the quote lifted for Gelf's review.
While we're tweaking SI: the magazine ran an embarrassing NBA item a few weeks ago. Entitled "True Gauge," it announced: "Want to separate contenders from pretenders? NBA coaches and executives do so by looking at what is key come the playoffs: the ability to protect home court and to win on the road. Subtract home losses from road wins, and you'll find these teams on top." What follows is a list of the teams with the best differential between home wins and road losses; stunningly, these teams are the Pistons, Spurs, Mavericks, Heat, and Sunstandings don't. And yes, a won-and-loss record is a "true gauge" of an NBA team, and it's important to both win at home and on the road; or, you might also say, to win games. Also key, therefore, is to score more points than your opponent; look out for a chart revealing that insight on a newsstand near you.
It's a good lesson, in two parts. The first is that using a results-based statistic as a measure of results is often not-interesting; usually, you want some form of predictor instead. But the second lesson is more telling. If you have a perfect metric, that correlates exactly with your end state, you're probably measuring your end state.
For those of you too lazy to do the math, SI's True Gauge (road wins - home losses) is exactly half of the standings metric (wins - losses) whenever a team has played the same number of road and home games. There is one benefit that SI's metric has: two closely matched teams, one of which has one or two more home games, and one of which has one or two more road games, can be more readily compared. But that's a pretty fine comparison, and clearly not in the intent of the quote lifted for Gelf's review.
