Wednesday, May 30, 2007

Graphickry: Team Salary, Team Performance

Ben Fry has concocted an ingenious interactive graph showing the relationship between each Major League Baseball team's salary and won-lost record for any day of the season.

To criticize such a graph--certainly the most interesting and useful presentation of this data that I've seen--would smack of churlishness and ingratitude. Hey, that's my cue!

The problem with the graph is that it distorts the relative positions of the team's salaries and performance by presenting them in evenly spaced lists on the two sides of the graph. Fry seems to recognize the problem with this approach on the salary side and therefore represents each team's salary by its line thickness as well as its position on the right side of the graph. The result is two visual representations of team salary that contradict each other: the Yankees' position on the right side of the graph inaccurately presents the teams a one evenly-spaced slot above the Red Sox, whereas the thickness of the line accurately but unintuitively reveals the huge gap between the top two teams. Combined with the deceptively even spacing of the team records on the left side, this flaw creates line slopes that can get seriously out of whack: at some points, one win or loss creates a deceptively large change in the slope of a team line, and the Yankees and Red Sox should be more obviously in their own leagues at the top of the salary side.

I've done enough web programming to know that my ideal graph would be vastly harder to program than Fry's already-complex one, so I understand his decisions; I just hope someone will surmount the technical hurdles to make an even better version of this.

I would also add a point of substance: it seems to me that team salaries in this context should include the costs and benefits of the luxury tax system. Which means that the Yankees are doing still worse.

In other words, Fry's graph makes the Yankees' season (to date--let's be clear) look disastrous, but it's actually much, much worse than it looks.

Sunday, May 20, 2007

The Preakness: Odds and History

I was struck by the headline in The New York Times on the day of the Preakness Stakes: referring to Kentucky Derby winner Street Sense, it read, "Favored in Preakness by Odds, Less So by History."

The article, by Joe Drape, shows the opposite.

It opens with a reference to Street Sense's status as the 7-to-5 favorite in the Preakness. That means that the odds gave Street Sense just under a 42% chance of winning. Like other sports betting markets, that of horse races has a fantastic track record, so we can reasonably expect horses with 7-5 odds to win about 42% of the time.

The article then moves to this paragraph, quoting Street Sense trainer Carl Nafzger:

“What I know is that I have a 9-to-1 chance to win,” he said, referring to the number of starters in Saturday’s race. “And that’s a lot better than it was at Churchill Downs for the Derby when we were 19-1.”

Because I am a generous man and have a good dessert in my belly, I will assume that Nafzger likes to toy with reporters rather than that he is as silly as this comment. (By this logic, horse owners might as well toss any old entry into the Preakness; heck, but a bunny in the gate--it's a lot cheaper to feed than a horse, and it will still have the same chance of winning!) Even so, this is an inspired bit of nonsense. It combines obviously false egalitarianism with a classic confusion of probability and odds ("9-to-1 chance" rather than one in nine).

Drape makes more sense in the paragraph that follows:

History suggests that the odds are much better than that for Street Sense: 52 percent of Preakness winners were sent off as the post-time favorite, as Street Sense certainly will be. In the past 10 years, six Derby winners have won the mile-and-three-sixteenths race and headed to Belmont Park with a chance to sweep the Triple Crown.

And here the more interesting point arises: if pre-race favorites had won 52% of Preakness runs, and Derby winners have done even better in recent races, then Street Sense's 7-5 odds reveal a weaker favorite than one would expect, a horse given less of a chance to win by bettors than the performance of similar horses in the past would indicate.

"Favored in Preakness by Odds, Less So by History"? Nope--favored by history, a little less so by the odds.

Thursday, May 10, 2007

A blog is born

I started this blog after composing this little rant on May 10th, 2007, so I'll date this post to that day, call it a blog, and then try to explain myself.

I just realized something: Bill Simmons, the Sports Guy, is the anti-Bill James. Like James, Simmons is an excellent writer with a great knack for popularizing his ideas. I often enjoy Simmons's columns immensely; I am always reminding myself to cut blog subscriptions, and Simmons always makes the cuts. But Simmons's reasoning is TERRIBLE. He's just a dreadful analyst in many ways, foremost among them his constant use of misleading statistics combined with reflexively anti-intellectual bashing of stat geeks (like me, it should be noted). Perhaps the next blog in my life needs to be sportsguytalkincrazyagain.blogspot.com.