Baseball Musings
Baseball Musings
February 16, 2009
Selfish Stats

I'm reading Michael Lewis's latest column in the New York Times, and this paragraph fascinated me. He speaking with Daryl Morey, a former STATS, Inc. employee who went on to the MIT Sloan Business School and became GM of the Houston Rockets:

When I ask Morey if he can think of any basketball statistic that can't benefit a player at the expense of his team, he has to think hard. "Offensive rebounding," he says, then reverses himself. "But even that can be counterproductive to the team if your job is to get back on defense." It turns out there is no statistic that a basketball player accumulates that cannot be amassed selfishly. "We think about this deeply whenever we're talking about contractual incentives," he says. "We don't want to incent a guy to do things that hurt the team" -- and the amazing thing about basketball is how easy this is to do. "They all maximize what they think they're being paid for," he says. He laughs. "It's a tough environment for a player now because you have a lot of teams starting to think differently. They've got to rethink how they're getting paid."

There is such a stat in baseball, runs. For an individual to score a lot of runs, he needs to be skilled in a number of baseball abilities; getting on base, running and hitting for power. If a player wants to drive in 100 runs in a season, he can do that by delivering in about 80 plate appearances, and do little the rest of the time. A run scorer, needs to constantly set up teammates, and take advantage of stolen base opportunities, be able to score on long hits, or hit for power himself. Another reason runs scored is a bit of an underrated stat.


Posted by David Pinto at 12:46 PM | Statistics | TrackBack (0)
Comments

No, that can't be right, David. I can think of many situations where a player maximizes his expected (personal) runs scored at the expense of his team's win expectancy.
One really obvious one: any time a sacrifice increases win expectancy. A successful sacrifice yields 0 expected runs for the player.
Another: the expected value of a steal attempt will be higher for the player than for the team (getting thrown out will often snuff a rally, but a rally is only a single run to the player).
And here's a strange one: reaching on a FC will be a positive value for a player, but negative for the team. This would mean batters should work at hitting grounders with men on base more often. That's got to be a bad incentive...

Posted by: James at February 16, 2009 01:33 PM

The whole time I read that article I kept thinking about how basketball is going through the statistical evolution that baseball went through in the 90's.

I liked Lewis' comment that baseball is an individual's game masquerading as a team game. Very true - how much better would most front offices be if they would accept certain basic truths and embrace the use of advanced statistics? (I'm looking at you, Reds)

Posted by: D-Rock at February 16, 2009 01:37 PM

No, runs scored is a deceptive stat when it comes to the individual. You can get on base and do everything right, but if you don't have someone batting behind you to drive you, you're not going to score many runs.

Posted by: sabernar at February 16, 2009 01:59 PM

Runs scored is an underrated stat? How's that? Runs, are THE most important stat in the entire sport. Hands down.

Posted by: Devon Young at February 16, 2009 02:40 PM

Devon, when David says the stat is underrated, he of course means that it is more important than is generally recognized. So when you it is *the* most important stat, you are agreeing with David!

Posted by: James at February 16, 2009 02:49 PM

Runs scored is a somewhat deceptive stat -- but no more so than RBI, which are far more highly prized by general managers and MVP voters. That's why runs is an underrated stat, and RBI overrated.

Yes, runs are dependent on the performance of others. But it's even more dependent on the individual's ability to (a) get on base, and (b) advance himself through power or speed.

Posted by: jvwalt at February 16, 2009 03:29 PM

Well, of course Michael Lewis has a bit of incentive to paint basketball at a similar place stats-wise to baseball 10 years ago, but certainly interesting...

My biggest take-away is that basketball is a game in inverse probability. The expected outcome on any given possession is a scoring event. This is inverse to baseball where the expected outcome on any play is an out. Even on plays without an out, a run scoring event is still a very low probability.

This explains why counting stats are not terribly useful in basketball. It's meaningless to say Kobe scored 60 points in a game if you don't measure that against how many possessions it took him.

The ultimate stat in basketball for defense would be Possessions Thwarted--of course, measuring that objectively is where all the difficulty lies.

Posted by: billy mueller at February 16, 2009 03:59 PM

FT%

Posted by: bandit at February 16, 2009 09:20 PM

If the player's incentive is FT%-based, then the player has a disincentive to try to get to the free throw line once he achieves the needed % and the needed attempts (assuming there is a minimum attempts requirement).

Posted by: Barron at February 17, 2009 11:29 AM
Post a comment









Remember personal info?