May 27, 2004
Good, Bad or Luck?
What does it mean to be a .500 team? Normally, we think of a .500 team as one that wins as many games as it loses, and that's certainly a good definition. But a .500 team could also be a team that had an intrinsic .500 winning percentage. That is, a team that was like a fair coin, as likely to win as to lose.
Now, when you flip a fair coin a number of times, it's not always going to come up heads and tails the same number of times. Sometimes, it's not even going to be close. If you flip a coin 45 times, probability predicts that 95% of the time, you will get between 16 and 28 heads.
When a team that's instriniscally .500 plays, the result of the games are like the flip of a coin. A win is heads, and a loss is tails. Now, most teams are about 45 games into the season, so we'd expect teams that are intrinsically .500 to have between 16 and 28 wins. As it turns out, all but three teams have between 16 and 28 wins! Does that mean that those 27 teams are intrinsically .500? No, but it does mean we can't tell. A team that is 16-29 could really be an unlucky team that is intrinsically .500.
It also means that if you set up 30 teams that are really evenly matched, you could get standings like the ones you'll read in the papers tomorrow morning by pure luck. That's why when you see a team like San Diego playing well, or a team like Seattle playing badly maybe there's been a change in talent, or maybe just a change in luck.
That's why, when I see a team like Detroit, that's playing better than it's intrinsic winning percentage, I'd like to see them build on the luck by actually improving the team.
Posted by David Pinto at
11:40 PM
|
Standings
|
TrackBack (0)
Agree with much of their perspective or not, the folks over at Baseball Prospectus have some interesting ideas, as well as statistical tools...my favorite is the "Pythagenport" third-order winning percentages. The Pythagorean Winning Pct (Winning Pct = (RS^2)/(RS^2 + RA^2)) usually is rather a good match for a team's winning pct over the course of a season...better yet, current pcts tend to more accurately describe future outcomes than even their actual record. BP takes it a step further by taking out what they would call the "luck" factor, as well as park effects, etc., to derive Adjusted EQR and Adjusted EQRA...and basically, it supports the notion that Le Tigre actually have rather a good chance this year. In fact, their offense has developed into one of the most potent...their actual runs scored likely being further suppressed by all their time at Comerica...for them to compete (and according to Pythagenport they're about 1/4 of a game behind the Chisox) they're going to need pitching...you figure, why not? Pick up a prime starter and a solid reliever, and you've got a chance. At the very least, you ride what clearly is a sense of excitement in Detroit and breathe a little life into what can be a tremendous baseball town...
But prime starters and solid relievers are hard to come by cheaply. While I agree that lowering runs allowed is a more efficient way to increase your winning percentage (see these graphs for the reason why, in the Tigers case it's probably easier for them to increase their runs scored with a better 1B/DH, which should be readily available.