Of course! The Padres have played poorly over the past two weeks; but, considering how poorly we fans expected them to play from Opening Day on, I think we're still basking in gratitude that they've done so well. Call it the Gift Horse Effect.

*I don't think the Pythagorean % over a 10-game period means very much, unless I misunderstand the statistic. I've always thought it was an empirical formula adjusted over many, many games to predict the teams's wins based on runs scored and runs allowed.*

That's my understanding too. The only reason I mentioned the losing streak in connection with Pythagorean percentage is that, purely on the basis of runs scored to runs allowed, a 10-game losing streak ALWAYS decreases a team's overall season Pythag over the interval from game *x* to game *x*+10, because the formula won't "predict" a zero-win streak unless the team is shut out in every game. As for using Pythagorean percentage to predict whether a team will win an individual game . . . that's as much of a fool's errand as anything Las Vegas can offer. (-;þ

Believe me, I'd love Pythagorean percentage to mean more than it does. Earlier this season, when the Pirates were winning most of their one-run games and losing all the blowouts, I tried to figure out some way to show that they were having a historically strange season. I couldn't do it. Having something like an eight-game difference between expected and actual W-L % over a forty-game stretch is interesting, but it really doesn't mean very much. According to the Pythagorean formula, now that the season is more than 80% over, the Pirates are four games off their expected percentage. That's too close to "normal" to be interesting, even retrospectively.

]]>I didn't want to write a "simple" comment about the Braves, I wanted directly to critique your word choice because I think you often aren't careful enough with your words. You speak as if certain statistics (Pythagorean W-L, W.A.R.) offer truth when in fact they offer one perspective. You'll trot out statements like "Who's the worst .300 hitter this year?" or "Who was the worst player to ever knock in 100 runs" and sure enough there'll be a list, sorted by Wins Over Replacement, as if this is the magic statistic that divines all facets of a player's value. Or if Pythagorean W-L is something more than another way of looking at a team's win-loss record that sometimes has value and sometimes doesn't.

I like reading what you write, but I wish you'd be more careful to realize that no statistic even comes close to evaluating a player's value, a team's "luckiness," or anything else really. That's why we have so damn many statistics.

]]>Thank you, David. Hmmm.... I need to re-take my Statistics 101. ]]>

No Neil, that's the Gambler's Fallacy. If you flip a coin ten times and get heads ten times, the likelihood that the eleventh flip will be heads is 50%. Regression does not say that good luck is offset by bad luck going forward (or vice versa). Regression says that good or bad luck is replaced by *random* luck going forward. Because of this, there's no way to predict when streaks end (or when they start).

The only thing you can look at is the underlying signal... good teams will win ~60% of the time, bad teams ~40% of the time. So good teams will have more winnning streaks than losing streaks (and vice versa).

You always get a big Pythagorean swing with a late season streak. During such a streak, you'll change the season WPct significantly while barely making a dent in the full-season RS/RA ratio. For the Padres recent streak, I would have guessed a swing of 6 or maybe seven. Using full-season RS/RA they were likely expected to win 6 or 7 but instead they lost them all.

]]>Tuna, don't expectations always color how critical fans and media are of a team's performance in a given season?

With respect to the 2010 Padres compared to the 1982 Braves, I'm not sure what it says for the Padres post-season chances. In their favor, the Giants have only gone 5-5 over the current losing streak.

I'm interested in the mid-season Pythagorean as an indicator of how likely they are to snap out of the losing streak.

I don't think the Pythagorean % over a 10-game period means very much, unless I misunderstand the statistic. I've always thought it was an empirical formula adjusted over many, many games to predict the teams's wins based on runs scored and runs allowed.

How can one distinguish the -3.3 Pythagorean as a predictor of Padres wins in the next few games as compared to random probabilitiy that a team will lose 11 straight games? The random probability the Padres will win tonight is (1.00 - 0.5^11) = 95%.

]]>