A week ago, I asked whether Hampton's bad game against the Nationals tells us much about his pitching ability. I looked at his games this season compared to his exceptional 22 win season in 1999 and concluded that the "dud" games aren't really that different in 1999. If anything, the difference between 1999 and 2009 for Hampton is revealed by the higher number of very good games in 1999. This raises the question: which Astros' pitchers have been the most consistent from game to game this year. And I intend to answer the question by the end of this article.
However, first I will detour to some recent discussion of Barry Zito's jekyll-hyde tendency this season. A San Francisco newspaper writer had suggested that Zito's high rate of "dud" games made him a terrible pitcher, i.e., worse than other pitchers who put up similar ERAs. Dan Symborski at Baseball Think Factory posted a recent article analyzing this contention. Read his article for yourself. But his finding is interesting: Zito's extreme variabiliity in performance from game to game actually made him more valuable than if he was perfectly consistent. Given the Giants' offensive pattern, the Jekyll-Hyde pattern for Zito can achieve a winning percent of .433, which is better than Zito's Pythagorean derived winning percent of .367. A theoretical "Mr. Consistent" Zito has a .317 winning percent, which is less than the Pythagorean prediction.
That raises some interesting questions for the Astros. The most obvious conclusion challenges our preconceptions about whether inconsistent pitchers are worthy of the rotation. However, because of the Astros' tendency to over-perform Pythagorean records, maybe this helps us answer an elusive question raised many times here: is there a characteristic which makes the Astros more likely to beat their Pythagorean prediction? Perhaps the variability of the pitchers in the starting rotation has some bearing on the qustion. Below the jump, I will look at the consistency of Astros' pitchers in the rotaiton.I used the game log for each Astros' starting pitcher, calculating a Runs Allowed Per Nine Innings (RA/9) for each game. I excluded relief appearances. Standard deviation and variance are statistics frequently used to evaluate the dispersion of data. A lower standard deviation is associated with less variability in the data, and vice versa. Shown below are the Astros' pitchers, from most consistent to least consistent, as measured by standard deviation of RA/9.
PITCHER -- Std. Deviation of RA/9 per game
The fact that Roy Oswalt is the most consistent pitcher is no surprise. Perhaps surprisingly, Hampton and Ortiz are the next most consistent pitchers. Perhaps reflecting the popular criticism of Wandy as inconsistent, Wandy is less consistent than Hampton and just behind Ortiz in consistency. Keep in mind that this isn't a measure of best pitchers, just the consistency from game to game. Wandy's overall runs allowed are clearly better than the more consistent pitchers above him in this ranking. Paulino and Moehler, by far, are the most inconsistent starters in the rotation. I expected Paulino to be the most inconsistent, and I was surprised that Moehler beat him out for that distinction.
However, based on the BBTF article referenced earlier, inconsistency may be a positive attribute, in terms of winning percentage, for the lower end of the rotation. Whether inconsistency has a positive impact depends on the pitcher and the offense behind him. As a simplistic example, Paulino has a 5.80 ERA as a starter. But he has received average run support of 2.79 runs/game. If both the offense and Paulino were perfectly consistent, the Astros can't win any games during his starts. However, Paulino actually has three games in which he had a RA/9 which was less than 2. Thus, Paulino's variability gives the Astros a chance to win some of his games.
In the near future, I may explore this issue further.