Dispelling the Andy Pettitte Myth

A friend of mine shot me a text the other day asking whether sabermetrics takes into account the ability of certain pitchers to perform better in "big games." The question seemed to be prompted by Andy Pettitte's recent 7 inning, 5 hit, 2 earned run ALDS victory over the Minnesota Twins.


I asked my friend quite honestly which pitchers have this reputation, as I really have no idea which players are considered "clutch." I mentioned James Shields, since I knew he's sometimes called "Big Game James," and my friend added Pettitte and Matt Garza.

The player I wanted to focus on is Andy Pettitte, mainly because I fear that his gaudy W-L numbers (240-138 over 16 seasons) may provide ammunition for those who will later argue his Hall of Fame case. I have nothing against the guy personally (I actually kind of like him as a player), but he's pretty much Exhibit A for why using wins and losses as an evaluation of pitcher performance is, well... wrong.

The first thing I wondered is why we were looking just to Pettitte's performance from 2004 onward. This seems like an arbitrary starting point. Taking a quick look at Pettitte's traditional numbers, it appears that he is the exact same pitcher from regular season to postseason:

Career regular season numbers: 3.88 ERA, 2.34 K/BB in 3055.1 IP
Career postseason numbers: 3.87 ERA, 2.33 K/BB in 256 IP

Pettitte's postseason ERA since 2004, however, is 3.46. He has gone 6-1 in that span, in 11 starts and 69.1 innings pitched.

My friend mentioned that the commentators for Pettitte's win against the Twins seemed to imply that Pettitte has just been able to perform better in big games since 2004, as if something just clicked with him. Two thoughts immediately came to mind. First, using wins and losses to evaluate an individual pitcher's performance is one of the worst metrics one can use, for reasons that have been discussed at length by many writers more able than myself. Second, 69.1 innings is just not an adequate sample size on which to base any real conclusions about a pitcher's skill set.

After digging into the numbers a bit more, however, I noticed that even aside from these two points, Pettitte actually has not really been that good in the postseason since 2004. Since I couldn't get Baseball Reference to calculate FIP for me over a particular postseason span, I just did it myself. What I found was this: from 2005 to 2010, Andy Pettitte's postseason FIP is 4.31. For his career, Andy has a 4.19 FIP.

Pettitte has been the beneficiary of a decent amount of luck, and a whole lot of run support during that time. On the luck side, Pettitte has a .286 BABIP over that span, compared with a .315 number for his career and a typical .300 average across all pitchers. On the run support side, the Houston and New York lineups during that time frame provided him with an average of 5.18 runs per game.

Looking to the numbers over which Pettitte actually has substantial control, we see that his performance from 2005-2010, while certainly not poor, do not provide any evidence that he pitches better in high-pressure games. He has walked 2.6 batters per nine innings, struck out 6.5 per nine, and has allowed 1.2 home runs per nine since 2004. Compare these numbers to his career 2.8 BB/9, 6.6 K/9, and 0.77 HR/9. While he has walked fewer batters in this limited sample size, he has allowed home runs at a significantly higher rate.

Upon further reflection, maybe we should be asking ourselves "Why has Andy Pettitte pitched relatively poorly in postseason games since 2004?" Have the nerves gotten to him? Is New York too much of a pressure-filled environment? Have the PEDs worn off? The answer, of course, is the same as before. Don't be fooled by small samples, and don't take at face value traditional numbers that often obfuscate a pitcher's true skill set.