Why ‘Expected’ Is Important

Roy Oswalt has been one of the best and most consistent pitchers in baseball this decade. He never got enough credit, likely due to pitching with Roger Clemens for the last few years, and does not have an intimidating frame out on the mound, but any team would be better with him on their staff. This year, though, many analysts have questioned what is wrong with Astros righty. Baseball is a game of reputations, and it is very possible for a player to have an entire season generalized on great or poor performance either very early or very late into a season.

This seems to be the case for Oswalt, who pitched poorly early on, but has been great recently, arguably the best starting pitcher in baseball over the last thirty days. Don’t believe me? Heck, based on his performance early on and the lack of national publicity directed towards the Astros, I wouldn’t believe that either, but his 1.47 WPA/LI in this span outranks everyone else.

His recent performance has lowered his seasonal ERA and FIP both to 3.91 which, while good, are definitely worse than we have come to expect from Oswalt. His FIPs since 2004 have been 3.17, 3.16, 3.30 and 3.59, leading to this year’s 3.91; it appears his controllable skills are steadily declining. Are they? His K/9 is the highest it has been since 2004, while his BB/9 is almost a half-batter lower than last year. His HR/9, however, which has ranged from 0.59 to 0.73 recently, is currently 1.09, which explains quite a bit.

How has the home run rate increased? The culprit is a HR/FB rate that has gone from 6.7-8.7% over the last four years to 13.8% this year. The league average in this category is generally 11-11.5%, so Oswalt had outdone it by a good margin for a few years. This year, though, his rate has been worse, despite decreasing quite a bit over the past month.

This leads into the title of the post. Yesterday, Pizza Cutter, my colleague at Statistically Speaking, wrote an article about Kelly Johnson in which he discussed how performance is equal to talent and luck; last year, Johnson’s performance was more luck-driven, whereas this year his performance is worse, but comprised of more talent. How does this apply to Oswalt? Well, since his HR/FB was so much lower than the average for those years, his FIP might not be the most accurate indicator of controllable skills. Look what happens when we introduce xFIP, which normalizes the home run factor of FIP:

Year   FIP     xFIP
2004   3.17    3.73
2005   3.16    3.56
2006   3.30    3.75
2007   3.59    4.08
2008   3.91    3.74

According to these numbers, Oswalt’s FIP is expected to be right around where it should have been from 2004-2006. The difference between FIP and xFIP is at its smallest point this year, meaning that Oswalt’s current performance is more talent driven than the result of “luck,” which I use very loosely because there may or may not have been something in his repertoire that limited home runs. His flyball rates have stayed stagnant, however, eliminating the possibility that perhaps he just had very low percentages of flyballs in those years.

We might not know how much talent or luck Oswalt has or benefits from, but he isn’t as bad as some people made him out to be earlier in the season. His xFIP has been in the same general range for the last five years. If there is something wrong with him, it would be that this is the first year in that span in which his FIP has not surpassed what was expected of him.