Saturday, October 30, 2010

The Luck of Clay Buchholz

There's been a lot written about Clay Buchholz over the years. Prior to being drafted by the Red Sox in 2005, he was arrested for stealing computers from a middle school. In 2007, in his second career start, he threw a no hitter against the Baltimore Orioles on 115 impressive pitches. In 2008, amidst incredibly high expectations, Buchholz struggled to the point where the Red Sox eventually demoted him to AA. After a rebirth in 2009 (which resulted in a Game 3 ALDS start), Buchholz was given a spot in the Red Sox rotation.

In 2010, Buchholz had a very successful season, particularly by traditional standards. He put up a 17-7 record with a 2.33 ERA, and was frequently named as a serious contender for the AL Cy Young. On the surface, Clay Buchholz appears to have been one of the majors best pitchers. While Buchholz was very good, I'm here to tell you that he's probably not worthy of serious consideration for the Cy Young Award. (Gasp! I know. A Red Sox fan is actually being objective about a player on his team. This is incredibly rare.)

Buchholz seems to have benefited from an extraordinary amount of luck in 2010. How do I know this? Well, for starters, I looked at his batting average on balls in play (BABIP). In a given season, most starting pitchers will register a BABIP between .285 and .310, with the majority of pitchers being within a few points of .300. Pitchers that register BABIPs outside of that range in one season will typically see that number regress back to the expected range the next season. Based on the numerous studies, analysts have found that pitchers have little, if any, control over their BABIP. Essentially, a pitcher's success with this statistic is largely (but not entirely) based on luck and the quality of the defense behind them. So how does this affect Buchholz's performance? In 2010, Buchholz gave up 142 hits, resulting in a .265 BABIP, which is significantly below the expected range. To determine that amount of "luck" Buchholz had on balls in play, I calculated his expected BABIP (xBABIP) based on his rate of giving up line drives, fly balls, and ground balls. After stripping away luck and quality of defense, I was able to calculate that Buchholz probably should have registered a BABIP closer to .301 (nearly dead on with the expected .300). This would've resulted in Buchholz giving up 154 hits, as opposed to 142. While 12 hits might not seem like that big of a difference over the course of 28 starts (and it isn't), it's the timing of those hits that matter.  Using linear weights, a single is, on average, worth 0.9 runs (based on wOBA).  Multiplying 12 singles by 0.9 runs, you get 10.8 runs which we'll round up to 11 runs. This would have a significant effect on his ERA, raising it from 2.33 to 2.90.  His ERA could, of course, be higher if he gave up those 12 hits with runners on second and third in each situation. 

Buchholz wasn't just lucky on balls in play. He was also lucky on balls hit out of the park. To determine his rate of luck in this factor, I chose to use a statistic called home run per fly ball (HR/FB) rate. Like with BABIP, HR/FB rate seems to be a normalized rate. By and large, pitchers typically register a HR/FB rate between 9.5-11.5%, with the league average being at 10.5%.* If a pitcher registers a HR/FB rate well above or below that range in one season, he'll typically regress back to that range the very next season. After registering HR/FB rates of 14.7% and 15.7% in 2008 and 2009 (albeit in smaller sample sizes), Buchholz registered a 5.6% HR/FB rate in 2010. This stands out as a little fishy for two reasons. One, his 2010 rate is well below the expected rate of 10.5%. Two, his 2010 rate is significantly lower than his rates in 2008 and 2009. While I can surmise that Buchholz was somewhat unlikely in 2008 and 2009 on HR/FB rate, I'm fairly certain he benefited from the pendulum swinging the other way last year. So how lucky did Buchholz get? Well, in 2010, he gave up nine home runs. After normalizing his HR/FB rate at 10.5%, I found that Buchholz was expected to give up 17 home runs based on his fly ball rate. At minimum, that's an additional eight runs toward his ERA--and that's only if we assume all of those expected home runs were solo shots.  Assuming all eight home runs is not only overly simplistic, but it's also unlikely.  Instead, we'll use the linear weights estimate (derived from wOBA) for run expectation of a home run--1.95 runs.  Taking 1.95 runs and multiplying it by eight home runs, you get 15.6 runs, which we'll round to 16 runs.  After adjusting his ERA for the additional sixteen runs (and the 11 runs from the BABIP adjustment above), that bumps his ERA up to 3.73--not far off from his 2010 Fielding Independent Pitching ERA (FIP) of 3.61.

* There are factors that can push this rate above or below this range for certain categories of pitchers. For instance, pitchers that pitch in extreme environments could have their HR/FB rate affected. Fenway is not one of these environments, as it favors doubles over home runs.

Additionally, the rate at which a pitcher gives up ground balls does not factor into the equation. A pitcher that has a 50% ground ball rate should theoretically have a similar HR/FB rate to a pitcher that has a 35% ground ball rate. That said, the pitcher that gives up ground balls at a lower rate should give up more home runs because of his elevated fly ball rate.

I don't want to give the impression that Buchholz wasn't a good pitcher last season--he was I also don't want to give the impression that his future isn't rosy--it definitely is. I do want to caution that there are factors that contributed to his performance in 2010 that we can't (and shouldn't) expect to be repeated in 2011. If he was to repeat his season exactly (in terms of strikeouts, walks, batted balls, etc.), we would likely see him register an ERA closer to 4.00 rather than 2.30.

Buchholz does have a bright future ahead of him.  He gives up ground balls at a rate above 50%, which means he'll be one of the better pitchers when it comes to avoiding home runs.  He induces contact at a rate below 80%, while inducing a whiffs at nearly a 10% rate. This indicates that we should see his K/9 rate increase from 6.22 to around 7.50 or higher (based on pitchers with comparable rates). In turn, this would improve his K/BB rate, while reducing the number of balls put into play (and therefore hits allowed).  One area that Buchholz needs to improve upon if he's ever to become an elite pitcher is his walk rate.  His 3.47 BB/9 rate was in the bottom quartile of American League qualifying pitchers.  Simply by reducing his walk rate, he'll be able to avoid the larger variances in performance that stems from poor luck (through BABIP and HR/FB rate).


  1. of course you can add his high strand rate and average line drive rate.

  2. The average line drive rate is already accounted for in his xBABIP that I calculated to be .301 (versus a BABIP of .265). Additionally, the additional 27 theoretical runs scored as a result of normalizing his BABIP and HR/FB rate would've drastically lowered his strand rate from 79% to something closer to the league average (around 72%).