Every winter, there seems to be at least one topic that sparks a massive debate that nearly causes a civil war within Red Sox Nation.  Last winter, it was Jacoby Ellsbury’s UZR.  This winter, the debate has raged around the apparent luck (or talent, depending on your take) of Clay Buchholz.  On Monday, Peter Abraham of the Boston Globe decided it was his turn to give his take on the matter.

“Buchholz said he felt lucky to be named co-Red Sox pitcher of the year with Jon Lester and joked to the fans attending the banquet that they shouldn’t expect him to repeat that performance in 2011.

(snip)

But the peripheral statistics suggest that Buchholz was, in fact, lucky and that fans shouldn’t expect a repeat of that performance.

‘I don’t know the exact stats. But I know in my head it seemed like I had one point in every game where I caught a break,’’ Buchholz said last night after throwing two scoreless innings against the Minnesota Twins in an 8-4 loss. “That adds up after a while.’’’

Was Buchholz lucky?  Based on the statistics, it appears so.  According to Fielding Independent Pitching (FIP), a metric created by Tom Tango that focuses only on a pitcher’s “controllables” (strikeouts, walks, and home runs allowed), Buchholz’s peripherals indicate his actual performance (2.33 ERA) far exceeded the expected outcome (3.61 FIP).  While I’m not going to dispute that his ERA was outstanding, it’s important to note that ERA is not very predictive from year-to-year.   According to a study done by Baseball Prospectus a few years ago (that was published in their spectacular book Baseball Between the Numbers), they found the following year-to-year correlations with regards to certain pitching statistics:*

Statistic Year-to Year Correlation
Winning % .204
BABIP .272
ERA .380
HR/Batters Faced .470
Hits/Batter Faced .499
BB/Batter Faced .676
K/Batter Faced .790
GB % .807

* For those not familiar with correlation coefficients, the scale ranges from -1 to +1, with +1 meaning a perfect correlation; -1 meaning a perfect inverse correlation; and 0 meaning no correlation at all.

Based on the findings by Baseball Prospectus, we find there are five factors with stronger year-to-year correlations than ERA:  home runs, hits, walks, and strikeouts per batter faced and ground ball percentage.  So what does this mean?  It means that we should probably look at factors other than ERA when trying to predict a pitcher’s future performance.

The problem with ERA is that it’s so dependent on external factors like luck, defense, and park effects. For example, if you put Jon Garland in front of the mid-1970s Baltimore Orioles defense, he’d likely produce an ERA that far exceeded his FIP.  Why?  The Oriole defense during the 1970s was so efficient that they converted more batted balls into outs than the average defense.  Logically, fewer hits mean fewer runs. The same is true if you were to allow him to pitch half of his games at Petco Park, as he did last season when he posted a 3.47 ERA and 4.41 FIP (against career rates of 4.32 ERA and 4.69 FIP respectively.  Garland’s peripherals weren’t really any better than they had been in previous seasons with the Angels and White Sox.   While there was a noticeable improvement in his strikeout rate, that improvement was canceled out by his unusually high walk rate.  Other than that, his peripherals were nearly identical to his career norms.  Instead, he was helped by an efficient defense and an offense crushing ballpark.

Now, back to Buchholz…

While Abraham points out that defense likely wasn’t a factor in Buchholz’s unusually low .261 BABIP (or his absurdly low .188 BABIP with RISP) due to the Red Sox being a “poor defensive club” in 2010, I’m not 100% sure that’s true.  Obviously, losing Kevin Youkilis, Dustin Pedroia, Jacoby Ellsbury, and Mike Cameron certainly hurt the Red Sox’s ability to prevent runs, but defensive metrics like Ultimate Zone Rating (UZR) and Defensive Efficiency Rating (DER) indicate that the Red Sox had a league average defense in 2010.  Under this assumption, the most obvious conclusion is that Buchholz benefitted from some degree of luck last season.  In the case of BABIP, his luck could’ve been created by defensive positioning, a high number of spectacular (out of the norm) defensive plays in timely situations (like with RISP), base running miscues, etc.  Over the course of a season, a few plays here, and a few plays there can really add up.

The most popular arguments against Buchholz’s apparent luck are:  (1) these types of variances tend to even itself out over the course a season, and (2) Buchholz’s BABIP was caused by his ability to induce weak contact.  The problem with argument #1 is that it incorrectly assumes that 150-200 innings is enough for a pitcher’s ERA to become a reliable measure of performance.  While 150-200 innings might seem like a large sample size, it’s not.  In truth, ERA needs approximately 500-600 innings (approximately three seasons) to become reliable; hence the reason ERA has such a poor year-to-year correlation to performance.  Still not convinced?  Derek Lowe’s run from 2002-2004 is an excellent example of this.

The second argument implies that pitchers have an innate ability to not only guide a batted ball toward a particular fielder, but also purposely induce weak contact.  Let me tell you right now, if pitchers had the ability to guide a batted ball to a particular fielder, hitters would rarely get hits.  It’s just not possible.   While the pitcher has full responsibility with regards to the quality of the pitch (velocity, type, location, and movement), he has no control over what happens to the ball once it leaves his hand.  At that point, the batter at the plate and the defenders behind the pitcher are in control.  Additionally, while Buchholz may’ve induced a lot of weak contact last season (which is definitely up for debate) that neither means he’ll continue to do so in the future, nor does it mean that the weak batted balls will be fielded and converted into outs by the fielders behind him.  Like pitchers, batters make adjustments.  This winter, after having a chance to review video of Buchholz, hitters may choose to lay off some of the out-of-zone pitches that allowed weaker (not necessarily weak) contact last season.  As for the defensive aspect, choppers and bloopers are often considered to be weak contact.  Those types of batted balls are often not converted into outs because either a speedy base runner beats a throw to first, or the ball falls into play just outside of the reach of a fielder.  Most luck naysayers tend to ignore these facts when constructing their arguments.

Buchholz is a ground ball pitcher (50%+ ground ball rate in each of the last two seasons).  While this is certainly helpful in terms of abilities to avoid home runs and induce double plays, it actually doesn’t help him produce a lower than expected BABIP.  For each batted ball type, there is an expected BABIP (xBABIP).  For line drives, the xBABIP is .730; ground balls it’s .240; and fly balls it’s .150.  Just by reviewing those numbers, it’s pretty clear that fly ball pitchers, not ground ball pitchers, are more likely to produce low BABIPs.  Just for fun, I ran a query on Fangraphs for pitchers with the lowest BABIP between 2002-2010.  To limit potential outliers due to small samples, I’ve included only pitchers that pitched at least 1000 innings during that time period.  Here are the top ten pitchers:

Pitcher BABIP GB%
Matt Cain .266 36.3%
Ted Lilly .267 33.9%
Barry Zito .268 37.8%
Tim Wakefield .269 39.7%
Johan Santana .270 37.5%
Ryan Franklin .272 39.5%
Jason Schmidt .274 38.2%
Carlos Zambrano .274 48.9%
Jarrod Washburn .275 36.2%
Jamie Moyer .276 39.6%

A couple of things jump out at me.   One, other than Zambrano, each of the pitchers on this list have fly ball tendencies.  Two, Tim Wakefield, as a knuckleball pitcher, is a unique case in that low BABIPs are typical for his breed of pitcher.  (Knuckleball pitchers are the only proven exception, at least to date.)  Three, several of these pitchers (Cain, Zito, Schmidt, Washburn, and Moyer) all spent several seasons during the nine year sample pitching in ballparks with unusually large amounts of foul territory.  As fly ball pitchers, this clearly aided in their ability to garner a few extra outs that ground ball pitchers like Buchholz or Sabathia would not have been afforded pitching in stadiums with small foul territories.

What does the bottom ten look like?  Here’s the list:

Pitcher BABIP GB%
Sidney Ponson .318 51.2%
Carlos Silva .310 48.0%
Jason Jennings .309 45.2%
Aaron Harang .309 37.9%
Andy Pettitte .308 48.5%
Mark Hendrickson .307 44.7%
Mike Mussina .307 43.3%
Zach Greinke .307 39.9%
Curt Schilling .307 39.5%
Jeremy Bonderman .306 46.5%

This list is a little less conclusive, although he majority of pitchers on the list are those with ground ball tendencies.  After doing a little extra research, I found that the reason that fly ball pitchers like Harang, Schilling, and Greinke are included on this list is due to their high line drive rates.  (Remember, the xBABIP for line drives is .730.)  Furthermore, it’s interesting to note that most of the pitchers on this list have spent considerable time pitching in hitter friendly ball parks with smaller foul territory as well.

So how does this relate to Buchholz?  Well, for starters, considering his ground ball tendencies, it’s unlikely he’ll continue to post BABIPs that consistently register in the .260s.  The historical data doesn’t work in his favor.  Furthermore, his batted ball statistics indicate he should’ve given up 152 hits, as opposed to 142 hits last season (.293 xBABIP).  While this might not sound significant, using the linear weights as our guide, ten hits could potentially lead to as many as 8-10 runs depending on the scenario in which those hits are accrued.  This alone would theoretically bump his ERA up by half of a run.

Another factor Abraham mentions is HR/FB rate.  For the purposes of this discussion, I’m going to ignore 2007 because it includes only 22-2/3 innings (and a no hitter).  Between 2008 and 2010, Buchholz produced the following HR/FB rates:  14.7%, 15.7%, and 5.6%.  Putting everything we know about Buchholz aside (including the Recency Effect), which do you numbers do you think are more representative of his career norms?  In all likelihood, it’s probably none of them, but going by stats alone, I would assume the first two HR/FB rates.  Based on everything I know, Buchholz was probably very unlucky in 2008 and 2009, while receiving unusually good fortune in 2010.  While I can’t say for sure (because there’s not enough data yet to make an accurate evaluation), it’s a pretty good bet that Buchholz will see some regression next season; in fact, he’ll probably regress back toward the 10.5% league average.  For those of you who remain unconvinced that he’ll regress (even a little), I’ve done some more research.  Here’s a list (using the same parameters from my BABIP research above) of the ten pitchers with the lowest HR/FB rates:

Pitcher HR/FB Rate
Matt Cain 7.0%
Jason Schmidt 7.5%
Justin Verlander 7.6%
Cliff Lee 8.2%
Zach Greinke 8.5%
Barry Zito 8.5%
Carlos Zambrano 8.6%
Mark Redmond 8.8%
Jarrod Washburn 8.8%

There are a few fun things I noticed about this chart.  One, a lot of our friends from the BABIP charts have made an encore appearance on this one, including Barry Zito, Matt Cain, and Jason Schmidt.  Interestingly enough, all three have/had spent considerable time pitching in San Francisco.  I think it’s safe to assume that the ballpark has certainly benefited all three of them to a degree.  Two, with the exception of Zambrano, each of the pitchers on the list have spent the bulk of their careers pitching in environments that are averse to allowing home runs.  While Fenway is an environment that promotes doubles over home runs, it’s certainly known as an environment that gives up its share of long balls.  Lastly, no pitcher over the past nine seasons has put together an accumulative HR/FB rate under 7.0%.  That means that Buchholz’s 5.6% HR/FB rate will almost certainly increase in 2011.  Like I said above, there’s not enough data to say what his true HR/FB rate level really is, but it’s fair to guess his true talent level is around the league average of 10.5%.  If we were to regress his 2010 rate to the league norm, Buchholz would’ve given up 17 home runs, as opposed to nine.  Using linear weights as our guide, the additional eight home runs would have likely have added 14-18 runs on to his year-end total, depending on the situation.  Using the mid-point (16 runs), and adding the nine runs from the BABIP adjustment above, his adjusted ERA would’ve been 3.63.  Oddly enough, this is very close to the number FIP predicted, 3.61.  Obviously, my calculation is a just an estimate, but it shows that sometimes things aren’t always as good (or bad) as they appear.

So what does this mean for Buchholz?  A lot of that depends on how his peripherals shape up this season.  If they remain consistent with his 2010 rates, we’ll certainly see a significant regression in his ERA.  If he improves his K/9 rate (as his swinging strike rate suggests), lowers his walk rate, and maintains his outstanding ground ball percentage, he could (and I stress could) make a transition similar to the on Jon Lester had in 2009 and Justin Verlander’s had in 2007.  While I’m apprehensive about making such a prediction (or a comparison for that matter), his performance indicators show the possibility for some big improvements this season.  His swinging strike and contract rates, along with his fastball velocity, are particularly encouraging.  In 2010, it appears Buchholz was able to induce batters to swing-and-miss, but for some reason he tended to do so on strikes one and two, but not strike three.  As Buchholz mentioned in the Abraham piece, as he gains more experience, he’ll start accumulating more strikeouts.  I think there’s a lot of truth to that statement.

As a pitcher with five major league quality pitches (four seam fastball, two seam sinking fastball, curve ball, change-up, and slider), Buchholz certainly has a lot of potential.  Even though he appears to have been extraordinarily lucky last season, the future is still very bright.  I don’t see him overtaking Lester as the ace of the staff, but I can certainly see him as a very strong number two starter for years to come.