A Decade of ERA-xFIP: Is Clay Buchholz a Buy Low?

The other night was just another typical evening for Clay Buchholz. Five earned runs, including four early, caused frustrated fantasy owners to see 7.11 ERA, 1.74 WHIP in the box score. A quick Twitter search for Buchholz shows the bandwagon (if there was one) has emptied fast. Mass media is feeding the frenzy, especially with the Red Sox performing so poorly over the first few weeks of the season. Buchholz continues to say things about not getting breaks, such as “…ground balls when you want to get them hit at guys for double plays. It seems like the ball is finding a lot more holes right now.” But is he lying?

Below is a table of the top ten qualified starters sorted by ERA-xFIP. Larger ERA-xFIP values indicate that their actual performance seems to be worse than the peripherals involved in the xFIP calculation (K%, BB%, HR/FB%) would imply their results should be more in line with.

We see a few middling names on the list who have gotten off to bad starts. For example, Kyle Kendrick, Kyle Lohse, Mark Buehrle. While they have large splits, even assuming they are projected to pitch better going forward is probably not enough. Their version of “better” might not be satisfying enough for typical mixed leagues (although a few names on the list are probably nice targets in deeper or AL/NL-onlies). Toward the end of the top ten, we see two giant names from Chicago who haven’t had the greatest of starts this season, but should be expected to improve going forward. But one name sticks out like a sore thumb. At the top of the list, posting the best K%-BB% and second best xFIP on the board is Buchholz. The Boston righty has put up elite ratios to open the season (9th in all of baseball in xFIP) but with terrifying results (10th worst ERA among qualified SP).

So we can easily establish that Clay Buchholz has an enormous ERA-xFIP split. The easy gut reaction is to say Buchholz has gotten horribly unlucky, and his .401 BABIP should come down, especially for a guy whose hard-hit ball percentage is a little elevated, but still lower than Clayton Kershaw’s has been in the small sample size of 2015.

But maybe ERA-xFIP isn’t a good proxy for “luck” for Buchholz. Maybe he’s mentally weak. Maybe he is throwing well 4 out of 5 pitches, but his other 20% are just meatballs down the middle of the plate. Maybe it’s terrible pitch calling/framing/sequencing. Maybe he just is a terrible pitcher, and terrible pitchers are overly buoyed when it comes to DIPS-type metrics. Well, it’s quite easy to observe whether or not these terrible pitcher outliers exist. The next plot shows the frequency distribution of yearly ERA-xFIP among qualified starting pitchers over the entirety of the last decade, binned by 0.06 increments (772 data points, or approximately 80 pitchers per year). The two gray vertical lines highlight the lowest and highest deciles; that is, 80% of all seasons fall between these lines. In simplified parlance, points which fall towards the right side of the curve are “unlucky” (ERA worse than peripherals) and points on the left side of the curve are “lucky” (ERA better than peripherals).

We can see over the last decade that the median ERA-xFIP is slightly below 0.00 (technically, -0.15). The vast majority of these seasons are clustered around the origin in a nice peak, showcasing that over longer times, ERA tends towards xFIP. Tell me something we don’t know, right? What is perhaps more important for this discussion is how compressed this peak is. On the side of the curve we are interested in (right), only 10% of the seasons have an ERA-xFIP worse than +0.69, with the worst single season being owned by Ricky Nolasco’s +1.83 mark in 2009. Only 5 of the 772 data points are higher than +1.50, and only 33 (~4%) are above 1.00. Buchholz is now labeled as the red dot on the far left portion of the spectrum. With a +3.10 ERA-xFIP split, Buchholz’s start to 2015 (should it persist) would make him the biggest outlier over the last decade but an enormous sum. Perhaps he is the terrible outlier where the peripherals are deceiving and subjective analysis tells the real story, but if he was, he’d certainly be trailblazing ground utterly unheard of in this era of pitching.

What this result establishes is a feasible “upper bound” for an ERA-xFIP split over the course of a full season. There is undoubtedly evidence to envision why a pitcher’s ERA might not match his peripherals. He may give up exceedingly weak contact on fly balls, suppressing his HR/FB% (hello, Matt Cain). Alternatively, he may throw fastballs over the heart of the plate in an attempt to knock his BB% down. Maybe he’s just a giant headcase that melts down when the going gets tough. That said, these “hidden” anomalies which we have not quantified tend to be relatively small, even for the pitchers with the largest splits. Approximately 95% of pitchers post ERA-xFIP splits which are between -1.00 and +1.00 over the course of a full season. This range almost certainly narrows as we start clustering multiple years together. There is very little evidence suggesting that larger gaps (such as the one Buchholz owns) can remain once we start talking about 125 innings instead of 25.

The Boston media and fantasy sphere are filled with stories about how Clay Buchholz should be on the edge of being DFA’ed. As evidenced by his ownership rates, fantasy owners have abandoned ship, even though the righty has posted one of the best peripheral marks in the league. There is just no evidence to suggest his performance to date is sustainable (in a bad way) going forward. And it is tough to recall instances where you’ve seen a guy with his kind of rates floating around on the wire or who can be had for a song. Heck, I saw him go for David Phelps (!) in a 20-team dynasty league after this weeks’s game. Even if you think he can’t sustain his K%/SwStr% gains (i.e., his xFIP is somewhat of a mirage) and he will fall back to the middle, it seems almost a certainty that there’s plenty of profit to be had for patient investors, especially as the depth of one’s league increases. Savvy owners would pounce.