If there’s one truism in baseball that rings more true than all the others, it is probably that pitchers are just remarkably inconsistent. Even putting injuries aside, you have things like Ryan Vogelsong coming out of nowhere to solidify the Giants rotation or Javier Vazquez just randomly vacillating between one of the league’s best pitchers and serving as a batting practice machine. No matter whether you look at ERA or xFIP, the reality is that predicting how a given pitcher will do going forward is challenging.
So, it’s understandable why Mark Buehrle is such an attractive free agent this winter. He’s thrown 200+ innings in every season since 2001 and he’s produced remarkably similar results in each year since coming to the big leagues. Here are his ERA- for each year of his career:
His career average ERA- is 84, and he’s hit that mark almost on the nose in three of the last four years. With only a couple of seasons where he gave up runs at an average rate, he rarely varies from his overall norms. His 2005 and 2006 seasons were essentially outliers to both sides, but beyond that two year stretch, he’s been just a solidly above average starting pitcher with hardly any variation.
Of course, looking at his underlying metrics paints a somewhat different story. His career FIP- is 92, quite a bit worse than his career ERA-, and shows that he’s consistently posted lower ERAs than we would have expected based on his walk rate, strikeout rate, and home run rate. It’s even more extreme when you look at xFIP-, where his career mark is 97. That’s a substantial gap, especially considering Buehrle has thrown nearly 2,500 innings in his career.
There are other pitchers who have sustained similar gaps between their ERAs and their FIP/xFIPs over the last 10 years in significant numbers of innings; Jered Weaver, Matt Cain, Barry Zito, Tim Wakefield, Johan Santana, Jarrod Washburn, and Carlos Zambrano have all outperformed their peripherals for most of their careers. However, for most of these guys, the answer as to why is pretty simple – they’ve sustained low BABIP and/or low HR/FB rates, the main variables that aren’t measured in FIP (in the case of BABIP) or xFIP (which doesn’t include either). These guys have excelled in the areas that most pitchers have not been able to, and so their positive results are simply the byproduct of their skills not being captured entirely by the model.
However, Buehrle doesn’t really fit into that same mold. His career BABIP is .291, just barely lower than the league average of .294 since the start of the 2002 season. He’s not really a guy who prevents hits on balls in play in any kind of real significant way. Likewise, his 9.8% HR/FB rate is just slightly lower than the 10.3% average over the same time frame. While he’s been marginally better at these things than a normal starting pitcher, the magnitudes are very small, and don’t put him in the same category of guys like Weaver or Cain. In reality, his BABIP and HR/FB rates explain hardly any of the difference between his ERA and his FIP or xFIP.
So, if he’s not preventing hits and he’s not preventing home runs, then he must be stranding a ton of baserunners, right? Roy Oswalt, for instance, has kept his ERA down by posting a 76.1% LOB% over the last decade, the fourth highest mark in baseball for pitchers with at least 1,000 innings since 2002. And, after all, Buehrle is very good at holding on runners, so it would make sense that he’d strand more than his fair share on base.
Except, this doesn’t really fit with the numbers either, as his career LOB% is 72.3%, a rather pedestrian mark that is just slightly above the league average. In fact, if we use the formula that Dave Studemund provided to predict LOB% from a pitcher’s xFIP – 86%-(.033*xFIP) – than we can see Buehrle’s expected LOB% (72.1%) is almost exactly in line with his actual mark. Buehrle is stranding basically the right amount of runners given his actual walk rates, strikeout rates, and ground-ball rates.
So, uhh, what’s left? He’s not preventing hits. He’s not preventing home runs. He’s not stranding runners. How is Mark Buehrle keeping so many runs off the board?
The answer might be that he’s not, and a real part of the explanation for Buehrle’s gap between his ERA and his FIP/xFIP is actually a bias in how ERA is calculated. As you know, Buehrle is a ground-ball pitcher, and pitchers who put their infielders to work see a larger share of their balls in play result in errors. Errors result in unearned runs, and unearned runs don’t count against a pitcher’s ERA.
In fact, if we look at Buehrle’s career, 10.1% of all the runs Buehrle has allowed have been labeled as unearned. For starting pitchers since 2002 with 1,000+ innings pitched, that’s the ninth highest ratio of unearned runs in baseball. Some of the pitchers ahead of him include Brandon Webb, Felix Hernandez, and Derek Lowe, which illustrates the point of how ground-ball pitchers tend to have ERAs that are driven down because many of the runs they actually do allow are counted as unearned.
For that group of starters, only 7.8% of their total runs allowed were unearned, and if we adjust Buehrle’s earned-run total to reflect that average, he’d have been charged with an additional 26 runs, raising his career ERA from 3.83 to 3.92. That doesn’t bring his ERA completely in line with his FIP or xFIP, but it cuts the gap by 29%, and shows that a significant part of the perceived gap between Buehrle’s results and his peripherals is really just due to the earned run bias that runs in favor of ground-ball pitchers.
This doesn’t invalidate Buehrle as a good pitcher, as his durability and consistency are still intact, and even if you judge him by something like FIP-, he still comes out as one of the better pitchers in the league. However, it’s perhaps a different type of good pitcher than you might think by just looking at ERA.
If we sort all starting pitchers with at least 1,000 innings pitched since 2000, Buehrle’s 3.83 ERA puts him as a peer of Ben Sheets (3.79), Zack Greinke (3.82), Josh Beckett (3.84), and Andy Pettitte (3.86). If you ignore the subjective earned runs tag and sort by runs allowed per nine innings, however, Buehrle’s 4.26 RA/9 makes him a peer of Al Leiter (4.23), Barry Zito (4.24), Randy Wolf (4.29), and Jarrod Washburn (4.29).
There’s nothing wrong with the second group of pitchers, but I think we’d all agree that they aren’t as good as the first group. If a team thinks they’re getting more of the former than the latter, they may very well end up disappointed.