Several months back, I wrote about Carlos Carrasco’s potential as a member of the Indians’ rotation going forward. The basic takeaway: the guy can be really good, so long as he limits his homerun rate. He strikes guys out, doesn’t walk many, and seems to be one of the more effective options for the Indians rotation for the next several years. He just seems to let up too many home runs.
In fact, look at these strikeout and walk numbers among Tribe starters this season—usually pretty good gauges of a pitcher’s ultimate efficiency and success:
I know it’s only two starts, but Carrasco is striking out nearly as many batters as Justin Masterson, but walking half as many (the fewest on the team, at 2.03 per nine innings pitched), leaving his K/BB ratio at a healthy 3.33. You’ll notice that no other starter has a ratio above 2.00. That’s good stuff, and really useful.
And it’s not entirely unexpected stuff, either. For his AAA career—57 starts and 344 innings pitched—Carrasco has an 8.55K/9 rate and a 2.72 BB/9 rate, good for a 3.14 K/BB ratio. Those peripherals are really good, and that’s why Carrasco has such front-end potential.
But I’m always the bearer of at least some bad news—I do write about the Indians, you remember. So let’s get to it, and along the way we’ll touch on some of the important differences between some of the ERA estimators out there and why we should care about them.
First, there was the caveat that Carrasco would have to limit his home runs allowed to be a successful pitcher. Has he done that? Not really. So far, in 13.1 innings pitched, Carrasco has given up three home runs—good for a HR/9 rate of 2.03, or about double the MLB average. For reference, no other starter on the 2010 Indians has HR/9 rate above 1.58 (David Huff, ladies and gentlemen!). That’s not such good news, considering that the one thing Carrasco has to do to be successful is limit home runs.
Second, though, is the same old luck-story that I’ve harped on several times before in this space. Carrasco has stranded 92.6% of baserunners (MLB average ≈ 72%; Carrasco’s AAA rate was 72.5%) and has limited opponents’ batting average on balls in play (BABiP) to a .246 clip (MLB average ≈ .303; Carrasco’s AAA rate was .297). So Carrasco has gotten a bit lucky through his first two starts—not that there’s anything wrong with that. But luck is hard to replicate, and to make up for it, he might want to start limiting those home runs.
Which, I think, presents a good time to discuss a few ERA-estimators, what they mean, and why we should care about them.
The first question I often get when I talk to people about these numbers is why we would ever use them. After all, why would we want to “estimate” a number that we can simply calculate or look up? Why care about what some system “thinks” an ERA should be rather than knowing what it actually is.
And that’s all true. ERA is a great stat that tells us what happened in a very efficient way. But the reason that I like to look at some other numbers is to strip out some of the things a pitcher can’t control (luck, defense, randomness, and yes, some skill too) to give you an idea of what we can expect going forward.
So here is Carrasco’s ERA this season, along with a few estimators:
What’s the first thing that jumps out at you? To me, it’s that his FIP is almost two runs higher than his ERA, while the other estimators look right in line (though all higher, suggesting some good luck). You might remember that FIP counts the number of home runs a pitcher allows, whereas xFIP suggests that pitchers don’t have control over home run rates, so it substitutes a league-average number. tERA, on the other hand, sorts batted ball types (line drives, flyballs, groundballs, etc.) and assigns linear weights to each type to estimate what we would expect a pitcher’s ERA to be.
Why include all these numbers? Shouldn’t there be one number that works—one number that makes the best prediction? Well, in a perfect world, I think we would have a better feel for which numbers do the best job of predicting future performance, but for now, we just don’t know enough about what is skill and what isn’t, so I like looking at all the numbers and then use common sense to make some inferences.
And what I notice about Carrasco is that he has had a history of allowing home runs at a higher rate than most pitchers. So for him, I might look a bit more closely at FIP than at xFIP—simply because in his case, there seems to be some evidence to suggest that he won’t allow home runs at a league-average rate.
But if you believe that Carrasco’s HR rates will eventually normalize to league-average, then maybe you’ll want to put more stock in a number like xFIP. Or maybe you think that batted ball types are repeatable skills, and then you’d like to consider tERA as your predictor.
The point is that we have many tools at our disposal, and when choosing which tool to use, we have to think about the job at hand. We don’t use these tools to determine what a pitcher has done—ERA works perfectly well for that—but we use them to think about what a pitcher might do in the future. And which tool we use tells us as much about ourselves and what qualities we believe are repeatable and skill-based as it does about the pitcher we’re studying.
When it comes to Carrasco, he’s still a guy that I see letting up too many home runs, so I’ll favor the FIP model for now, meaning I believe his success will correlate pretty well with his HR rate. But if he does end up limiting those home runs? He’s a 3.50 ERA pitcher. A handy guy to have around. Exactly what we thought he could be.