ERA is commonly used as an indicator of pitcher performance, but it’s dependent on several things—pitcher performance (it has to be there; he is pitching), defense (better defenses make more outs), randomness (balls drop in, good pitches get hit hard, bad ones don’t, etc.), and other pitchers’ performance (allowing, or not allowing, inherited runners to score). Obviously, not all of those measure pitching performance. Essentially, the pitcher only has control over a play when he throws the ball. Once the bat swings through the zone, the only thing he can usually do is watch. ERA, however, adds in all of that stuff after the ball is hit and evaluates the pitcher, though what happens after is up to randomness and the defense.
But a pitcher can make better pitches to get weakly hit balls!!!! Not really. Most pitchers over the course of their careers will have BABiPs (Batting Average on Balls in Play) around .290-.310. Now, those are just averages, and like a home run hitter who averages 32 homers a season, he doesn’t actually hit 32 every season. Sometimes, they can have seasonal averages of .250 or .350 due to randomness, and over the course of many years, that will even out as they’ll have average seasons and opposite outlier seasons. But major-league quality pitchers will stay within that range for a career (though knuckleballers are a notable exception), and if they don’t, it’s pretty much always high and an indication they don’t belong in the majors. If they have an abnormally high BABiP (hey, Joba’s is .341), it will likely come down the next season because, as abnormally high, it isn’t likely to happen the next season, though it is possible as randomness can act like consistency occasionally but will show it’s true colors eventually. Don’t believe me? Roy Halladay, the “best pitcher in baseball”, has a .299 career BABiP, and CC Sabathia’s is .295. Go check some more. I’ll wait.
Are we back? The next question is what can pitchers actually affect, and the answer is strikeouts, walks, and types of batted balls. Strikeouts reflect the quality of stuff a pitcher has, and walks reflect control and command. A pitcher can affect batted balls through their location (high pitches usually end up as flyballs and low ones as groudballs), stuff (sinkers get a lot of groundballs, for example), and movement (little movement ends up as line drives). The quality of their stuff and control along with what type of pitcher they are then reflected in home runs as stuff and control are found to be lacking (more home runs means this may happen more often for some pitchers, but there is some uncertainty over the 10.6% HR/FB theory) and the defense cannot affect them (yes, they can rob home runs, but they don’t do it very often in relation to the number of home runs hit).What FIP, and other DIPS statistics, do is simply measure these things together to isolate only the pitcher’s performance.
That last sentence reveals the key—isolate only the pitcher’s performance. FIP, and other DIPS statistics, really do measure the pitcher’s performance, but the difference between them and ERA is that you cannot see the direct correlation between them and the number of runs scored. The problem is that FIP does not directly. What FIP really does is state how many of those runs the pitcher is really responsible for by taking out the effect of defense and randomness. Therefore, it does measure pitcher performance, and it does so without the outside effects that ERA is dependent on.
Now, here’s where the statistic has come under fire. It also has predictive ability. Essentially, it’s a two-for-one statistic. It measures pitcher performance, and because it does isolate it so much better, it will be more predictive of future performance. Because ERA is wildly unpredictable and fans are more used to ERA, sabermetricians usually say something like, “that ERA is bound to come way back next season.” Because of the abstract nature of FIP (ERA is something you can easily see, whereas FIP tries to parcel out certain things which also results in the multitude of different DIPS), people think it means what the pitcher should have done or what the player might do. But it isn’t that way. It is the pitcher’s performance and only the pitcher’s performance, but it also has some predictive power because stats such as strikeouts and walks don’t fluctuate so much year-to-year (because, well, a pitcher’s ability does not vary that wildly from year-to-year). Defense and randomness can vary, which causes ERA to be unpredictive.
ERA isn’t all that bad of a team statistic, though. Sabathia has thrown 209 innings so far, but the Yankees, as a team, have thrown 1,253 innings. That’s almost 6 Sabathia seasons, and the more innings someone, or a team, throws, the closer ERA comes to describing run prevention, though not Yankee pitcher quality. Remember, that also includes defense, but the randomness has evened much, much, much more than it would have 1,000 innings ago. ERA doesn’t work well for an individual pitcher’s season because that randomness and defense haven’t evened out, but over the course of a career, FIP and ERA begin to correlate quite well. Roy Halladay’s career ERA is 3.33, and his career FIP is 3.43. CC Sabathia’s career ERA is 3.57, and his career FIP is 3.59. Amazing, isn’t it? This is what we call good science. We would expect FIP to correlate well because it was designed to measure how good a pitcher is, which comes down to how many runs they give up. It’s just better year-to-year because of its isolation properties, but ERA is fine to use over an entire career of more than 5 or 6 years.
So how does this work for Joba? This season, he is striking out 9.15 batters per nine innings while walking 3.10. K/BB ratios of 2 are average. Joba’s (2.95) is well above average. That, along with giving up home runs at a 7.6% clip, suggests something hinky going on to give him a 4.71 ERA when his FIP, responding to the K/BB ratio and home run rate, is 3.21. The hinky thing is randomness. Joba’s BABiP is .341, and because of that, his LOB% (Left On-Base %) is low at 66.2%. This bit of unfortunate randomness has caused Joba’s ERA to skyrocket over his FIP. Because relievers pitch in much fewer innings than starters, their ERAs will vary more wildly from year-to-year, and in-season fluctuations are greater as well (one home run barely over a fence or one ball just over Jeter’s head with a man on second affect a reliever’s ERA more than a starter’s). Joba is likely to have fewer runs score during his innings next season. It won’t be because he’ll be “better” but because randomness won’t hurt him so much, in all likelihood.
But he just can’t handle the pressure, right? Again, that’s wrong. He walks into all of those eighth innings, whether you or Marchand like it or not, and performs. Otherwise, he wouldn’t have such a great K/BB ratio and give up few home runs. He CAN focus and perform in those situations. The stats show that. So, why does he blow saves? Runs happen to everyone. That’s why we have ERAs and FIPs. Pitchers give up runs. To get a blown save, you have to have the right combination of a lead and then runs scored against that outnumber the lead. Even with an FIP of 3.21, one would expect Joba to give up a run every 3 appearances, but because we know pitchers don’t just give up one run at a time, sometimes he gives up more than one. Sometimes, the situations converge and screw the team. It happens. Don’t let a few games here and there destroy your perception of a player. We’re much more likely to remember the bad than the good (There’s a saying that you know god is doing his job when you don’t know he’s there).
But Joba isn’t as good as he used to be! Even your FIP says that he’s worse! That’s true as well. He’s striking out fewer hitters and allowing a few more walks. His stuff may have also diminished a bit as his fastball only clocks 95 instead of the 97 in 2007, though his curve and slider are as hard as ever. Yes, he’s gotten worse, but he’s still an excellent option out of the bullpen. He is a good reliever, 16th best in the AL. But I realize that doesn’t live up to expectations.
What I hope to have done is show you that FIP, indeed, measures player performances. When deciding how much faith to put in a pitcher, look at his FIP for that season, not his ERA. It isolates his performance, and his performance only. I realize it’s not the easiest thing in the world to reconcile that ERA isn’t what you thought and were taught it was. But it is possible. I grew up with ERA, just like everyone else, and I had an issue with it. I still do sometimes. I just try to keep an open mind and question myself repeatedly if a good argument is presented. It feels weird at first, but it’ll be okay, I promise. And it will make our discussions so much better.