Welcome back to Advanced Metrics and How to Use Them, your very brief introduction to some of the statistics that help us judge a player’s overall value. Last week, we delved into batting, examining some of the problems with traditional offensive statistics and looking at some good alternatives, like OPS and wRC+. Now it’s time to tackle pitching.
Before we get into the stats, let’s remember what we’re trying to get at: true talent. We’re trying to understand how good a player is apart from luck, chance, and the baseball gods in order to determine how he’ll perform in the future. These are the kinds of things we want to look at when we’re playing General Manager and want to see what’s really there in all of those free agents and trade targets. So, with that in mind, let’s begin!
The Ugly: Wins (W)
Let’s start with the bane of the pitching analysts existence: the win. In each and every baseball game, one pitcher is credited with the win, and another is credited with the loss. Presumably the guy with the win was the better pitcher, right? Why can’t we analyze based on this stat?
The Problem: Pitcher A goes out and throws 9 innings of one-run ball, striking out 17, walking none, and allowing just one hit: a solo home run to lead off the game. Pitcher B goes out and barely makes it through 5 innings, strikes out just 3, walks 4, and gives up 10 hits, and 6 runs. But Pitcher A’s offense is held scoreless, and he takes the loss because of that solo home run while Pitcher B’s offense, puts up a ten spot in the first five innings, so he gets the win. Seems kind of backwards, doesn’t it? If you think that doesn’t actually happen, look at Shelby Miller in 2015, who won just 6 games in spite of his solid 3.02 ERA. On the other side, Rubby de la Rosa won 14 games with an ERA of 4.67. That kind of seems like a problem to me.
How to Use It: Use it when you want to point out how cool it is to win twenty games in a season and how impressive that is. Don’t use it if you’re trying to evaluate how good a player is or has been.
The Bad: Earned Run Average (ERA)
ERA can’t be all bad, because I just used it, right? If we want to know how good a pitcher is, wouldn’t it make sense to look at how many runs he allows on average? I mean, that kind of is his job – to prevent runs. So what’s the problem with ERA?
The Problem: Much like batting average, ERA isn’t inherently bad; it just doesn’t do what we want it do. Remember, we’re trying to figure out how a pitcher will perform in the future apart from luck, defensive advantages, and random coincidence. ERA only tells us how well a pitcher has performed in the past, inclusive of all those things. Because of this, it frequently can fluctuate from year to year.
For instance, Chris Sale is one of the best pitchers in the American League (and all of baseball, for that matter), but this year he put up an unusually high ERA of 3.41. It’s not bad, but certainly high compared to his career average of 2.91 and his 2014 level of 2.17. So is he declining at 26? A quick glance at BABIP tells us that in 2015, his BABIP was .323 – almost thirty points higher than his career average. So that high ERA is probably a result of bad luck or bad defense or both.
How to Use It: ERA is great if you want to evaluate what already happened, but a pretty poor tool if you’re trying to figure out what will happen in the future. If you’re trying to see past the luck and defense to get a good view of what the pitcher – and only the pitcher – did, then look elsewhere.
The Good: Fielding Independent Pitching (FIP)
Alright, so if ERA doesn’t give us an accurate view of what we want to know, then why is FIP any better? For that matter, what is FIP? What does it mean that it’s “fielding independent”?
Why It’s Good: In order to understand FIP, we have to understand DIPS (defense independent pitching statistics) and the theory of pitching built around it. If you want a great intro, check out Voros McCracken’s piece on Baseball Prospectus, but here’s a brief explanation: pitchers can’t control what happens on balls that are put in play (at least not very well), but they can control a number of other things, including strikeouts, walks, and home runs.
To put this in practical terms, what’s the difference between a ground ball between the first and second baseman, and a ground ball right to the shortstop? Nothing, except he gets credit for an out on one and blamed for a hit on the other. To make things more complicated, how do we tell how much having Jason Heyward in right instead of Carlos Beltran affects pitching success? The answer is that it’s really hard, and maybe impossible.
Given the fact that we want to separate a pitcher’s performance from luck and his defense, what do we have left to work with? Strikeouts, walks, hit by pitches, and home runs. It’s hard to argue that the pitcher doesn’t have pretty much absolute control over those four things, and so FIP cuts out luck, pares away the defense, and plugs those four factors into a formula before multiplying them by a constant to put the whole thing on an ERA scale. The result is a familiar number that attempts to get at the core of how well a pitcher has pitched, not simply how good his results have been.
How to Use It: When trying to determine what might happen in the future, FIP is a much more accurate tool than ERA. It’s useful when you want to examine or compare a hurler’s raw pitching prowess. So if you’re interested in seeing past luck and a team’s defensive prowess, then go straight to FIP for your answers.
Other Useful Metrics: BABIP, K/9, BB/9, and GB%
BABIP: As with offense, BABIP can be a useful measure of luck. For instance, in April of 2015, Clayton Kershaw had an ERA of 3.73, which was unusual, but his BABIP for the month was also up at .378, indicating that he was suffering from a lot of bad luck, not a loss of ability. If you see incredibly high or incredibly low numbers, check the BABIP — you may be looking at luck.
K/9: Since pitchers can’t really control what happens to a ball once it enters play, it’s in their best interest to not let it enter play. In short, the more batters a pitcher strikes out, the better, and K/9 gives us great insight into that. Look for a high K/9 to correlate with a consistently good pitcher, and a low K/9 with a consistently worse pitcher.
BB/9: BB/9 is just the opposite side of K/9, providing the same info for walks. This time, you’re looking for a low number for a good pitcher. As you probably already imagined, striking out lots and walking just a few is a large part of what makes a pitcher valuable.
GB%: If you aren’t totally convinced by the defense independent pitching stats theory, then ground ball percentage is a great tool. In short, balls hit on the ground lead to fewer hits, and fewer extra base hits, than line drives and fly balls. If it seems like a pitcher is defying what FIP and other DIPS stats say he should be doing, it’s probably because he has a high ground ball percentage and is, in fact, controlling what his balls in play are doing. Taken with K/9 and BB/9, it can build a pretty complete profile of how good a pitcher is overall.
Once again, if you’re interested in exploring this stuff more, I’d encourage you to check out Fangraphs. In addition to wonderful articles and lots of player data, they have a phenomenal glossary that offers both simple definitions and in depth looks at a whole host of useful pitching stats.
So there’s the primer on how I like to look at pitching and what advanced metrics I find most useful. Come back next week and we’ll dive into fielding and how best to measure the value of defense.