The thing about statistics that I hope is really obvious by now is that each one measures something very specific. The factors that go into the calculation of a statistic determine what conclusions you can draw from it, which is why, in order to be confident in your analysis, you have to make sure you know what you’re measuring.
That’s why it also helps to be able to look at a few different statistics. If you’re trying to examine something as nebulous as “pitching performance,” it helps to have several angles on it. And if you’re measuring essentially the same thing in slightly different ways, the agreement or disagreement of your various statistics can tell you a lot about the pitcher in question.
Last week we talked about the outside factors that influence pitching performance, and the most basic way we have of getting rid of those, which results in Fielding Independent Pitching (FIP) and its counterpart, Expected FIP or xFIP. SIERA, one of the stats we’ll talk about today, is marginally better at predicting ERA than xFIP, but those two are the best we have.
So while some numbers won’t be as predictive as others, they incorporate more information, and give us yet another way to look at a pitcher’s abilities. tERA and SIERA reintroduce batted-ball data, which we’ve talked about before as being able to give us some insight into a batter’s skills. Well, the contact type that a pitcher induces can also give us some pretty good guesses as to future performance.
True Earned Runs Allowed. tERA follows in the footsteps of FIP, xFIP, wOBA, and all those other awesome weighted stats. It just incorporates more information, for a somewhat more descriptive picture of what a pitcher did.
When we last discussed batted-ball data, we talked about the values of the various contact types. Line drives are most likely to fall in for a hit. Fly balls can be good or bad – long ones are dangerous, especially depending on the park, but popups are much easier to handle. Ground balls most commonly get turned into an out.
tERA takes all of these and weights them according to their run expectancy (remember that one?), park-adjusts each of the components where necessary, and spits out a number that’s on a comparable scale to ERA.
tERA is based on a slightly different stat, one that you can still see on some other sites, called True Runs Allowed or tRA. Some people prefer this version, as it eliminates the subjectivity involved in the hit/error distinction. But it also means that the final number doesn’t compare well with ERA. FanGraphs uses tERA; StatCorner uses tRA, if you’d like to take a peek.
Skill-Interactive Earned Runs Allowed. Remember last week, when I said that xFIP was the best-correlated with future ERA? I lied. SIERA is actually a tiny bit better at predicting the future, and it does this by being freakishly complicated.
If you want to get into the nitty-gritty of how SIERA works, FanGraphs has a multi-part primer for you. The short version is that SIERA improves on tERA by examining trends within each of the subcategories that tERA takes into account.
It turns out that the effects of things like walks, strikeouts, and ground balls can increase or decrease disproportionately based on their frequency. For example, the individual walks of a low-walk pitcher matter even less, per walk, than those of a high-walk pitcher. The low-walk pitcher limits additional baserunners, making one walk even less likely to turn into a run.
SIERA spots these trends and takes them into account in its calculations. But that doesn’t mean it makes sense to abandon our other numbers in favor of it. Looking at all four of the statistics we’ve gone over so far will give you the most complete picture of a pitcher – the variations can tell you a lot about a player’s strengths and weaknesses. Next week we’ll take a practical look at these four statistics and how they play out for a few different pitchers.
Questions/corrections/math? Take it to the comments!
