This morning, we rolled out several new pitching metrics, and I outlined their uses in an overly long introductory post. If you haven’t read those posts, go do so now, as they essentially set the table for this post.
As we noted this morning, our goal in introducing Fielding Dependent Pitching is to help quantify some of the missing aspects of run prevention that are not captured in Fielding Independent Pitching. However, you also have undoubtedly noted that we have not changed how we are calculating pitcher WAR, and FDP is not included in those calculations.
I promise that this is not because we are stubborn and refuse to admit that pitchers have some control over hits on balls in play. In actuality, the decision to leave FDP out of pitcher WAR for now was actually a difficult one, and was not our original intention when we developed FDP. The genesis of creating metrics to measure the wins added on balls in play and runner stranding was an effort to improve the way we calculate WAR, and we planned on modifying WAR to account for both FIP and FDP. Trust me, we don’t like some of the weird-looking results that a FIP-based pitcher WAR produces any more than you do.
However, when it came to actually modifying the formula, we came to the same crossroads that caused us to choose a FIP-based WAR when we created our initial implementation several years ago, and that was a trade-off between being more comprehensive at the cost of making an arbitrary decision about the level of defensive support a pitcher received. For whatever flaws FIP-based WAR has, it is strikingly good at being transparent in exactly what it is measuring and not measuring, and making no claims beyond what it knows it can support with data. Because walks, strikeouts, and home runs only really involve two parties — or three, if you count the umpire — it is easy to assign full responsibility for the outcome of these events to the pitcher. FIP knows what each of these events are worth, and judges a pitcher solely on the things that we can say were the direct result of their actions.
When you introduce balls in play into the equation, those blacks and whites become very gray. How much of a pitcher’s BABIP is he responsible for, and how much is the result of his defenders? We honestly don’t know.
And so, in not knowing, any decision we made now to add some portion of FDP into pitcher WAR would have required an arbitrary decision. In reality, the things that make up FDP are messy, acting more like football or basketball plays with multiple variables interacting together, and much less like the kinds of baseball plays that make it fairly easy to say “this guy did that, and he deserves this much credit for it.” Even if we decided that a pitcher should get half credit for his BABIP — my initial position, in the interest of full disclosure — what do we do with strand rates that are highly affected by BABIP distributions?
For instance, let’s look at Jordan Zimmermann‘s line this season. His 2.63 ERA is nearly a full run better than his 3.43 FIP, and while his .280 BABIP is a little below the league average, only +0.6 wins of his FDP come from BIP-wins. Most of the difference between his ERA and his FIP have come from runner stranding.
Bases Empty: .280/.324/.420, .323 wOBA
Men On Base: .192/.240/.305, .241 wOBA
RISP: .154/.201/.225, .180 wOBA
If those splits were the result of a drastic improvement in his FIP, we would probably want to give Zimmerman nearly all of the credit for his LOB-wins. After all, pitching better with men on base is clearly more valuable than melting down and letting everyone score, and a pitcher should be rewarded for his ability to buckle down under pressure.
However, we can’t say that the results are completely due to Zimmerman buckling down in those situations.
Bases Empty: 3.59 FIP, .330 BABIP
Men On Base: 3.27 FIP, .215 BABIP
RISP: 2.61 FIP, .185 BABIP
Yes, he’s pitched better with men on base, but his rate of hits on balls in play is the primary driving force behind his strand rate. What amount of credit should Zimmerman get for these results? Should he get more or less credit than Johnny Cueto, who has also posted an extremely high strand rate, but has done it without significant BABIP splits?
I think we could probably all come up with a number that we could justify for each pitcher, and maybe all those numbers would even be pretty similar, but I have yet to see a methodology that would make that pick anything other than arbitrary. Our strong hope is that a methodology will be discovered soon, and advances in our understanding of how to split credit between pitchers and fielders will give us a systematic way to incorporate some percentage of FDP into pitcher WAR.
What that percentage should be, I don’t think we really know yet, and rather than impose our best guess onto the calculations and hope that we’re in the ballpark of reality, we’ve decided to keep WAR transparent about what it is and is not measuring, and display all of the various components of FDP so that you can can make whatever adjustments you feel are warranted. Essentially, we have decided that it is better to provide you with as much information as possible in a way that is free of our personal opinions on what percentage of hit prevention is pitching or fielding.
Our decision to leave FDP out of WAR means that it is not comprehensive in measuring all aspects of run prevention, but we think it is better — for now — to leave it based solely on FIP until more research produces a consensus, systematic way to reward pitchers for some aspect of FDP that does not require us to simply pick an arbitrary number and force it upon you. And, hopefully, by displaying all these components separately, we’re providing tools that could be useful in researching the various aspects of run prevention, and may even aid in the creation of a logical way to give pitcher’s credit for some portion their FDP in WAR.
Our hope is that pitcher WAR will not always face these same hurdles, but we feel like it is better to be up front about what kinds of compromises would have to be made in order to attempt to be more thorough than it is to simply force decisions that couldn’t be defended on an empirical level. For now, I’d encourage you to look at pitcher WAR as a baseline for what we know a pitcher was responsible for, and then make your own decisions about how much you want to adjust for each aspect of FDP. Personally, I’m likely to give more credit to LOB-wins than BIP-wins, but I don’t believe I have enough data to defend a challenge of that opinion. So, for now, WAR is still based on FIP, but we’ve attempted to give you the tools to make rational adjustments where you see fit.
If you want to simply evaluate a pitcher on his runs allowed, you can now do that on FanGraphs. If you want to blend FIP and Runs Allowed evenly, simply cut FDP by half and add it to his current WAR. If you want to give more credit for runner stranding and less for hit prevention, you now have a better starting place than you did yesterday. Our hope is that these tools empower you to be more comprehensive in your own evaluation of a pitcher, however you deem it best to do that.