Apples to Apples

As we drag through the offseason, I’ve tackled topics that move further and further away from specific Cardinals. I promise that today is as zoomed out as I plan on getting this winter, and that I’ll refocus on minutiae soon. That said, I want to finish up my gaze into the abstract with one last discussion of WAR. As I mentioned last article, discussing WAR in a baseball article is a great way to get a lot of people mad without accomplishing much. You might, for example, provoke a former manager of the site into sniping at you on Twitter for your article’s crude attack on WAR in an article where you defend the very thing he derides you for attacking. Just a hypothetical example. Still, though, I can’t resist touching the stove- we’re going back in for one last pass. Today, I want to talk about why something like WAR is an ideal way to analyze baseball, and why questioning individual parts of the statistic isn’t an effective way to refute the more general idea of WAR. This isn’t going to be everyone’s cup of tea- but hopefully if nothing else it’s an interesting thought piece.

Let’s start with the proposition that we’d like to build a model that can describe how valuable baseball players are. This is worth questioning- not everyone wants to think of baseball that way. Describing how valuable players are isn’t necessary to enjoy baseball, or even to enjoy reading about baseball. My mom, for example, has been a huge baseball fan her whole life. When she comes to visit me, we usually go to see a game. I’m not sure she can break down WAR, and why would she need to to have a good time at a ballpark? Still, though, I probably wouldn’t want her to become GM of the Cardinals. The fun of talking about WAR is thinking like a GM- assigning value to players and thinking about how to spend scarce resources, or win as many games as possible, that kind of thing. Let’s stipulate that right away- describing how valuable baseball players are in relation to each other isn’t necessary to enjoy the sport, but it is necessary to do the job of a front office. When the team is deciding which of two players to spend resources on, ‘they’re both unique and incomparable’ isn’t an acceptable answer, so we’re going to need something.

Okay, so we’ve settled on needing a valuation framework. That’s a great first step. For a valuation framework to function, we need to express things in consistent units. What do I mean by this? Well, let’s take fruits for an example. If you’re trying to figure out which is going to provide more energy, you might look at the difference in calories between an apple and an orange. A calorie is a consistent unit here. You might say that an orange has 130 calories and an apple only 110. What are inconsistent units? Well, you might say an orange has 130 calories and an apple weighs four ounces. Those things, while true, aren’t comparable. To be able to compare them, you need some kind of medium of exchange. The parallels in baseball are clear. If one player hits three singles and another is fast, well, those aren’t on the same scale. We need some consistent units to be able to compare them.

If we’re looking for a consistent unit in baseball, runs are a natural fit. Every game is about scoring and preventing runs. The only way to win a baseball game is to outscore your opponent, of course. You can think of everything a baseball player does as creating or destroying some amount of runs. Turn a double play? You saved part of a run. Hit a home run? Pretty self-explanatory. Do a really sweet bat flip? Well, we can include that too- it’s just probably worth about zero runs. It’s pretty convenient that runs are the only unit of account in baseball. If you had to, say, outscore your opponent and also write a persuasive five-paragraph essay, things would be a little trickier. That essay doesn’t score any runs, but it’s still valuable. Those runs aren’t going to write your conclusion. That would complicate things. Instead though, we get to look on only one measure.

How do we convert individual plays to runs? Thankfully, smart baseball minds have focused on this question for more than a decade. The all-in-one offensive stat du jour, weighted runs created plus (wRC+), has runs created right in the name. It is itself an offshoot of wOBA, a statistic that takes the average run value of each plate outcome and gives players credit based on those run values. Figuring out the run value of defensive plays is a logical offshoot of figuring out the run value of offensive plays. Defensive Runs Saved, or DRS, has it right in the name. Baserunning? The same general principle applies. When we know many runs will be scored on average in one state, and then compare that to how many runs will be scored in a new state, the play that makes the game switch between the two states is worth the difference between the two values. The overarching importance of runs puts all of these different baseball acts on one single scale, which makes our job as analysts much easier.

It’s worth stopping to mention here that this feature isn’t inherent in all sports. In football, for example, points are still king, but they don’t arise in a linear fashion from actions. Pass blocking leads to some points. Having a good running back leads to some points. You can’t just have ten left tackles or ten halfbacks, though. In football, emergent properties of combining different types of offensive players produce yards, and the yards produce expected points. That’s why you’ll see a lot of quarterback value metrics or overall offense metrics, but the individual players get graded in more bespoke and qualitative ways. Baseball’s nice that way- every single hitter has the same ideal outcome, and it’s the same with every pitcher. Without that convenient fact, our job would be a lot harder.

This might sound reductive, but that’s basically all the information you need to know that a stat like WAR is an ideal stat to measure value in baseball. Players are all trying to do the same thing. They do the same thing in a finite number of ways, all of which produce outputs that can be expressed in the same units. That’s honestly it. Runs are the currency of baseball, so any value metric is necessarily going to be expressed in runs. Now, WAR has some other twists and turns to it. It’s expressed in wins rather than runs, for one. That said, there’s nothing too fancy about that- you can just figure out how many extra runs a team needs to score on average to win an extra game, and that gets you to the conversion between runs and wins. There’s replacement level, a clever little trick for appropriate scaling that I talked about last week. Positional adjustments let us compare across positions- it would be less analytically daunting to compare second basemen only to other second basemen, but positional adjustments let us transcend that limitation. At its core, though, the correct way to value the contributions of a baseball player HAS to be in runs. There’s simply no other unit of value we can use.

When you understand that, criticisms of WAR become questions of methodology rather than fundamental flaws. WAR, the general concept of it, is right. It just is. There’s no chance that counting up how many runs better or worse a player makes your team isn’t the relevant metric. Do you happen to think that pinch hit plate appearances have worse expected outcomes than regular plate appearances? Tweak WAR to adjust the baseline run values produced in those situations. Your output is still going to be in runs, though. Do you think that speed on the bases makes opposing pitchers perform worse? Quantify it. Add that value to baserunners. Subtract it from the runs produced by the hitters at the plate- if the guy on the bases generated the value, the guy at the plate didn’t. Still, though, your output is going to be in runs.

No serious sabermetrician would ever say that WAR the way it’s calculated today is unassailable. We haven’t quantified all the ways that players can create expected runs, and there’s value that baseball analysis hasn’t satisfactorily accounted for yet. Statcast blew the lid off of contemporary baseball analysis, and it will probably continue to do so in the future. Expressing player value in wins and runs, though, is here to stay. That’s the real power of WAR. Objections to WAR aren’t actually objections to the framework of WAR. They’re objections to the transformation we perform to turn actions on the field into runs. Those transformations won’t always be right, and over time the way we think about them might change. That’s small potatoes, though. The value of WAR has always been in the framework.

This distinction gets lost in a lot of discussions of whether WAR is a good statistic. People will raise some objection to the way WAR is calculated and use it to say it’s a bad or invalid metric. That’s just wrong, though. First of all, the vast majority of the time their objection is wrong. WAR not accounting for RBI’s is a feature, not a bug, and the fact that it doesn’t care about them is intended. Take my baserunning example above, though. It’s hard to believe that baserunners who require constant attention don’t distract pitchers from the job at hand. Just because WAR doesn’t account for that, though, doesn’t mean WAR shouldn’t be used. It just means that the current calculation of WAR hasn’t quite gotten down to the true elemental value of the runs a player is worth. It needs to be refined, not discarded. When people raise objections to WAR, what they’re mostly doing is suggesting improvements. As far as I’m concerned, the question of whether to measure players in runs and wins is settled. It was honestly never really a question- there was only ever one valid answer. Ten years from now, the way we calculate WAR and wRC+ and fielding runs will probably be different, because we’ll probably know more. The central unit of scorekeeping, though, is going to be the same.

If you take only one thing away from this article, from the whole series of articles I’ve written over the past month, let it be this. Don’t slavishly adhere to the exact way things are done now. Think about them, and if you feel like it, verify them. The exact way that things are done now isn’t what’s going to be important in ten years. Do, however, trust in the framework, in the concept. If you want to admire the majesty of baseball, do whatever makes you admire that majesty. Watch some nasty sliders. Obsess over how rarely Matt Carpenter grounds into double plays. Do whatever you want- beauty is in the eye of the beholder. If you want to measure value, though, there’s no escaping the framework of scoring runs and preventing your opponent from scoring them. Either your value metric reduces to that, or it’s wrong. End of story.