WAR And Salsa

A lot of discussion started yesterday about WAR thanks a very thoughtful article on It’s About The Money, Stupid. Discussion followed in two article on The Book Blog (here and here) and then on Replacement Level Baseball Blog. I’ve mostly been ignoring the defense part of WAR for some time now. Ever since Colin Wyers showed how useless those metrics really are, I see no reason to include them.

This was originally titled The Problem With WAR because it irritates Tango because he thinks it should be “My Problem With WAR.” Except it’s not my problem. I changed it because we hadn’t had an article with salsa in the title and that’s just wrong.

A basic salsa recipe is essentially a framework for all other salsas. You can replace the tomatoes with peaches, blackberries, mango, apples or whatever the hell you want. You can take any basic salsa recipe and replace the tomatoes with avocados to get guacamole. Salsa isn’t the only recipe that can be easily adapted and turned into something entirely different. We don’t care about those because we’re going to use a basic salsa recipe to illustrate the problem with WAR. I’m just making this recipe up. It’s probably pretty close to a basic salsa recipe, but I’m too lazy to look it up.

Basic Salsa
4 tomatoes, chopped
1 medium onion, chopped
1 jalapeno, minced
1 t. fresh lime juice
1/4 C. chopped cilantro
1/2 t. minced garlic
1 t. salt
1 t. crached black pepper

With this recipe you can use anything you want to replace tomatoes to come up with your own salsa. Like a black bean and corn salsa? This recipe works. Want a blackberry and apple salsa? Go ahead. Is guacamole your favorite? You’ve got the recipe right there.

We can even write that recipe out as a formula if we wanted to. Pound cake and vinaigrette are the easies formulas to remember. All recipes are formulas. You want a certain percentage of tomatoes, onions, jalapenos and so on in the above recipe to equal salsa.

Imagine I write that recipe differently.

Basic Salsa
4 tomatoes, chopped
1 medium onion, chopped
1 jalapeno, minced
1 t. fresh lime juice
1/4 C. chopped cilantro
1/2 t. minced garlic
1 C. salt
1 t. crached black pepper

You know nothing at all about food. So you make this recipe as it is, taste it and realize it sucks. You’re not sure why because you know nothing about food. You instead make a peach salsa thinking maybe the tomatoes weren’t ripe and the peach salsa sucks too. Every salsa you make with that recipe is going to be horrible. Now the forumula is broken.

When one of the ingredients is wrong, the recipe fails. I don’t know if you’ve ever been around any chefs or even professional cooks, but you’ll sometimes hear them talk about a recipe that works or one that fails. In that position you create so many new food items and some of them are going to work and some will fail. This salsa fails because of one ingredient.

WAR is like the 2nd basic salsa recipe. There’s an ingredient in WAR that messes it all up. The inclusion of defense makes it unreliable at best. But defense is obviously part of a player’s value so it belongs in the framework of calculating WAR. There’s no doubt about that.

You could eliminate the salt altogether in the salsa recipe and it would be fixed, but adding just a bit of salt will bring out more flavors. Eliminating the salt doesn’t break the recipe. It’s not as good as it could be with a little salt, but it works. The same is true here with defense. Defense breaks WAR. If we eliminate it we know we’re not capturing the entire value of a player, but the recepe now works. It’s not perfect. Being able to accurately measure defense and include it would highlight much of what we’d miss without it, but its inclusion breaks the damn thing.

Should we keep making salsas with 1 C of salt and pretend that something isn’t wrong? Should we keep adding in defense when it is undoubtedly broken and pretend that something isn’t wrong? Yes, defense is part of the framework, but should it be part of it right now? If the best defensive metrics we have are UZR and Total Zone then I say no. If I’m left to invent my own fielding number then I say no. If I’m left to rely on scouting reports, which are often incorrect, I say no. If I’m left to include the fans scouting report, I say no. There is nothing about any fielding metric in existence today that makes me confident it improves WAR. Quite the opposite actually.

If someone tells me my salsa recipe is broken, I’m not obligated to fix it. I’m not selling it so perhaps I wouldn’t care. Maybe I don’t even care about my reputation in which case there would be no reason at all for me to fix it. But as long as I have published an awful recipe I can’t tell others to stop complaining about it. If people say “my problem with the salsa recipe” I’d tell them that it’s not their problem but the recipe’s problem. “The problem with that salsa” is an accurate way to describe it. It takes one ingredient to fuck up a recipe. I know salt belongs in the recipe as much as I know defense belongs in the calculation of WAR, but as long as the two result in making the end product worse, I see no reason at all to include it. When a better metric comes along I’d be happy to add it back in, but what’s the point of adding in useless numbers? Are we supposed to think that adding in these useless numbers makes the metric better somehow? That’s ridiculous. That’s as true as improving salsa by adding Peanut Butter Crunch to it.

WAR’s problem is that it includes defense. It’s not my problem with WAR. The WAR formula is as broken as the second salsa recipe. This doesn’t mean we eliminate WAR altogether. There’s no need to eliminate salsa. It just needs to be fixed. Until it is, adding in defense only makes the number less reliable.

I like WAR and I’ll continue to use it. It’s a great stat, but based on what we now know about defensive metrics, including it seems silly.


Quantcast