Let's talk about run differential

So Rob Neyer ran a piece over at Fox Sports yesterday about early season run differentials. We need to talk about it.

I recommend reading the whole thing first so you can get the tone of the piece. One of the big "aha!" parts is that the Reds and Brewers have been about equally effective teams so far because of their similar run differentials. Here he is talking about the strong Brewer start:

How? We keep talking about their pitching! Except they've given up 11 more runs than the Cardinals, 13 more than the Reds. It's not their pitching. Not really. It's really as simple as this: the Milwaukee Brewers are 12-3 in games decided by one or two runs. Despite what you might hear on the radio, a team's record in close games is determined largely by luck, along with a small dollop of relief pitching.

Run differential is often a very good way to find out if a team is overperforming or underperforming over a period. The general idea is that players are not really all too concerned with the game situation when they are playing. A pitcher pitching with a 1-run lead isn't going to try harder to not give up a run than a pitcher down 3 runs. A hitter batting down 4 or up 3 isn't trying to make an out in one situation or the other. Over a longer period, the logic goes, this sort of thing tends to even out. A team getting unlucky and bunching up their runs in wins but going cold in other games, or in contrast spreading out runs evenly on defense, is probably better than their record indicates.

This is generally good logic. It's also our rationale for the context-neutral hitting and fielding statistics that build WAR. But it breaks down a bit when we're talking about a team winning mostly based on a great back-end of a bullpen like the Brewers have so far.

Let's say a team's 5th or 6th best reliever had a FIP and ERA of around 4 through a month and a half of the season. That reliever is probably going to have a fairly low average leverage index, which is a measure of the importance of a game situation in which the reliever is used. But that reliever is probably going to have to pitch in some high-leverage situations from time to time-- when other relievers are tired or when games go to extra innings, for example. If that reliever happened to be pitching well in the high-leverage situations, and not well in the low-leverage situations, that would be some pretty good evidence that the team is not as good as its record. But if it's Zach Duke, and he's been stellar in mop-up duty as well as extra innings, your team might just be playing really well.

Overall, the Brewers have both had pitchers throwing extremely well overall and have also done a pretty good job of getting those relievers high-leverage innings. An average starter's inning has a leverage index of 1, and a leverage index of 2 is about twice as important as average, and .5 is about half as important. So far, here are the Brewer LI when they have entered the game. They also haven't been too concerned about punting a couple of games with Wei-Chung Wang on the mound.

Here's a good argument for why the Brewers will not keep playing .800 ball in 1 run games and .600 ball overall: their relievers are not as good as they have pitched so far, with their four best relievers all checking in with a FIP of around 2. Those relievers are probably going to fall off as the year goes on. But here's a not-so-great argument-- the fact that they have given up more runs than the Reds means their pitching staff has been worse so far, and they've mostly gotten lucky to win so many close games. That obscures what's been happening. Given their usage and performance so far, the great 1 run game record and fantastic record overall is pretty much what we should have expected.

Neyer does end by acknowledging that the Brewers have a good enough head start on the division that they do need to be taken seriously. But...

None of this means the Brewers aren't a good baseball team. Right now, they're ... actually, I'm not convinced they're good. Take away these unsustainable things -- by the way, I haven't even mentioned Carlos Gomez and Mark Reynolds yet -- and they're just fair. The Brewers have a lovely head start on winning 90-some games. But they'll need a new formula for the rest of the season, probably something that includes Jean Segura, Aramis Ramirez and Khris Davis hitting a lot better than they have. Because there's very little help on the way.

Doesn't that sound more like a reason for optimism? The Brewers have gotten to this impressive division lead with some very underwhelming offensive performances (and Neyer can say what he wants but I think Carlos Gomez is a bit more than "fair" at this point and more of "really good"). But if we're going to just bank that the bullpen will be getting worse just because it's been good, can we also bank that those hitters will get better because they have been pretty bad?