ERA+ is a funny, messy statistic. Unlike OPS+, it doesn’t tell us how much better a player is than the league: it instead tells us how much worse the league is than the player. This ignores our intuition and causes unnecessary confusion. Furthermore, it makes it harder to use as a tool for direct comparison: someone with a 200 ERA+ is not twice as good as someone with a 150 ERA+, while that would effectively be true with OPS+ (and similar offensive statistics.)
To fix this, we introduce a minus sign. ERA- is the proper league- and park-adjusted statistic that is analogous to OPS+ and their ilk. As before, I won’t dive into the precise calculation because it’s not that interesting. Instead, let’s make sure we understand how it works.
Because ERA is a statistic that we want to be lower, ERA- follows the convention so that the lower it is, the better. The league average is 100, and an ERA- of 50 means the player’s ERA is 50% better than league average. Armed with that improved knowledge, let’s properly look at how Bob Gibson and Dwight Gooden compare; we’ll also look back at Zack Britton’s 2016 season.
Best ERA
In 1968, Bob Gibson’s ERA was 1.12, giving him an ERA- of 38. So, he was 62% better than the league. In 1985, Dwight Gooden’s 1.85 ERA was good for an ERA- of 44, putting him at 56% better than the league. These ERAs are actually quite comparable in the context of their seasons, although Gibson remains on top.
Relief
Zack Britton had an 803 ERA+ in 2016. How helpful is that as a statistic? Not very. It seems rather absurd compared to other values we saw. Instead, let’s ground it by noticing his ERA- was 13.1Note that 100/803 is about 0.125, which rounds to 0.13 or 13%. That’s the equivalence between ERA+ and ERA-, although the numbers don’t quite work out because Baseball Reference and Fan Graphs are using slightly different numbers. That’s still properly fantastic!
Derivative Statistics
We can compute FIP- and xFIP- in analogous ways, allowing us the best of both worlds: a less biased base statistic to compare against, but put on a scale that is contextualized agains the league environment at the time. In terms of raw ability and potential, we’ve arrived at the key statistics to use when comparing pitchers.2Sadly, Baseball Reference doesn’t use these “-” statistics; they use ERA+. That confusion (and my assumption that ERA+ was exactly analogous to OPS+) led me to rewriting major parts of the ERA+ post, and this post, to correct my mistakes. Luckily I realized all this before any of this was published!
While we still have a couple more statistics to explore — WAR, which we’ve seen before, and a statistic somewhat analogous to WPA for pitchers — those two represent value in different, more esoteric ways.
Due to FanGraphs being the only good place to get these “-” statistics, it’s harder for me to find fun pieces of trivia to share. So, we’ll leave the discussion here and revisit trivia tomorrow.
Continue to Day 19 – Leverage Index
- 1Note that 100/803 is about 0.125, which rounds to 0.13 or 13%. That’s the equivalence between ERA+ and ERA-, although the numbers don’t quite work out because Baseball Reference and Fan Graphs are using slightly different numbers.
- 2Sadly, Baseball Reference doesn’t use these “-” statistics; they use ERA+. That confusion (and my assumption that ERA+ was exactly analogous to OPS+) led me to rewriting major parts of the ERA+ post, and this post, to correct my mistakes. Luckily I realized all this before any of this was published!