ERA: better, but still noisy
Two pitchers give up nearly identical contact all year. One has a 3.20 ERA, the other 4.60. How?
ERA — earned runs per nine innings — is the most respectable traditional stat. It at least measures runs, which is what pitching is about. But three big contaminants sneak in.
The three contaminants
Defense: a shortstop with great range turns hits into outs; a bad one does the opposite — same pitch, different ERA. Ballpark: the same fly ball is a homer in one park and an out in another. Sequencing luck: allowing a walk, single, and homer in the same inning costs 3 runs; spread across three innings, maybe 1. The pitcher barely controls the order.
Over a full career these mostly wash out, and ERA becomes fairly honest. Over one season — the window fans actually argue about — the noise is huge. Level 2 introduces the luck detector (BABIP), and Level 3 builds FIP, which strips out the defense entirely.
The formula (optional — skippable)
ERA = (Earned Runs × 9) ÷ Innings Pitched
'Earned' excludes runs scoring after errors — a small nod to the defense problem that doesn't come close to fixing it.
How this stat lies to you
- It charges the pitcher for his defense's range and errors (partially).
- It ignores ballpark — a Coors Field 4.30 can be better than a pitcher-park 3.80.
- One disaster inning can distort a month; sequencing luck is invisible.
Check yourself
1. Which of these does a pitcher control least, despite it moving his ERA?
2. Why is one-season ERA risky for judging a pitcher?