Expected ERA — a pitcher's earned run average if every batted ball returned a league-average outcome.
Earned run average is the oldest pitcher rate stat in the book, and also one of the noisiest. A pitcher who has been hit hard but bailed out by his defense can post a sparkling ERA over fifty innings; a pitcher who has been pitching beautifully into bad batted-ball luck can carry an ugly one. xERA — expected earned run average — strips most of that luck out.
The metric is built from Statcast data. For every batted ball, the league record holds a probability of it becoming a single, double, triple, home run, or out, conditioned on exit velocity and launch angle. xERA assigns each pitcher the league-average outcome for the contact he actually allowed, layers in his real strikeout and walk rates, and converts the resulting line into runs per nine innings.
Because xERA replaces sequencing luck and defense with neutral inputs, it tends to settle faster than ERA. Two months into a season, ERA is still being yanked around by a handful of bunched-up hits; xERA already reflects what the pitcher is doing on contact. The gap between the two — call it ERA minus xERA — is one of the most common regression flags in pitcher modeling. A pitcher running an ERA a full run below xERA is, on average, a sell candidate; one a full run above is a buy.
The metric isn't perfect. It assumes league-average defense behind the pitcher, which is often wrong; it treats every park as neutral; and it can't distinguish a pitcher who genuinely suppresses hard contact through deception or sequencing from one who is simply running good for now. Used alongside the rest of the Statcast suite, however, it's the single most useful one-number summary of pitcher quality available in the public data.
Our pitcher ratings lean on xERA — and the gap between it and ERA — as the single biggest regression input. A starter trending toward an xERA-implied true talent line is one of the cleanest signals we use when grading totals and run-line edges.