Baseball stats glossary

How do you read any baseball statistic?

A counting stat rewards playing time: thirty home runs beats twenty, but the second player may have missed two months. A rate stat divides to fix that — hits per at-bat, earned runs per nine innings. Rates compare a part-time player to a regular, but they get noisy in small samples.

An adjusted stat asks what a rate was worth where it happened, since runs are not equally easy to score in every season or park. Numbers with a plus — OPS+, wRC+, ERA+ — set league average at 100, and higher is better. Numbers with a minus — ERA-, FIP- — use the same baseline but measure cost, so lower is better. That sign is the most common misreading here.

So check whether a number is adjusted before you compare anything with it. An unadjusted rate from a pitcher’s era and one from a hitter’s era are not comparable, and care about the decimals will not make them so.

Hitting statistics

The scales are rough on purpose, and they shift with the run environment.

Pitching statistics

Pitching scales move with the run environment. A 4.00 ERA has been below average in some decades and above it in others.

Fielding, batted ball and situational statistics

Three sources: human charting, ball-tracking cameras, and win-probability models. The fielding numbers disagree with each other more than hitting numbers do.

What is WAR in baseball, and why do two sites disagree?

WAR is not one statistic. It is a recipe, and the two best-known kitchens use different ones. Baseball-Reference publishes bWAR, FanGraphs publishes fWAR, and for the same player they usually print different numbers.

The largest split is pitching. Baseball-Reference starts from the runs a pitcher actually allowed, then adjusts for the defence behind him, the parks and the opponents. FanGraphs starts from FIP, so it credits a pitcher only for outcomes fielders had no part in. A pitcher who allowed fewer runs than his strikeout and walk numbers suggest looks better at Baseball-Reference. The two also use different fielding inputs, so a shortstop can sit several runs apart. Neither is wrong; a disagreement is information about the player.

One rule follows. A gap of about a win between two players is inside the noise — smaller than the gap between the two published versions of the same player. WAR sorts players into broad groups. It does not rank two players a few tenths apart.

BABIP, and the word “lucky”

BABIP is batting average on balls in play: how often a ball hit into fair play became a hit, leaving out home runs and strikeouts. League average sits close to .300, and that stability is why anybody far from it gets called lucky or unlucky and the conversation stops there.

That reading fits pitchers better than hitters. Most pitchers have limited control over what happens once a ball is in play, so a starter with an unusually low BABIP has often been helped by his defence, and he tends to drift back toward normal. FIP is built on that reasoning.

For hitters the claim is much weaker. Individual hitters carry real, repeatable levels, driven by speed, which turns choppers into infield hits; by line-drive rate, since line drives fall in more often than anything else; and by contact quality. Hitters with all three have held marks above .350 across whole careers. That is not a decade of good fortune.

So a high BABIP supports a hypothesis, not a verdict. Check the batted-ball evidence beside it, and whether the same level appeared in earlier seasons. A mark that has held for four years describes a player.