What is FIP — and why our MLB model trusts it over ERA

Two pitchers, both with a 3.80 ERA. One plays in front of the best defence in the league; the other behind statues. Are they equally good? ERA says yes. FIP says: let's check.

What FIP counts

Fielding Independent Pitching uses only the outcomes a pitcher controls alone: strikeouts, walks, hit batters and home runs. Everything that depends on fielders — whether that grounder becomes an out or a single — is deliberately excluded, replaced by league-average outcomes. The result lands on the same scale as ERA, so a 3.20 FIP reads like a 3.20 ERA, minus the luck.

Why it predicts better

Study after study finds the same thing: this year's FIP predicts next period's ERA better than this year's ERA does. That's the property a prediction model needs — we don't care what a pitcher's record looks like, we care what he's likely to do tonight. When ERA and FIP disagree sharply, FIP usually wins the argument in the long run — and a pitcher whose ERA flatters his FIP is exactly where a market can be slow to adjust.

How Modal uses it

Every game on the MLB board shows both probable starters with their FIP, and the model's starter rating weights FIP first, with ERA as the fallback when FIP isn't available. Fused with the team Elo and blended against the market, it's a probability you can interrogate: if you disagree with our number, the starter line usually tells you where the disagreement lives.