FIP is only as independent as its inputs
Fielding Independent Pitching was one of the great clarifying ideas of the sabermetric era. It began from a genuine insight — that a pitcher’s earned run average is polluted by the defense behind him, the sequencing of hits, and the whims of balls in play — and it proposed a fix: judge a pitcher only on the three outcomes a defense cannot touch. Strikeouts, walks, and home runs. No fielder catches a strikeout, no shortstop ranges to a walk, and a home run leaves the yard before anyone can react. Reduce a pitcher to those events, scale the result to look like an ERA, and you get a number that strips out the noise ERA drags in. FIP earned its place, and it belongs in any serious evaluation. But the name oversells the achievement. FIP is independent of fielders. It is not independent of luck, of ballparks, or of the hitters a pitcher happened to face — and two of its three inputs are shakier than the clean formula suggests.
What FIP actually isolates
FIP takes a pitcher’s strikeouts, walks (plus hit-by-pitches), and home runs allowed, weights them by their run value, divides by innings, and adds a constant that puts the whole thing on the ERA scale. What it deliberately throws away is everything that happens once a ball is put in play and stays in the park — the ground balls, line drives, and fly balls that fielders convert into outs or fail to. That exclusion is the entire point, and it is defensible: batting average on balls in play is famously noisy and heavily influenced by the seven fielders behind the mound. By refusing to credit or blame a pitcher for those events, FIP removes a real distortion from ERA. So far the name holds. The problem is what survives the cut.
The home run input is barely stable
Of FIP’s three ingredients, the home run is the least reliable. Home runs are rare, which makes their per-season totals volatile, and whether a fly ball clears the wall depends on things the pitcher does not control: the dimensions and altitude of the ballpark, the wind, the temperature, and the year-to-year behavior of the baseball itself. A pitcher can allow the same quality of contact two seasons running and post very different home-run totals purely because of where he pitched and how the ball was manufactured. This is why analysts invented an alternative — expected FIP, which replaces actual home runs with a rate estimated from fly balls — precisely because they noticed that FIP’s home-run component swings around too much to trust in a single season. When you have to build a second metric to stabilize a third of the first one, the first one was not as solid as the name implied.
Strikeouts and walks import the opponent
The two stable inputs, strikeouts and walks, are the parts of FIP most in the pitcher’s control — but even they are not context-free. A strikeout is a negotiation between pitcher and hitter, and the hitter’s tendencies matter: a pitcher who faces a lineup of free-swinging, high-strikeout batters will rack up more punchouts than the same pitcher facing a disciplined, contact-oriented order. Walks carry the mirror version of the same effect. FIP does not adjust for the quality or the approach of the hitters a pitcher faced, so a reliever deployed against the soft underbelly of a lineup and a starter who works through the heart of the order three times are scored on the same scale. The strikeout and walk rates are genuinely the pitcher’s to a large degree, but they still carry a fingerprint of who stood in the box.
What the name quietly leaves out
FIP is silent on things that clearly are pitching skill. Some pitchers really do suppress the quality of contact against them — they generate weak grounders and lazy fly balls at a repeatable rate — and FIP, by design, gives them no credit for it, treating all balls in play as equally out of their hands. That was a reasonable simplification when batted-ball data was thin, but modern tracking shows contact suppression is partly a real, persistent skill for a subset of arms. FIP also ignores a pitcher’s ability to control the running game, to hold runners, and to pitch differently from the stretch. None of that makes FIP wrong. It makes FIP incomplete in a specific direction: it protects pitchers from being blamed for their defense, but it also refuses to reward the ones who genuinely make hitters hit the ball badly.
Why FIP still beats ERA for prediction
For all of that, FIP does the job it was built for. Because it leans on the most repeatable outcomes and discards the noisiest ones, a pitcher’s FIP predicts his future ERA better than his current ERA does. A starter with a shiny ERA and an ugly FIP is usually getting help — from his defense, from favorable sequencing, from a low home-run rate that will not last — and he tends to regress. A starter with a bloated ERA and a strong FIP is usually pitching better than his runs allowed suggest. That gap between the two numbers is one of the most useful early-warning signals in baseball analysis. And because the stat is transparent — three inputs, one formula, no black box — anyone can see exactly why a pitcher’s FIP and ERA disagree, which is part of why it became the default sanity check on a hot or cold run. The value is real; it just comes from FIP being a better noise filter, not from it being a complete account of pitching.
How to read it without overtrusting it
Treat FIP as the fielding-adjusted starting point, not the last word. Read it next to ERA to spot luck and defensive help, and next to expected FIP to check whether the home-run component is inflating or flattering the number. For contact-suppressing pitchers, lean on batted-ball data and expected stats that FIP throws away by construction, because for those arms FIP will systematically understate the skill. And remember the sample: over a few starts, FIP is dominated by home-run variance and tells you little. Over a full season it becomes trustworthy, and over multiple seasons the strikeout and walk components — the genuinely stable core — carry most of the real signal.
The honest read
FIP delivered exactly what it promised in its most literal sense: it took the fielders out of the pitcher’s line, and that alone made it more predictive and more fair than ERA. But “fielding independent” is a narrower claim than “pitcher controlled.” One of FIP’s three inputs swings on ballparks and baseballs, the other two carry the imprint of the hitters faced, and the whole metric ignores the real skill of weak contact. Use FIP as a superior estimate of a pitcher’s true talent and a strong predictor of where his ERA is heading — just do not mistake the word “independent” for a promise that the pitcher earned every bit of it himself. The defense is gone from the number. Plenty of context is not.