Fielding percentage hides what defenders actually do
Fielding percentage is the oldest defensive stat in baseball and the one that has resisted every analytical revolution that has swept through the game. It still appears on broadcast graphics. It is still the stat quoted when a Gold Glove is announced. It is also the stat most likely to be hiding a bad defender from his own evaluation. Fielding percentage doesn't measure defense. It measures the willingness of an official scorer to charge an error, and the willingness of a defender to attempt a play that might result in one. Those two things are not the same as actual defensive value, and the gap between them has been protecting weak gloves for over a century.
What the stat actually counts
Fielding percentage is defined as putouts plus assists, divided by putouts plus assists plus errors. It is, in other words, the percentage of times a fielder successfully completed a play they attempted. A shortstop who fielded 400 ground balls cleanly and committed 10 errors has a .976 fielding percentage. A shortstop who fielded 300 ground balls cleanly, never attempted a play on another 100 balls that he probably could have reached, and committed 5 errors has a .984. The second shortstop looks better in the box score. He is much worse at defense.
The stat rewards passivity. A fielder who lets a tough chance go past for a hit takes no statistical hit — the play becomes a base hit against the pitcher, not an error against the fielder. A fielder who attempts the same play and bobbles the ball takes the error. The incentive structure has been encouraging defenders to play conservatively for as long as the stat has existed, and a careful veteran fielder can extend his career by simply choosing not to attempt the chances that might cost him on the back of his baseball card.
The range problem
Range — the ability of a fielder to reach balls in play — is what actually separates good defenders from bad ones. Two shortstops can have identical fielding percentages while one of them is fielding 250 more balls per season than the other, because he gets to chances the second shortstop never even attempts. The first player is creating outs that don't appear anywhere in his own stat line. The second is failing to create outs that don't show up as failures.
This is why every advanced defensive metric — UZR, DRS, Statcast's Outs Above Average — starts by trying to measure range, and treats fielding percentage as one input among many rather than the primary output. The advanced metrics consistently disagree with fielding percentage about who the best defenders are. They agree with the scouts, the film, and the actual run prevention. Fielding percentage is the only stat in the disagreement, and it is the one still printed on the broadcast graphic.
The scorer-judgment problem
Fielding percentage isn't even a clean record of what happened on the field. Errors are awarded by an official scorer who is exercising a judgment call about whether a play "should" have been made. The same play in two different ballparks can be a hit or an error depending on which scorer is working that night. The same play by two different fielders can be scored differently depending on the scorer's view of those fielders' baseline range — a scorer is more likely to award an error on a play that a "good defender" would normally make.
This is why historical fielding percentage comparisons across eras are essentially meaningless. The standards have shifted. The scorers are different people with different priors. The stat is a moving target dressed up as a count. A 1955 shortstop with a .960 fielding percentage and a 2025 shortstop with a .985 fielding percentage are not being measured against the same yardstick. The numbers on the backs of their cards suggest they are.
Why the position adjustment makes it worse
Fielding percentages are reported by position, which seems sensible, but the position adjustment hides as much as it reveals. The league-average fielding percentage at first base is roughly .994. At shortstop it is roughly .974. A first baseman with a .994 fielding percentage is exactly average for his position. A shortstop with a .984 fielding percentage is well above average for his.
But these numbers are not telling you anything about the difficulty of the positions or the relative defensive contribution. A passable first baseman and a passable shortstop are both posting near-average fielding percentages for their roles, but the shortstop is creating roughly 20-30 more outs per season through range that the first baseman doesn't need to provide. The position-adjusted comparison treats these contributions as equivalent. They aren't. The shortstop is doing the harder job and being measured against a more forgiving standard for completing it.
What replaces it
The defensive metrics that have replaced fielding percentage in serious player evaluation are imperfect but they are measuring the thing fielding percentage was always supposed to measure. Outs Above Average uses Statcast tracking to compute, for every ball in play, the probability that an average defender would have turned it into an out. A defender who gets to a ball that average defenders convert 30% of the time gets credit for 0.7 outs above average. A defender who lets a ball go past that average defenders convert 80% of the time gets debited 0.8 outs below.
This stat is doing what fielding percentage pretended to do: measuring the defensive contribution. It captures range, it captures reaction time, it captures positioning, it captures arm strength on plays in the outfield. It also doesn't reward passivity, because the credit for a play is awarded based on the difficulty of the chance, not on whether the fielder chose to attempt it. The OAA leaderboards and the fielding percentage leaderboards rarely overlap. The OAA leaderboards are the ones that match the players whose teams actually prevent runs.
How to read a defensive stat line
The shortcut is to treat fielding percentage as a hygiene check, not a measurement. A fielder posting a .950 has a hands problem and is making errors on routine plays. A fielder posting .990 is reliable on the chances he attempts. Neither tells you whether he's a good defender. For that, you need to know how many chances he attempted and how many an average defender at the position would have attempted in the same situations. That comparison is what every modern defensive stat is built to make.
The reason fielding percentage refuses to die is that it is simple, intuitive, and has the longest historical record of any baseball stat. The reason it shouldn't be the lead defensive stat on a broadcast graphic is that it has been hiding what defenders actually do since the day it was invented. The defenders it has been hiding are not always the ones with the prettiest fielding percentages. They are the ones reaching the balls nobody else gets to, taking the risks that occasionally show up as errors, and creating the outs that the pretty-percentage guy next to them never even tried for.