You Are Here > > Blog >

SITE NEWS: We are moving all of our site and company news into a single blog for We'll tag all B-R content, so you can quickly and easily find the content you want.

Also, our existing B-R blog rss feed will be redirected to the new site's feed. ยป Sports Reference

For more from Andy and the gang, check out their new site High Heat Stats.

The Problem with Fielding Percentage

Posted by Sean Forman on November 10, 2010

Citing fielding percentage as evidence of the best fielder is like saying the best quarterback is the one with the lowest interception rate, or the best basketball defender is the one with the fewest fouls.

Not making errors has value, but its far from the most important thing you can do as a defender.

This entry was posted on Wednesday, November 10th, 2010 at 9:50 pm and is filed under Announcements, Awards, Fielding Stats. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

44 Responses to “The Problem with Fielding Percentage”

  1. Fielding Percentage is a problem. Most defensive statistics, new old and old age, have problems.

    In fact, this post could have just as easily been called, The Problem with UZR.

    Or some other stat.

  2. It's a PART of the story. But it's far from being able to trump anything else.

  3. If you have the range of a statue but good hands, you can have no errors and a great fielding pct.

  4. Charles Saeger Says:

    Why pick on fielding percentage? It isn't the be-all, end-all, but it does tell you how often a fielder makes errors. As opposed to raw range factor, which has six million biases keeping it from telling you anything.

  5. Is Fielding Percentage perfect?


    But it's still a damn sight better than anything else.

  6. Frank Clingenpeel Says:

    Even though it is a viable starting place for judging fielding performance, fielding percentage taken alone can be extremely misleading. Take, for example, the case of Joe Morgan. While Morgan, a valid Hall of Famer, was undeniably a good-fielding second baseman, his fielding percentage didn't on its own reflect his limited range -- which in turn resulted in his not being penalized by the larger number of chances that simply slipped beyond his reach and bounded into the outfield as hits.

  7. @3 - I came here to say this.

  8. dukeofflattbush Says:

    Someone on this blog ripped the Triple Crown a few months ago, by saying "it didn't prove anything!" He was very adamant. But he was wrong, it proved who led the league in AVG/HR/RBI in a given year.

    Josh Hamilton led the AL in AVG not Hits.
    I get it.

  9. Fielding percentage tells us absolutely nothing. Anyone who watches the games knows that the scorers are beyond abysmal. The word does not exist in english for how bad they are. At least once a game, a play occurs where it is obviously an error and they give the batter a hit, and often, more than once. This is why we've had so many records for consecutive errorless games and high fielding percentages in recent years. And it certainly doesn't help ERAs and BAAs. But it certainly makes the offensive numbers look better.

  10. Wade@9- You're right, the biggest problem with fielding percentage is the with the official scorers. Anything close and it comes down to who was the fielder, the pitcher and the hitter. If Joe Blow is the pitcher and Cal Ripken is the fielder and it's a close play at all, you can make good money betting on how it's going to be scored. Plus, you have the additional problem of it not telling you about the balls they didn't get close enough to to make a play on.

    That being said, it's not like Jeter was Dave Kingman or Dick Stuart or something. He was good enough at short that his team finished with the 3rd most wins in the majors. The reason New York didn't win the AL East was because of Vazquez & Burnett. Yeah, he didn't get to a few balls that a better fielding shortstop would have and that might have cost them a game or 2. But there ain't no shortstop ever been born that's going to turn a ball that lands in the upper deck into an out.

    He didn't deserve the Gold Glove either, however.

  11. Mike Felber Says:

    When you consider how arbitrary error decisions can be, and how FP does not reflect range at all, I cannot see that even with certain flaws, that some variation of range factor is not the best measure of overall defensive ability. Either go with a zone related fine tuning of that, or use range factor with a team & pitcher adjusted estimate of how many balls were actually put into potential fielding range.

    A team could easily be good enough that you could have a Dr. Strange glove at short & be that good. There are way too many factors to go by standings. Though Jeter was not THAT bad, he is below par even when factoring in "hands". Though I guess he can at least hit better for 1 or 2 more years than in '10.

  12. true fa and rf arent perfect if i had to choose id choose fldpct over rf

  13. Baseball is a game of averages and people are trying to pin down everything on a a number. Not possible, even though the folks here want you to believe that.

    Fielding % is clearly not the most important stat. However for a middle infielders DP's certainly are. Also, although not perfect comparing a player to league averages in range is also pretty solid to get a idea on how much ground they cover. Strikeouts can effect this a little.

    Lastly, no % is perfect, because the actual game is not a number. Stats are important but not everything.

  14. Fielding percentage isn't as bad as you think. I once went through a bunch of box scores and found that the team that had more errors in a game lost something like 80 percent of the time. Considering how lenient scorers are in giving errors, you've got to really screw up to get one, and when you do, it often spells disaster. I think range may actuallly be overrated. Make the routine plays and you'll be alright. Plus, range is sort of accounted for in fielding percentage because the more plays you make, the higher the fielding percentage.

  15. Tim @14

    Are you suggesting that given two players with 50 balls hit into the general area where they play defense you would prefer the player that got to 35 of them and only made 1 error, thus converting 34 of them into outs to the player who got to 43 of them but made 3 errors thus converting 40 of them into outs?

    Range doesn't have nearly the impact on fpct that you suggest either. The number of outs and assists is a much larger number than the number of errors. So additional balls fielded have much less impact than an additional error or two. Also, fielders with large range are more apt to make errors for a variety of reasons including that they are fielding the ball in a less controlled position and need to make more hurried throws.

  16. @14

    Ummm... "the more plays you make, the higher your fielding percentage" is not a true statement. You could make 1,000 plays, and the next guy could make 50. But if he made no errors, and you made 100, guess whose fielding percentage is higher?
    The truth is, range alone is incomplete, because we need to know how many plays a guy made, not just how much ground he covered. But that's what stats like Rfield do. They see how many plays a player would have made as compared to how many plays an "average" fielder would have made. That lets us know if a guy's getting to balls and making plays that other players are missing, or if a guy is missing balls another player would get to.
    For those who don't understand why fielding percentage is a bad stat, imagine a player with the best glove and arm in the history of the world. He plays shortstop. Unfortunately, he has no legs. Wherever he is placed at the beginning of the inning, that's where he stays. Now, if you have a player like that, his fielding percentage will be 1.000. However, he'll only make plays on like 10% of the balls an average major league shortstop would get to. So, essentially, 90% of his plays are errors, because he can't get to the balls that anyone else would get to. I hope that helps.

  17. mr.baseballcard Says:

    Question: On a routine pop-up between SS & 3B (or SS & 2B) etc when either player could take it and the 3B (or 2B in the second scenario) calls for and catches it, does this "chance" get included in UZR or RF9? I ask as it seemed to me that although DJ's range has declined it seemed to me that this year at least one time per series that Cano was taking a pop-up that Jeter could have handled? The result, if included in the various stats being discussed here would be DJ's fielding stats suffer needlessly and Cano's get inflated needlessly. Note that Cano won his first GG and his PO's were off the charts.

  18. Get over it, everyone. Jeter won the GG. Move on.

  19. Johnny Twisto Says:

    Larry, this is a forum for discussing baseball. Since you are not interested in discussing baseball, this is probably not the place for you. We will continue to discuss baseball here, and many of these discussions will involve the past, since it is easier to discuss things that have happened than those that have not.

    (Much as it is easier to count grounders fielded than those that were not, which is why fielding stats are so tricky.)

  20. Johnny Twisto Says:

    Question: On a routine pop-up between SS & 3B (or SS & 2B) etc when either player could take it and the 3B (or 2B in the second scenario) calls for and catches it, does this "chance" get included in UZR or RF9? I ask as it seemed to me that although DJ's range has declined it seemed to me that this year at least one time per series that Cano was taking a pop-up that Jeter could have handled? The result, if included in the various stats being discussed here would be DJ's fielding stats suffer needlessly and Cano's get inflated needlessly. Note that Cano won his first GG and his PO's were off the charts.

    RF9 is putouts plus assists per 9 innings. In your example, the player who caught the ball would be credited with a putout, so yes, it helps his range factor. UZR is more complicated, but I believe it works along the lines of how often a ball hit in that zone is fielded. If this is a routine popup, it will be caught 99% of the time, so the player that makes the catch may get a very tiny credit, and the other player should not be penalized.

    For infielders, we are mostly concerned with assists. Most putouts, as you note, are simple plays and/or discretionary. If Cano is taking putouts that Jeter could make, yes, that will increase his RF and his fielding%, but the advanced stats don't really care about that. They are trying to measure how many groundballs were hit in each player's area and how many each turned into outs.

  21. TZR and UZR only consider ground balls, so Cano can not steal chances from Jeter in that way.

  22. The basic problem with trying to find meaning in fielding percentage is that the spread from best to worst, even at a busy position like SS, never amounts to more than a couple dozen plays. All other things being equal, yes, you'd prefer the more sure-handed SS. But all other things are rarely equal; in particular, the spread in range factor (or any other measure of chances handled) absolutely dwarfs the spread in error rate. For example:

    Derek Jeter was charged with just 6 errors this year, the fewest of any regular AL shortstop. In almost exactly the same number of innings, Oakland's Cliff Pennington made 25 errors, the most of any AL regular. So in a head-to-head comparison, Jeter is up 19 plays in this department.

    However, Jeter handled just 553 defensive chances. Pennington handled 739 chances, 186 more than Jeter. (And the bulk of that difference -- 130 plays -- was in assists, so please let's not hear that Pennington is hogging all the popups while Cap'n Jetes defers to his teammates.)

    In this rough calculus, Pennington is now 167 plays ahead of Jeter -- more than one per game. Sure, a more advanced analysis would adjust for the K rates of the respective pitching staffs, the breakdown of LHPs and RHPs, ground-ball and fly-ball tendencies, and even a measure of balls actually hit into each fielder's "zone." But I don't see how anyone can think about the basic defensive counting stats -- chances and errors -- and come to the conclusion that the error rate has any large meaning in the big picture of evaluating defense.

    (Aside: I think's one of Sean's opening analogies sidetracks the point a bit. Assuming that we're comparing real-world regulars at the position, a quarterback having the lowest interception rate is immensely more meaningful than a shortstop having the lowest error rate, simply because an interception is a vastly worse outcome than an error.)

  23. The best way to judge a stat is to look at it on a team level. The Yankees had 69 errors, the Royals had 121. Which is better? Obviously 69. Most of the teams with low error totals were good, most of the teams with high error totals were bad. (Texas was an exception, though). Bottom line, it's not good to make errors.

  24. Fielding pct. has always been a known problem. Advanced stats weren't even needed to see the flaw.
    I believe that was one of the reasons Rawlings put it to a vote rather than giving the GG to the fielding pct. leader at each position.

    Now I have a question about UZR, even though it is from Fan Graphs. Andrus had a UZR of 12.1 in 2009. This season it was 0.1. How is that possible unless he had some horrible injury?

  25. JT @19,

    That is fine. Then let's please talk about baseball. I think we all truly understand how terrible Jeter is. Maybe there is another subject out there?
    Four of the last five baseball related posts are:
    1) "Jeter has Yanks over a barrell" - discussing how bad he is but that he will still be grossly overpaid
    2) "My head just exploded...because Jeter won another gold glove" - self explanatory
    3) "Best Defensive Shortstops" - Jeter is featured again showing how he ranks on the bottom of the list of SS statistically. Nice positive shout-out to Tulo though - that was refreshing.
    4) "Problem with Fielding %" - I admittedly may be reaching here (due to having Jeter on the brain from the 3 previous posts) but a new thread about how unreliable the one decent statistic that Jeter had in all of 2010 seems a little coincidental, especially since that very subject was covered at least 20 times in the previous 2 posts. But if everyone just all of a sudden had a desire to discuss the merts/failures of Fielding % completely unrelated to the Jeter GG fiasco, then I truly apologize for overreacting to my perception of a trend.

  26. I think that sometimes evaluating someone s fielding percentage is largely dependent on his reputation.

    I saw Brooks Robinson play many times against the Yankees in new York, especially after 1970 when he was the MVP of the WS against the Reds.
    I saw Robinson make some spectacular plays, but usually, he was just THERE, a few steps and he was in front of the ball. and that came from years of experience in the league. He won 16 GGs, whether he deserved all of them, I don t know, both Clete Boyer and Graig Nettles were superior fielders as well.

    And the difference between very good to great to extraordinary fielding (I htink) for 3B and SS is best measured by plays attempted. For 2B, maybe the best measure is DPs.......and that is why Bill Mazeroski is in the HOF, the best ever to turn a DP.

    I do agree that scorers are too lenient. During the last game of the WS, i think it was the inning that Renteria hit his HR, I saw Nelson Cruz of the Rangers attempt a diving catch and he came up was socred as a hit, the next play he made a very nice running catch agaisnt the wall. But on the play for which he dove...he closed his eyes!!!!! That is an error!!!!!!!

    And as a side note one key element the best professional QB is the one who makes the fewestt mistakes. Brett Favre is a great, great quaterrback but he does make mistakes...thows intercepetions, many times at inopportumne monents for his teams.

    If i had to choose a quarterback for an immensely important game, Bart Starr would be the man. He threw very, very few interceptions. Still the only QB to win 5 NFL chqampionships and with a 9 and 1 record in must win to avoid elimation games. he lost his first, the 1960 NFL champuionship game agaisnt the Eagles and after won in 61, 62, 65, 66 and 67 including the famous Ice Bowl and first two Super Bowls..

  27. Johnny Twisto Says:

    Andrus had a UZR of 12.1 in 2009. This season it was 0.1. How is that possible unless he had some horrible injury?

    Last year he slugged .373, this season it was .301. How is that possible?

    Guys have good seasons and bad seasons. One's performance in a particular game, month, season, etc, is just a sampling of one's true talent. Maybe Andrus just didn't make as many plays this season -- sore knee, bad positioning, slower reactions, more balls hit just out of reach, who knows?

    Or maybe UZR is "wrong." It's not a Truth. It's just an estimate, and it should be supplemented with other defensive numbers as well as observation.

    FWIW, TZ, DRS, and the Fans' Scouting Report all rate Andrus lower this season than last season (to varying degrees). So I'd guess he probably didn't play as well this year. Which is not to say we shouldn't expect him to play better next year, if we think he has the talent/skills/tools to do so (and most seem to think he does).

  28. Johnny Twisto Says:

    But on the play for which he dove...he closed his eyes!!!!! That is an error!!!!!!!

    An error is when a play should have been made with usual effort. That was not an error.

    But this is the advantage of ignoring errors. Whether it's an error or hit, in each case the result is the same. A man is on base and it's a play not made. The advanced defensive systems will treat it the same.

  29. Johnny I know the definition of an error.

    What I was saying is that he made a mistake or an error by closing his eyes....he might have caughtt the ball if he had kept his eyes open.
    he could have played the ball on a hop.

    But I think if a player makes an above average effort for a ball and after his effort....he comes up just short...there should ne some other statistic to reflect effort....

    .But closing the eyes is a fundamental mistake for a fielder in any posiiton.

  30. I think most rational folks that follow baseball can agree that fielding % is not perfect. But it also shouldn't be tossed out. Let's also not try to pretend that other ratings, like UZR also factor judgment into play. A batted ball must be classified into Soft, Medium or Hard. My understanding is that this is a judgment made--or am I wrong and the actual speed of the ball is tracked with a radar gun at the time of the play?

  31. John DiFool Says:

    "I once went through a bunch of box scores and found that the team that had more errors in a game lost something like 80 percent of the time."

    That may be more because good teams tend to avoid errors, not that avoiding errors helps you win ballgames.

  32. To me, fielding percentage is like batting average. It's a fine stat in and of itself, but it only shows what it shows and nothing else. Batting average is a great place to start looking at a hitter's ability, but you need to look at extra bases, walks, strikeouts, etc. as well. Same with fielding percentage - start there, but don't stop there.

  33. Player A, no range, 10 balls in play, gets to 7, no errors
    Player B, better range, 10 balls in play, gets to all 10, 2 errors

    not much difference, but player B at least in position to make all 10 throws and fewer base runners

  34. how is rf biased? all it is is putouts + assists over 9, very straightforward and fair to me. no measure of what was in play, just results - did they get there or not. how often did they make a play. just look at it and errors/fldpct together and that's good enough for me. what the pitchers get in terms of contact can't be changed, ignore it when evaluating a fielder. trade two guys who have opposite biases in rf, why wouldn't the result be reversed to the appropriate degree? to me just observing rf with err/fldpct is plenty. i mean, yeah, you'll need a historical curve to compare eras, and you can say that the scorer is biased, but all players in the league are subject to similar biases. then there's the ballpark factor, which is minimal. i feel like these comments are splitting hairs even more than other baseball stat arguments.

  35. Johnny Twisto Says:

    It's biased because not all players have the same number of opportunities.

  36. Jason @34, let me expand on JT's correct answer @35. The NL average last year was that 31% of all balls in play, (not counting bunts), were ground balls. For the Cards it was 36%. This is a huge difference. If we did not adjust for this the StL infielders would look great, and the outfielders bad, because the opportunities were distributed so much differently from the average. When someone says that player A is better because he had more RBI than player B, we are using the unstated assumption that all else is equal. Like RBI, this is not true for simple RF.
    As John Autin said @22, pitchers K rate GB/FB tendencies, and # of RH batters vs LH batters faced can all significantly change the number of balls hit to a particular fielder. There are also possible park effects; think of the large foul ground in Oakland, or all the balls that would have been caught in other parks but bounce high off the wall in Fenway.
    Overall, the biggest problem in judging a players defense is trying to figure his opportunities. UZR and Dewan's plus/minus, (often called DRS), get data in which someone has given a direction and a speed estimate, (soft, medium, hard), for all ground balls, and spot landed or fielded for outfield balls with a notation of line drive or fly ball. There is subjectivity here and there likely is systemic bias. (Calling balls not fielded hard hit, or line drives, while calling balls fielded more often softly hit or fly balls.) There now exists Field f/x, at least at some parks, which would give us exact figures for ball direction, speed and hang time, as well as fielder positioning. At this time the information gathered has not been publicly released as the earlier Pitch f/x data was from the begining.

  37. Also, RF is poorly constructed. It is put outs plus assists per 9 innings. But this mixes apples and oranges. For an outfielder the put outs are what we want. Adding in assists just puts together two somewhat different skills that we should want to measure separately. For infielders it is worse. We want to know balls successfully fielded, but assists include the pivot on double plays, relays from the outfield and rundowns in the basepaths. Put outs include pop ups (easy), line drives, (often not so easy), throws from other infielders, throws from outfielders to double off a baserunner, pick offs and caught stealing, as well as ground ball plays made unassisted. If we have play by play info, as from a scorecard, then we can sort out a lot of this. Some plays can be made by more than one player, e.g. pop flies and caught stealing at 2nd. Some plays can be made in more than one way, like ground balls near 1st. The 1B can take it unassisted, or throw to the pitcher covering. There is a lot of noise in RF, some of it can be reduced of eliminated with play by play data; but we still don't know enough from that to say that 2 players faced the same number of balls that were easy, medium and hard to field.

  38. Hey, I'm relatively new to the advanced fielding stats. Can someone help me out here?

    Robinson Cano led the league in putouts, assists, and fielding pct. , but his Rtot for the year is "0" and the advanced fielding stats don't show him to be a great 2nd baseman it seems (in 2010).

    How is this possible?

    I also think he looks extremely competent with the glove, and his double play turn looks as good as they come with a great arm.

  39. When I was 12, I was playing CF. I had a sinking liner hit right at me. I charged, slid, but couldn't come up with the catch. I was able to knock the ball down and keep it in front of me. When I got back to the dugout, the scorekeeper asked if I had hit the ball with my glove before it hit the ground. "I'm not sure. I think I did," was my reply. "Then it's an error," he said.

    That was all I needed to know to conclude that errors were dumb. Had I simply laid up and played the ball on the hop, I wouldn't have gotten an error. By trying and ultimately failing to catch the ball, which might not have been catchable at all given my positioning, I was charged with an error. Again, I was 12.

  40. Johnny Twisto Says:

    Robinson Cano led the league in putouts, assists, and fielding pct. , but his Rtot for the year is "0" and the advanced fielding stats don't show him to be a great 2nd baseman it seems (in 2010).

    How is this possible?

    1. He led the league in PO and A in large part because he played every day. Only one other 2Bman in the league played more than 90% as many innings as Cano.

    2. PO for a 2Bman are not that important. Someone else in this thread or one of the similar ones mentioned that he thought Cano was catching a lot of popups this season which Jeter could have taken. Most PO are either routine plays (routine pops, receiving the throw on a force at 2nd) or discretionary (taking the popup that anyone could have taken), or both.

    3. Assists are more important, as most of them indicate the successful fielding of a groundball and throw to first. But, refer back to (1). Cano had a lot of assists because he played a lot of games. Of the 13 AL 2Bmen who played at least 500 innings, Cano was 9th in assists per inning.

    4. Fielding percentage is not that important. From the team's perspective, there is no difference between a grounder that is reached but booted, and a grounder that is not reached. Both result in a man on first. Traditional stats do not count the play not made against the defender, even though the result is the same. In fact, the result might be worse for the play not made -- runner on first can go to third -- while an error might keep the ball in the infield and the runner on first stops at second. (Throwing errors which allow runners already on to advance extra bases are different, of course.)

    5. Going back to (3), we still have not fully accounted for Cano's opportunities. The innings he played are only part of it. If the ball is never hit near him, he can't make any plays. So the advanced stats attempt to estimate his opportunities. UZR and DRS use a record of where every ball was hit (see kds's post 36). Total Zone (Rtot) goes off the play-by-play records. Essentially these systems are trying to break the field into zones, determine how many balls are successfully fielded when hit into a specific zone, and determine how many balls Player X successfully fielded.

    6. But these advanced systems are estimates. They try to eliminate the bias of just using your eyes, but they can have their own biases (again see #36). By and large they get it "right" (look at all the players rated highly, you will find they are generally considered good fielders, and vice versa). But for any individual player, the system can miss. Maybe the player faced an inordinate number of balls right on the edges of his zone, or hard-hit balls not recorded as such. People who know more about them could probably think of other reasons.

    7. I agree that Cano turns the DP well, but no major league 2Bman turns it poorly. I doubt the difference between Cano and the worst would be more than 10 completed DPs over the course of the season, maybe worth a few runs.

    8. Cano only rates average this season by TZ and UZR, but he does rate well by DRS.

    I think the advanced numbers have value, but no one should take one single-season number as gospel. I advocate looking at multiple numbers and supplementing that with subjective opinion.

  41. Phil Haberkorn in Indiana Says:

    OK, how would you have rated me as a Gold Glove candidate in my church softball league? One season our second baseman was my little brother, Little League age, and I was 18 playing first base. I'd have to go to my right more often for ground balls and flip the ball to him as he raced for first to beat the runner, but the runner sometimes beat him to the bag - so even though I fielded the ball, no assist was recorded because the batter got credited with a hit. Next year, the second baseman was only a year younger than me, quick as a cat, so I never took more than one step to my right, so now he's the one racking up the assists. We're getting more outs, and my fielding percentage was about the same but now it's based on more putouts, so did I have a down year that first year, or am I less likely to get Gold Glove consideration the second year just because we have a second baseman who can reach more of his "territory" and the stats never reflected my range during either year? I need to know, because I'm asking the League Commissioner (Dad) to act in the best interests of softball and overrule the award voting ... (I think there was a lot of ballot-stuffing in Hicksville which I believe skewed the results).

  42. Phil Haberkorn in Indiana Says:

    Re: post #1...."good teams tend to avoid errors, not that avoiding errors helps you win ballgames...."
    I think that statement is a hypotenusal impossibility, or a circular illogism, or whatever you call it.
    Ummmmmmmmmmm, OK, so how do you define "good teams" if it's not based on winning ballgames, which you do by avoiding errors? What's the last "good team" that led the league in errors? What's the last last-place team that led the league in fielding percentage?
    They say good pitching beats good hitting, which explains the 1966 World Series.
    But does anybody's fielding "formula" take into account the effect lousy pitching has on fielding?
    A shortstop commits more errors this year than last year, but he's getting more chances because the pitchers are giving up more ground balls, so is the shortstop having a bad year? Maybe bad enough that the manager starts thinking about moving him to first base?

  43. Johnny thanks for the explanation regarding Cano. I guess the key bit of info is that no otehr 2nd baseman played everyday or even close to it it seems.

  44. Phil @ 42,

    I think it is more likely that the manager would see that he was leading the league in PO and Assists, and vote for him to win a Gold Glove. If he could hit a little he might win.

    More seriously, I think it is very important to try to remove the effects of fielding when rating pitchers. There is strong reason to believe that the big improvement in ERA for TB between 2007 and 2008 was mostly fielding.