Remember me

Universal Fielding Percentages

April 19, 2008
Putting Fielding Percentages on a Common Scale
 
            There is a unique problem with reading fielding percentages, which is the multiplicity of standards.   We interpret baseball statistics in casual conversation by the use of standards—he’s a 17-game winner with a 3.80 ERA, or he’s a .270 hitter with 80-90 RBI a year.   The existence of universally understood standards makes sense of all of the numbers, not merely those that meet or exceed the standard.  
            Fielding percentages are different, because
            a) there are no magic numbers, and
            b) the standards are so different from position to position that nobody can carry a full set of them around in his head.
            A .988 fielding percentage for an outfielder. . .is that good or bad?   It’s tremendous for a shortstop, outstanding for a second baseman, bad for a first baseman. ..what is it for a catcher?    .964 for a third baseman in the 1950s. ..good or bad?   No peeking.    .978 for a shortstop in 2007. ..above average or below?   No peeking.  
            Not only are there different standards for each position, but these standards have changed tremendously over time.   Yes, it is true that there are different expectations for hitters at different positions; yes, it is true that standards of batting and pitching excellence change somewhat over time.   Not to the same level.   200 strikeouts for a pitcher was outstanding in 1940, it was outstanding in 1950, it was outstanding in 1960, it is outstanding today.   There was a period in the 1960s when it was a little less outstanding.   90 RBI was a middle-of-the-order hitter in 1940; it’s a middle-of-the-order hitter today.   Seventeen wins was a quality starting pitcher in 1930 or 1960 or 2007.   ERA norms may be 3.70 in one decade and 4.70 in another, but they’re not 1.70 or 6.70. 
            There are some people who have a pretty good grasp on fielding percentages, of course, and there are many people who are very good baseball fans and have been for many years and can tell you who won the NL batting title in 1970 and the American League MVP Award in 1984, but who can’t tell you whether a .951 fielding percentage for a shortstop will get you bronzed or get you benched.   It’s very hard to compare one to another.   One might know that .986 is not a good fielding percentage for a first baseman and .941 is not a good fielding percentage for a shortstop, but. . .which is worse?      One might know that .997 is a good fielding percentage for a catcher and .986 is good for a second baseman, but which is better?
            I have a method to deal with this problem. . .it’s actually a very good method and I’ve had it for a long time, but I just realized (because of a letter in “Hey, Bill”) that I had never explained it to the public.   The method scores everyone on a zero-to-one scale with an average player at .500, good fielders over .500 and poor fielders under .500.     It enables you to easily and meaningfully combine fielding percentages from different positions so that if a player played 400 games at first base in his career, 350 at first base, 275 in the outfield and 4 at second base, you can give him a career fielding percentage which is a meaningful number to the extent that fielding percentage is a meaningful concept.   You can combine all the positions into one—or, if you prefer, you can get a separate number for each position.  
            This is how it works.  
            A player’s expected errors are the number of errors he would have had if he had had an average fielding percentage at the position, and the number of total chances that he actually had.   
            We will call his expected errors X, and his actual errors E.
            The formula is
            2X – E
-------------------------
               2X
 
            That’s all.  
 
            Let’s take Bill Russell, Dodger shortstop of the 1970s/1980s, famous for being not the best shortstop ever.    In 1974 Russell made 39 errors, and fielded .946 at shortstop. The fielding percentage for National League shortstops in 1974 was .961, and Russell had 724 fielding chances.   He thus had 28.02 expected errors.   He exceeded his expected errors by 10.98, resulting in a Universal Fielding Percentage Score of .304 for the season—well below average:
 
            (56.04 – 39) / 56.04 = .304
 
            In his career as a shortstop Russell was below average more often than above average, although he was rarely as error-prone as he was in 1974.   He was actually over .500—over the league average—in both 1973 and 1975, and in three other seasons.   In his career he had 339 errors as a shortstop, with 301.53 expected errors—a Fielding Percentage Score of .438:
 
            (603.06 – 339) / 603.06 = .438
 
            He also played 62 games in his career at second base, with a score there of .212, and 299 games in the outfield, with a score of .545.   He had one fielding chance at third base, and handled that cleanly for a score of 1.000.   Adding it up, he has a Universal Fielding Percentage Score of .436:
 
            Shortstop         Expected Errors          301.53             Actual Errors   339
            Second Base   Expected Errors              5.71             Actual Errors       9
            Outfield          Expected Errors              7.69             Actual Errors       7
            Third Base       Expected Errors                .05             Actual Errors       0
            Total                Expected Errors          314.98             Actual Errors   355
            Universal Fielding Percentage Score               .436
 
            For players with limited playing time, you might occasionally get numbers less than zero.   I would just enter them as zero, but I guess I’ll leave that up to you. 
 
            Here’s a Whitman’s Sample of Universal Fielding Percentage Scores for some other players:
           
            Omar Vizquel at shortstop                  .713
            (Never played any other position)
 
            Brooks Robinson at third base            .696
            Brooks Robinson overall                     .695
 
            Ozzie Smith at shortstop                     .684
            (Never played any other position)
 
            Al Kaline in the outfield                     .652
            Al Kaline, all positions                        .640
 
Alan Trammell as a shortstop              .645
            Alan Trammell, all positions               .639
 
            Sherm Lollar at catcher                       .630
            Sherm Lollar overall                            .634
 
Orlando Cabrera at shortstop              .638
            Orlando Cabrera all positions             .632
 
            Bill Freehan at catcher                        .613
            Bill Freehan at first base                     .642
            Bill Freehan overall                             .616
 
            Bobby Grich at second base               .619
            Bobby Grich at shortstop                    .408
            Bobby Grich all positions                   .599
 
            Stan Hack at third base                       .598
            Stan Hack overall                                .597
 
            Bill Dickey at catcher                         .570
            (Never played any other position)
 
            Juan Uribe at shortstop                       .568
            Juan Uribe all positions                       .560
 
            Robin Yount at shortstop                   .506
            Robin Yount in the outfield               .742
            Robin Yount all positions                   .549
 
            Ray Schalk at catcher                         .548
            (Never played any other position)
 
            Derek Jeter at shortstop                      .548
            (Has not played another position)
 
            Jim Hegan at catcher                           .535
            (Never played any other position)
 
            Travis Jackson at shortstop                 .531
            Travis Jackson at third base                .569
            Travis Jackson all positions                 .535
 
            Darrell Porter at catcher                      .531
            Darrell Porter overall                           .531
 
            Gabby Hartnett at catcher                  .521
            Gabby Hartnett overall                       .518
 
            Bobby Abreu in the outfield               .516
            (Has not played another position)      
 
            Ernie Lombardi at catcher                  .505
            (Never played any other position)
 
            Bob Elliott at third base                      .496
            Bob Elliott overall                               .499
 
            Bill Russell at shortstop                      .438
            Bill Russell in the outfield                  .545
            Bill Russell all positions                      .436
 
            Wally Schang at catcher                     .464
            Wally Schang in the outfield              .245
            Wally Schang all positions                  .434
 
            Rod Carew at second base                 .392
            Rod Carew at first base                      .455
            Rod Carew all positions                      .421
 
            Roberto Clemente in the outfielder    .414
            Roberto Clemente, all positions          .407
 
            Reggie Jackson in the outfield            .166
            (Never played any other position)
 
            George Bell in the outfield                 .053
            George Bell all positions                     .057
 
 

COMMENTS (13 Comments, most recent shown first)

tangotiger
IIRC, John Walsh in the Hardball Times 2007 Annual looked at all OF throwing arms using Retrosheet data, and ended up with Clemente as the best thrower over the last 50 years (IIRC). I don't remember exactly everything he did, but you can find articles of his online that explains the methodology.
11:39 AM Apr 25th
 
rpriske
Could it be (not having watched Clemente play) that memory is reminding you that he thre very hard? It seems to me that the sort of player that would just uncork one and fire it like a cannon will have a higher rate of throws that go wild than an outfielder that just throws the ball in 'normally'.
10:38 AM Apr 24th
 
bjames
.500 is the mean, yes. Roberto Clemente is under .500 because he has relatively high error rates. This seems to me to be a relevant point in his case, because I would assume that almost all of those errors were on throws. If you ask any baseball fan over 50 who had the greatest throwing arm he ever saw, he'll probably say "Clemente". But. . .doesn't it bother us just a little that he was making an above-average number of throwing errors?
11:30 PM Apr 23rd
 
jollydodger
I don't understand...if STATS, Inc or whoever can chart every ball hit at x trajectory to y spot on the field (as they divide the field into hundreds of little portions), why can't they chart where the player who fields the ball started from? Would that not help with "great play" or "great placement"?
6:48 PM Apr 21st
 
rpriske
...And I see the right above me Tango has made the same analogy...
4:22 PM Apr 21st
 
rpriske
This is similar, but not exactly the same as a positional version of 'FPct+', which I just made up but I am sure others have already done so. Basically the same as ERA+ or OPS+. Works, no? (forgetting of course that using fielding percentage as a defensive metric has problems other than just ease of use.)
11:16 AM Apr 21st
 
tangotiger
This stat is simply to normalize an existing stat to give it more relevance, much as you would divide ERA by the league ERA, you divide Errors per chances by the league Errors per chances, or HR per AB divided by league HR per AB.

Why does this normalizing process have to give you more than that?
11:12 AM Apr 21st
 
aheho
Ayjay,
Perhaps I should make myself clearer.
At this stage I don't know why we need another fielding stat that won't help telll us who is the better fielder.
7:02 AM Apr 21st
 
ayjay
aheho, it's not "obviously false" that Vizquel has a higher fielding percentage, or a higher UFP, than Ozzie, it's demonstrably true. Just add up the numbers. That doesn't say anything about whether he's a better fielder, but Bill isn't saying anything about whether he's a better fielder, is he?
1:26 PM Apr 20th
 
tangotiger
The same objective can be achieved by doing E/X * 100. So, 39 Errors when the average would be 28 would come in at 139%, meaning errors at 139% of the league average. What Bill is doing is dividing this number by 200 (.695) and then subtracting 1 from it, giving you .305.

It's all a question of scale, and whether .500 is a better scale than 100, and whether you want "good" to be a low number or high number.

1:23 PM Apr 20th
 
sandy32
I'm confused about what this means -- or if a meaning is supposed to be inferred. Clemente at .414? Does that mean that a below average fielding percentage is meaningless? Also, what about Larry Bowa? I remember he had some great fielding percentages but wasn't one you thought was a great shortstop. But I did enjoy the article.
12:28 PM Apr 20th
 
aheho
Mildly interesting, but inherently flawed as an anaysis tool because of the well documented problems with using fielding percentage to rate fielders.

Case in point: Your UFP rates Vizquel over Ozzie. Obviously false.
10:17 AM Apr 20th
 
77royals
Bill,

Not being a math kind of guy, I know what I want to ask, but I'm not sure how to phrase. But here goes. Is .500 like the mean, with anyone over being above average, and every one below below average. Or is there another line, like .560 as the average?
8:27 PM Apr 19th
 
 
©2024 Be Jolly, Inc. All Rights Reserved.|Powered by Sports Info Solutions|Terms & Conditions|Privacy Policy