March 12, 1998

Shortstops and DFTs

by Clay Davenport

This year's BP includes a new way of measuring fielding. Significantly, it is, to my knowledge, the first book that has attempted to apply fielding measures to minor league players. To give everyone a better idea of how the fielding system works and, most importantly for judging current players, has worked in the past, I'm starting a retrospective look through major league history. This is especially important because the results I get are often different - and occasionally, dramatically different - from the results other analysts, primarily Pete Palmer's Fielding Runs (FR) and Sherri Nichols' Defensive Average (DA). Davenport Fielding Translations (DFTs for now) are much kinder to Roberto Alomar, for instance, than Palmer's FRs ever were, while not nearly so nice to John Valentin as Defensive Average was.

A problem with assessing fielding statistics is that, unlike batting stats, there are no clear, objective standards to serve as a basis for comparison. We are forced to compare them to conventional wisdom, and the fielding statistics produced thus far don't follow that wisdom very closely. I've done my share of bashing the baseball establishment, but I have to doubt whether that many people can really be as wrong as some fielding statistics say they are. They could be, or it could be that the statistics aren't being used correctly.

I. Overview of the method

Like Palmer's system, DFTs depend on the basic defensive statistics kept by organized baseball, not the ratings and information that has been accumulated in recent years by Elias or STATS. The biggest difference is the formulation of a sort of "park factor" for each position and team, a factor that depends not so much on the park, but on the teammates of the player involved. In principle, this "fielding factor" is an estimate of how many more, or fewer, real chances this player is likely to have had relative to the other players at his position. Only with that knowledge can you make a reliable use of the defensive statistics.

So what goes in to these factors?

1. Balls in play. The single biggest error in Palmer's work, in my opinion, is that it uses the game - 27 outs - as its rate counter. This is wrong, wrong, wrong, because every team, no matter how good or bad it is in the field, is eventually going to get 27 outs. Errors are the only way he'd differentiate a good team from a bad one; but in most cases it's not the errors, it's the hits that are getting through that define a bad defensive team. So I estimate the number of balls the team's pitchers allow into play, essentially at-bats minus strikeouts minus home runs (which is, in essence, James' Defensive Efficiency Rating), to get a much better foundation for the system. I do take the overall park factor into account here, an obvious flaw that I overlooked before the book went into print, which will cause some numbers I present here to differ from those in this year's book.

2. Ground ball/fly ball teams. Some pitchers, and consequently pitching staffs, have definite leanings towards one type of out or the other. There are ballpark tendencies that influence this; the Cubs and Red Sox almost always seem to have groundball-leaning staffs, while the Oakland As of the `70s and `80s put up the highest flyball ratios in history. Team groundball-flyball ratios can generally be estimated well by comparing total infield assists to total outfield putouts, but it is impossible to be certain whether a given ratio results from a genuinely uneven distribution of ground balls and fly balls or a genuine defensive imbalance between the two. In handing out the ratings, I assume that both are at work evenly; if a team's ratio is .90, I rate the team at .95. In biblical words, I split the baby.

3. Left/right splits - Primarily in the modern game, the handedness of the pitching staff affects the distribution of balls towards the pitcher's opposite hand. Lots of left handed pitching usually means more chances for third basemen, less for first basemen, with a smaller effect on second basemen and shortstops.

Currently, the ratings are set by considering the entire pitching staff as a whole. It is certainly conceivable that treating the pitchers as individuals, and summing their individual contributions, would result in different team ratings, and that's an area for future research.

The three factors above produce a rating for a position, like .91, that I interpret as a chance modifier; a player who played in 100 real games only got as many chances as an average league player would have gotten in 91 games, so we shouldn't be surprised that his defensive statistics are lower. It is no surprise to me that league-leading assist performances are often accompanied by ratings of 110-120 relative to their peers.

II. Dave Concepcion Dave Concepcion, shortstop for the Big Red Machine in the 70s, was considered by the people of the time as a terrific shortstop. He won five Gold Gloves in the six seasons from 1974-79, missing only in 1978; any chance at a Gold Glove after '79 had an Ozzie Smith-sized roadblock in the way.

Palmer's Fielding Runs doesn't think too much of him. During his five Gold Glove seasons he is credited with +57 runs, a respectable amount, but nothing special. Only once did he lead NL SS in FR. More remarkably, FR says he went downhill dramatically after the '82 season, charging him with a staggering -82 runs in 1983-85. For his career, Total Baseball says, he totaled -36 runs. Even without the last three seasons as a regular, he would only total +46 runs, or about +4 per 162 games.

These two things don't appear to have much in common. Which one do you believe is closer to the truth?

I think FR are way off base here. The Reds of the mid-70s were an outstanding team, both offensively and defensively. One of the problems that comes with FR's adherence to using the "game" as a time unit is that players are judged as much against their team as against the other players in the league. A good fielder on a poor fielding team will be rated higher, by Palmer's methods, than an equal fielder on a good team. The 1974-77 Reds featured four Gold Glove winners, which suggests the observers of the time certainly thought they were a great fielding team, and would lead to the entire squad being underrated in FR. The story as I see it is of a player who was always playing in an environment that was not favorable to piling up large defensive numbers, with fielding factors in the mid-90s during his Gold Glove years, and adjustments into the 80s during the `80s. The picture I get is very different from Pete Palmer's:

"Difficulty" is the park factor, if you will, for Cincinnati shortstops. "Player rating" is the DFT, as a rate of performance, with 100 being average, 86 absolutely terrible, and anything over 110 being absolutely wonderful. Runs above average is how my rate statistic, and his playing time, would convert into an FR-like number.

	Diff	Player	RAA	Palmer's DR
Year	Rating	Rating	(Me)	(Web)
1970	101	96	-5	-8
1971	102	95	-7	-6
1972	100	103	5	3
1973	106	105	7	2
1974	95	113	31	9
1975	99	111	22	9
1976	96	108	19	17
1977	91	113	29	8
1978	95	104	9	-5
1979	98	105	12	14
1980	91	102	4	-13
1981	89	107	11	4
1982	96	104	9	13
1983	87	106	12	-21
1984	89	97	-4	-30
1985	88	100	0	-31
1986	100	106	4	-2

Notice that there is a strong, though not perfect, correlation between the difficulty factor and the difference between me and Palmer. Concepcion played almost his entire career in an environment that depressed his defensive statistics as much as the 1960s Chavez Ravine suppressed hitting. He starts off as a slightly below average SS, and gradually improves for a few years before breaking through at age 26, rides a very strong four-year peak before falling back a notch, and continues at a roughly +10/year pace until falling back to average in his last two seasons as a regular. In other words, a very normal career progression, with everything easily explicable, including the Gold Gloves and keeping his job in the early 1980s. A rating that incidentally judges his career to have been not an extremely ordinary fielder, but an impressive +153, placing him amongst history's elite (11th on the all-time list).

Nor is Concepcion an isolated case. Here are some others where the difference between FR and my RAA differ by at least 50 runs:

Player			Me	Pete	Me-Pete
Gene Alley		8	70	-62
Dave Bancroft		54	204	-150
Jack Barry		-3	-69	66
Dick Bartell		40	160	-120
Larry Bowa		-8	-89	81
Al Bridwell		-58	50	-108
Don Buddin		-50	6	-56
Bert Campaneris		21	-93	114
Leo Cardenas		46	-30	76
Chico Carrasquel	60	-19	79
Dave Concepcion		153	-35	188
Frankie Crosetti	25	-26	51

All before I even get to the letter D.

So who's right? I don't know.

III. My Hall

I'll provide a long list of shortstops and their assessed fielding performances at the end. For now, I'm going to skip to combining the fielding and batting levels of these players to generate my own list of who was the best overall shortstop in major league history. OK, so "the best" isn't a debate. The list after that might provide some, though. I did it like this:

1. I only looked at players for the years in which they were a regular SS, i.e., they led their team in games at SS. Time spent as a backup or playing some other position gives you no credit.

2. The value I used was batting runs above replacement offensively, or BRAR (defined as a .230 EQA, where .260 is league average or 1.2 runs per 27 outs below average) plus fielding runs above replacement defensively, or FRAR (where 86 is a replacement level fielder -- think Hubie Brooks).

3. A player's best five seasons, in terms of total RAR, were triple counted, to emphasize a high peak (i.e., they count once in the total, and then twice more).

4. Military service was treated by giving the player 75% of his average annual value in the two years before and after service.

How many people should be in the Hall, anyway? I think it's fair to have called the best two SS active at any given time to be Hall of Famers; over most of baseball history, that means about 1 Hall of Fame SS per eight teams. Let's spread that over twelve-year careers, and call it one Hall of Fame SS per 96 team-years. Since 1893, there have been 1958 team-seasons, which suggests 20 Hall of Famers. Don't you love even numbers? Some of them are active or very recently retired; I'll note the ones who have already reached the top 20 level all-time.

An easy number one is still Honus Wagner. His bat earns him 1026 BRAR, one of the all-time great scores for any position, without including his years at first base, outfield, or as a multi-position star, which he was before settling on SS, and 400 more than any other SS. His fielding RAR, 533, is seventh-best all-time. He has the best peak by far at 706. His total, then, only as a SS, is 1026 + 533 + 706*2, or 2972.

The second best is still active, but no longer plays short. Cal Ripken gets 589 for his bat (3rd among SS), 557 for his glove (6th), and a 554 peak (3rd) for a total of 2254.

Third, and shamefully overlooked for the Hall for many years, is Arky Vaughan. His 626 BRAR and 571 peak rank second only to Wagner. He only netted 371 runs afield, which is why Cal passes him. Total: 2139.

Fourth best is the man long named as the American League's top SS ever, and prior to Cal I guess I'd agree. Joe Cronin was absolutely phenomenal afield in the early 1930s, though his numbers collapsed after moving to Fenway. But he was a great hitter, amassing 513 BRAR, a 410 FRAR, and a 4th-best 527 peak, for a total of 1977.

Fifth is, unbelievably in a complex system like this, a tie. Luke Appling was a wonderful leadoff hitter for many years, and earned 551 BRAR - more than any SS outside the top 3. His fielding was slightly above average, at 453, but his peak was much less than Cronin's, at only 448. He also gets a 40-run bonus for his service in the Second World War, giving him a total of 1940.

Also totaling 1940 is my highest-rated non-Hall of Famer, turn of the century shortstop Bill Dahlen. Often paired nowadays with similarly excluded George Davis, he is, in my eyes, clearly the better player. His bat is lighter than anyone above him (468 BRAR), but he has the best glove yet seen at 624. His peak totaled 424.

In the seven spot is, in my opinion, the best fielding SS of all time. Ozzie Smith's bat only earned him 333 BRAR over his career, but the 694 FRAR - 70 more than second-place Dahlen, and more than anybody except Wagner got with their bat - offsets much of that. His peak of 412 was fielding-driven to a large extent, and produces a total of 1851. Welcome to my Hall, Mr. Smith.

Eighth, despite a short career, is Lou Boudreau. An outstanding hitter and fielder, he picks up 449 with the bat, 364 with the glove, and a very impressive 521 peak for a total of 1835.

I think that #9 will be a controversial pick, not for the person but for being this high. His numbers don't really stand out in any way, 417 batting, 408 glove, 403 peak... but I figure that he's missing 100 from his batting numbers, and 55 off his fielding, for three years of service in the Second World War. Pee Wee Reese is comfortably above my Hall line even without that bonus, pushing him up to a total of 1786.

Rounding out the top 10 will be another real old-timer, longtime Brownie Bobby Wallace. As forgotten today as Dahlen, he would have deserved most of the AL Gold Gloves after the turn of the century if they had been thought of then, and piled up a 10th-best 521 FRAR. He wasn't a bad hitter, but not especially good either. He winds up chalking up 347 BRAR with a 442 peak, totaling him up at 1752.

Eleventh, a man who perhaps deserves to be higher because I have totally ignored what was virtually a second career as a first baseman, is Ernie Banks. His 444 BRAR and 254 FRAR are low, but a fifth-best 521 peak brings his total up 1740. An easy Hall choice.

Now we're to the point that choices will become a little more controversial, and my pick for twelfth is no exception. He's not in the real Hall, and isn't considered a strong candidate for it. I see his 586 FRAR as the fifth best of all-time, and his 424 peak is reasonably strong, despite just racking up 257 BRAR. This is the man I've already discussed at length, Dave Concepcion, scoring a 1691.

Thirteen isn't in the Hall either, but probably will be. In my eyes he was a good hitter and an average fielder, getting 443 bat, 399 glove, and a 420 peak, which includes a MVP award that he should have won. Step up, Alan Trammell, you've earned it with a 1682 score.

Fourteen is another glove man. His glove got him a third-best 619 FRAR, but his bat was the weakest of my Hall worthies at just 224. Likewise, he peaked at a sub-standard 398 (mostly from fielding). He's considered one of the real Hall of Fame's weakest choices, but Rabbit Maranville's 1639 makes mine.

At fifteen and climbing is Barry Larkin, who's already established his value in this little pantheon. So far, he's picked up 408 at bat, 320 with the glove, and has had a 454 peak for a total of 1636. A good year in '98 will carry him past Concepcion.

Vern Stephens gets the 16th slot. He was certainly not as bad fielder as anecdotes seem to make him out to be, usually in comparisons with Rizzuto. For me he rates as a little above average there. He had a booming bat, making for a relatively high 456 peak to go with 386 batting and 306 fielding - the lowest fielding score in this group. That's a 1604 total, though, and gets him in.

A step below, in 17th place, is Hall of Fame whipping boy Joe Tinker. Elected by a poem, some say. Well, his 522 FRAR are eighth on the all-time list, and his career fielding rate of 110 ranks first. He wasn't a strong hitter - I give him 259 - and his peak is only 402. But that's good for 1585, and that *doesn't* count his Federal League time. He's in.

Eighteenth is the man who, like Banks, had a long career at other positions. Robin Yount isn't getting any credit for his outfield years, but 370 BRAR, 317 FRAR, and a 435 peak just in his career as a shortstop gets him a combined score of 1557, and a spot on my list.

Nineteenth is a paradoxical player. He was a great fielder - DFTs give him seven Gold Gloves on his career, and his 590 FRAR are fourth all-time - but his career rate was only 102. Regarded in his day as a wonderful leadoff man, he was a terrible hitter, with only 284 BRAR in a very long career. His peak score, 341, is horribly low for a Hall of Famer. It is all just enough for me to get Luis Aparicio into my Hall with a 1556 mark.

And finally, in 20th place, is a man I thought wouldn't make it, because I gave his greatest asset, fielding, so much less value than Palmer did. Still, Dave Bancroft had 435 FRAR, to go with 291 batting and a 392 peak, and sneaks into my top 20 with 1510 - which, incidentally, makes 1500 a nice round number for Hall of Fame qualification.

How about the Cooperstown picks who didn't make mine? Well, Travis Jackson is a bad joke, with 252 at bat, 230 at glove, and a 332 peak for just 1146. His career comes in somewhere around 45th place.

Phil Rizzuto is a near miss, coming in 22nd for me. With just 224 at bat, 375 in the field and a 372 peak, plus a 140 bonus for World War II leaves him at 1483. I could see an argument that gives him 100% of the seasons he missed during his military service, which would net him about 50 more points and put him ahead of Bancroft, but I think he pretty much defines the proper bottom edge for Hall of Fame shortstops.

Hughie Jennings' career as a player was just too short. A regular SS for just six seasons, he gets 239 bat and 273 field, and 478 of that 512 total counts as peak. Considering he was playing 132-game seasons, that rivals Vaughan for second-best peak ever. He racked up a 1468.

Joe Sewell doesn't get any extra credit from me for avoiding the strikeout. 293 BRAR, 271 FRAR, and a 398 peak, totaling 1364, doesn't come particularly close.

What about George Davis? Where is he? With me he finished right between Rizzuto and Jennings, with a 1469 mark. He was sort of a reverse Yount: he started off in the outfield, then moved to the infield, spending a couple of years at third before going to short. He was a good, but not great, hitter, and picked up 395 BRAR for his SS years (and about 300 more from his other positions). And, yes, he missed the entire 1903 season because of an interleague dispute, a season that might well have earned him the 60 he needs to move past Bancroft. He does not emerge from my system as a good fielder... just average, with a 270 mark. His peak, 402, is very good, but not in any way remarkable. He probably should be in the Hall, but his performance at shortstop alone doesn't do it for me.

Bert Campaneris is another near miss. Like Concepcion, I give him considerably more for his fielding than TB does, but a 325 BRAR, 423 FRAR, and a 369 Peak comes out to 1486 and is just a little bit short. Donie Bush had a long career, a great batting eye, and was terrific his first couple of years in the field, but his fielding numbers dropped steadily for the rest of his career. He finished at 283 BRAR, 400 FRAR, a 391 Peak, and a total of 1465.

How about Ray Chapman? He had seven seasons as a regular SS for the Indians, picking up 229 BRAR and 182 FRAR. Defensively, he was behind Peckinpaugh and Scott, even if he was a much better hitter. He seems to have been fragile, missing considerable time almost every year, and only gets a 325 peak. That's just 1061. The what-if game says he may have been able to double the bat and field numbers, and perhaps up the peak to 375, and that would push him up to 1572, which puts him in my Hall, but not as a slam-dunk candidate.

IV. The Main Table

In this chart, I've summarized the players career fielding accomplishments. "Years" refers only to the player's years as a regular shortstop for his team; a notation such as Alley's 1965-72x69 means that he was the regular from the years 1965 to 1972 inclusive, except for 1969. In cases where a player was on two teams in one year, only the primary team is counted; the only exception I know of is the Bartell/Fletcher trade of 1920. Stats were counted for both teams in that one.

"Rating" is my rating of the player, with 100 = league average. If it seems that most players are above 100, relax; the list only includes players who were their teams regular shortstops for at least five seasons, and you'd expect that anybody who stays on that long was above average.

"RAR" is the number of career Runs Above Replacement the player earned for his rating. A replacement player has a fielding rating of 86, and an RAR of (obviously) zero; an average shortstop would have about 30 RAR per 162 games, although this number is slightly variable over time (especially in pre-1920 ball, when it was as high as 35).

"RAA" is Runs Above Average (average being a 100 player), which is the standard used by Palmer.

"Palm" is Palmer's Fielding Runs for only the years in question. Data was from Total Baseball II, except for players who were rated in years 1991 or later; for those players, the Palm value for their entire career came from the TB website (www.totalbaseball.com). Palm data for 1997 was not available, so the RAA and Palm for active shortstops are not completely comparable.

"Notes" is other things I wanted to add. "AGG" is Actual Gold Gloves; there weren't any before 1957. "CGG" are the seasons for which I would have rated this player the best in the league; this was determined by a combination of RAR and rating (2*rate+RAR, so a SS with a 110/41 season would get the nod over another who was 108/44).

It was apparent from making this chart that there is a time-dependent character to the ratings that I have not removed. A 110 rating was comparatively easier to achieve the farther back in time you go; turn-of-the-century leagues always had one or two players who were so much worse than everybody else that the rest of the players had their ratings inflated (since the league, as a whole, is always 100). While the occasional experiment with a Hubie Brooks still takes place, the top-to-bottom disparity is never so great, anymore, as it used to be, which is consistent with what Stephen Jay Gould has to say about changes in player populations over time.