Batter's Box Interactive Magazine Batter's Box Interactive Magazine Batter's Box Interactive Magazine
Evaluating shortstops is toughest of all. Defence is recognized to be very important, and yet our tools for evaluating defence are rudimentary. Fortunately, for most of today's candidates- Derek Jeter, Alex Rodriguez, Nomar Garciaparra, Barry Larkin and Julio Franco, defence is not the major issue. With that in mind, let's look at who is in the Hall and who is not.

OFFENCE

I'll divide up the prospective and actual Hall of Famers by their hitting ability:

The hitters

These guys hit so well that they only need to be average fielders to make the Hall, but most were much better than that- Honus Wagner, Arky Vaughan, George Davis, Joe Cronin, Robin Yount, Luike Appling, Lou Boudreau, Hughie Jennings and Ernie Banks. There was little doubt about them when their names came up. Rodriguez, Jeter and Garciaparra are on pace to be in this class. Larkin arguably falls within it, as well, but he can be put in the next class as well.

The combo guys

These ones were above average hitters, but had to be considered above average fielders to be considered. Bobby Wallace, Joe Sewell and Pee Wee Reese are in from this group, but many of the best hitters are not in from Bill Dahlen to Vern Stephens to Alan Trammell. Julio Franco fits neatly in this group. Cal Ripken, believe it or not, fits within this group, but is not yet eligible for the Hall.

The fielders

These shortstops were below-average hitters over their career, but made it with their glove- Ozzie Smith, Phil Rizzuto, Luis Aparicio, Rabbit Maranville.

The incomprehensible ones

Travis Jackson and Dave Bancroft were admitted to the Hall, but why they were is a mystery.

As defence is perceived to be so important here, I've chosen to use evaluation tools that are amenable to combining offence and defence. I'll use Lee Sinins' Runs Created Above Average to break down the retired shortstops by their offensive ability (all figures courtesy of Sinins' Sabermetric Encyclopedia):

the hitters-200 RCAA career or more (all in the Hall of Fame)
Player        RCAA
Wagner 1011
Vaughan 478
G. Davis 379
Yount 284
Cronin 243
Appling 239
Banks 207
Jennings 206
Boudreau 202


the combo guys 0-200 RCAA career-Hall of Famers
Player        RCAA
J. Sewell 124
B. Wallace 37
Reese 13


selected combo guys 0-200 RCAA career-non-Hall of Famers
Player        RCAA
Glasscock 188
Dahlen 186
Trammell 161
Stephens 157
McKean 131
Fregosi 128
Kuenn 111
Wise 94
Pesky 90
Hemus 89
Travis 76
Petrocelli 41
Fernandez 35


the fielders-Hall of Famers
Player      RCAA
Rizzuto -10
Tinker -38
Ozzie -80
Aparicio -180
Maranville -280


DEFENCE- THE NUMBERS

Before I start, a warning is appropriate. All defensive numerical evaluations from historical stats are rough approximations.

Unfortunately, there is no Runs Saved Above Average stat for fielders, but we can attempt to fix a range of values. In the New Historical Baseball Abstract, Bill James pointed out that shortstops on average across many decades average 28% of team assists, and that no player's teams over a career has exceeded the average by more than 504 (Ozzie Smith). Let us assume that all of the 500 extra assists can actually be attributed to the shortstop, as opposed to the particular circumstances of the team (3rd baseman, handedness of the pitching staff...), and that each of them would otherwise be a single. How much would that be worth in runs?

We'll try to make an estimate using Tangotiger's 1999-2002 run expectancy tables. If nobody is on or out and the ball is hit toward the shortstop, the difference in run expectancy between an assist (.297) and a hit (.953) is .656. If nobody is on, and there is one out, the difference is .456. If nobody is on and two are out, the difference is .251. It seems safe to consolidate the figures at .45 run/assist.

With a runner on first, the calculation becomes more complicated for the following reasons:

1. we do not know if an assist will lead to a 6-3 out at first, a 6-4 fielder's choice of a 6-4-3 double play, and
2. we do not know if a single will end up as a 1st and 2nd or a 1st and 3rd.

The range of difference is between .466 and 1 run/assist. We can safely consolidate at .73 run/assist.

With a runner on second, the range is between .538 and 1.228 run/assist. We can safely consolidate at .88 run/assist.

The figures increase with multiple runners on, reaching a high of between 1.815 and 2.538 for a bases loaded 2 out situation. Most plays occur with nobody on or one runner on, so I estimate the value of an assist (the out and the lack of a baserunner) at somewhere between .6 and .8 runs.

So, that would mean those 500 assists are worth between 300 and 400 runs; it's a safe bet that they would be worth many fewer runs in the 1960s or in the teens than they are now or as they were in the 20s. Now, there are other elements to shortstop defence- ability to turn the double play and fielding percentage being the two most important. It does seem likely that the very best defensive shortstops could save conceivably 500 runs in a long career above an average one.

The NHBA figures suggest that the numbers are smaller. Ozzie Smith is credited with 329 Win Shares, of which .43 are defensive, for a total of 141.47. 3 Win Shares represent an actual win, so dividing 141.47 gives 47.16. Ten runs is roughly equivalent to a win, so that would have Ozzie at 472 runs saved. Here are the figures for other players using this methodology:


Maranville 430
Wagner 480
Dahlen 470
Reese 368
Wallace 440
Vaughan 263
Banks 222
Cronin 330
Trammell 305
Ripken 447
Scott Fletcher 215
Felix Fermin 108


One has to bear in mind that all defensive Win Share numbers are positive, so the average shortstop will save some runs. Fermin achieved his 108 runs saved in 900 career games. Fletcher's and Fermin's stats appear to be roughly league average, so it is a fair guess that 200 runs saved for a 1800 game career is about average. I've put in Ripken too as a reference point for Trammell.

Putting it all together, I get the following scale:

good solid shortstops (e.g. Banks, Trammell,)- 5-10 runs saved above average/150 games at short
excellent shortstops (e.g. Wallace, Dahlen, Ripken)- 10-20 runs saved above average/150g
the best (Ozzie, Wagner)- 15-25 runs saved above average/150 g

The difference between the excellent and the best is unlikely to be more than 2-3 runs/season.


Summing up

The best hitting shortstops of all time are in the Hall of Fame. The best fielding shortstops, with the exception of Marty Marion (who had a short career) and Bill Dahlen (more on that later), are in the Hall of Fame. It's the fine, but not great, hitters, who are also solid fielders who tend to be on the outside. Trammell and Vern Stephens would be the classic examples of this kind of player. On my club, I'd rather have Trammell than Joe Tinker or Joe Sewell or Luis Aparicio or Rabbit Maranville, and the numbers do suggest that the sum of the runs he created on offence and saved on defence exceed what the others were able to do in combination:

The calculation goes as follows:
              Offence           Defence        Total
Trammell 161 70-140 231-301
Maranville -280 140-280 -140-0
Tinker -38 120-240 92-202
Aparicio -180 170-340 -10-160


Why is Bill Dahlen not in the Hall of Fame?

Here is Bill Dahlen's statistical line. He is one of the top 15-20 offensive shortstops of all time, and one of the top 5 defensive shortstops, and he had a long and productive career. Was he simply overshadowed by Wagner, or what? Perhaps he was just forgotten when the Hall elections began in the 30s. Or maybe it was because he was "Bad Bill".

In Part 2, we turn to today's contenders- Alex Rodriguez, Derek Jeter, Nomar Garciaparra, Barry Larkin and Julio Franco. We'll run a full chart then.
Hall Watch 2004-The Shortstops-The Standards | 21 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.
_Paul D - Wednesday, December 08 2004 @ 10:42 AM EST (#10834) #
Interesting article, but I think you need to add Rizzuto to the incomprehensible list. Or the, "Only in the Hall because they played for the Yankees and tons of people lobbied for them, despite the fact that they don't deserve to be in the Hall" list.
Mike Green - Wednesday, December 08 2004 @ 11:42 AM EST (#10835) #
I didn't add Rizzuto to the incomprehensible list because his defence, especially his ability to turn the DP, appears to have been superior. It's very difficult to measure how much weight to attach to the defence, and it is conceivable that he saved 250-300 runs on defence in his career.
_elston - Wednesday, December 08 2004 @ 01:58 PM EST (#10836) #
Guys, take it easy on the Scooter. He was voted in by the Veteran's committee which doesn't come with the same status as having been voted in by the writers. Moreover, there was a lot of lobbying and pressure on the vet committe from the NY media to induct him because his old rival Peewee Reese had been selected previously. Back in their time, there was a lot of debate over who was better and many thought that the Scooter was. Aside from all that, the guy did have a great career. He was the Yankee SS during their most dominant period in the '40's and 50's, several times an all-star, was a great world series performer and after his playing days was a very popular and much listened to broadcaster. Overall, he'd probably spent at least 50 years in baseball. You can look up how many pennant and WS rings he has and I would bet he has more than any of the other Hall of Fame SS's. He also did military service during the early days of his career which probably affected some of his numbers. Sometimes you have to allow sentiment into the equation when it comes to selecting people like the Scooter because of the affection people close to the game held for him and because he was on the scene for so long.
_Paul D - Wednesday, December 08 2004 @ 03:53 PM EST (#10837) #
Well, I'm basically basing my disdain for Scooter based on The Politics of Glory by Bill James, in which he makes it pretty clear that Scooter is very underserving, and questions whether or not he was considered a good defensive shortstop at the time, or if that happened later.

I don't give any weight to how popular a player was, whether he played for the team when they were dominant, or what happened after he retired. I also don't care how many pennants or World Series a team won, and I don't believe in allowing sentiment into the equation. I don't have a problem if you feel that way, but I don't consider any of those when evaluating a player's Hall worthiness.
_tangotiger - Wednesday, December 08 2004 @ 04:21 PM EST (#10838) #
http://www.tangotiger.net/customlwts.html
Mike,

Good work. I'm glad you went to the trouble of doing all the intermediate steps to get the run value of a play.

The short answer is about .75 runs.

A single is roughly equal to .45 to .50 runs (depending on the run environment). The out is roughly equal to -.25 to -.30 runs. Essentially, it works out to .75 runs or so. Your intermediate steps makes it clear why it is what it is.

You can use this chart:
(COMN)
if you want more precision by run environment.

Tom
_tangotiger - Wednesday, December 08 2004 @ 04:28 PM EST (#10839) #
Oh, and you are right that WS is not an above average scale, so you can't directly compare the two. I believe that the average SS gets 39/162 * (243*.17) Win Shares. That's 39 parts of 162 fielding parts for the SS, and 17% of all 243 WS are fielding win shares. That's 10 WS for the average fielding SS per 162 GP. (Hmmm... sounds high... I must have I made a mistake somewhere).

Ozzie played in roughly 15 complete seasons, so an average SS would have 150 WS, which is higher than Ozzie's total.

Anyway, that's how you'd figure it. Just have to figure out what that "39/162" term should actually be.
_Magpie - Wednesday, December 08 2004 @ 04:46 PM EST (#10840) #
I'm basically basing my disdain for Scooter based on The Politics of Glory by Bill James

By the time he did his new Historical Abstract, James had revised his opinion of Rizzuto's defense:

"...the best shortstop ever at turning the double play, almost beyond any dispute, was Phil Rizzuto. This is one reason Rizzuto ranks where he does here, the 16th best.... In another book I wrote at length about Phil Rizzuto. I did not know this at that time."
_tangotiger - Wednesday, December 08 2004 @ 04:49 PM EST (#10841) #
The correct term should be 36/200, and not 39/162. That makes the average fielding SS WS per 162 of 7.44. Over 15 seasons, that's 112.

That Ozzie gets 141 makes him 29 above average. That works out to almost 10 wins above average, or about 100 runs saved above average. Certainly sounds pretty low, but, it's probably in-line with Win Shares misvaluing fielders and pitchers anyway.
Mike Green - Wednesday, December 08 2004 @ 04:51 PM EST (#10842) #
Thanks, Tom.

I'll use the custom linear weights tool next time to analyze this kind of problem. Sometimes re-inventing the wheel is a positive learning experience.:)

Mike
_tangotiger - Wednesday, December 08 2004 @ 05:02 PM EST (#10843) #
Yes, definitely. The best way to buy into linear weights is to do those intermediary calculations. Once you realize that, you'll never go back.
Mike Green - Wednesday, December 08 2004 @ 05:07 PM EST (#10844) #
I agree that 100 runs saved for Ozzie over his career sounds low. He had 500 assists more than average, and that is giving him credit for just over 1/4 of them.

His best numbers were achieved (not surprisingly) in the first 5-7 years of his career, and they were relatively consistent over that period. He split that time between San Diego and St. Louis. The Padres did have a couple of soft-tossing lefties in the rotation while he was there (Randy Jones/Owchinko/Shirley/Curtis at various points), and their third basemen changed every year. In St. Louis, the rotation was right-handed, save for Dave LaPoint, and Ken Oberkfell was a respectable defensive third baseman if my memory serves me well.
_elston - Wednesday, December 08 2004 @ 07:44 PM EST (#10845) #
Just a few more points regarding Rizzuto. Casey said at the time that he was the best defensive shortstop he'd ever seen. Ty Cobb said he was one of the best bunters he'd ever seen. He was 2nd to Ted Williams in the 1949 MVP voting, won the MVP in 1950 and won the World Series MVP in 1951. These points don't necessarily qualify him for the Hall, but on the other hand neither should they evoke disdain.

In his time, the Scooter was better than most of his peers over a fairly long period of time and an ultimate team man. I remember towards the end of his career when Don Larsen was pitching for the Yanks. Larsen was a good hitter and a on a few occasions, Casey put him hitting 8th in the line-up with Rizzuto 9th. Rizzuto never complained and I always had the deepest respect for that. So, I got no problem with the Scooter being in the Hall, even if his stats don't qualify him by today's standards.
_Paul D - Wednesday, December 08 2004 @ 07:56 PM EST (#10846) #
the best shortstop ever at turning the double play, almost beyond any dispute, was Phil Rizzuto. This is one reason Rizzuto ranks where he does here, the 16th best

Wow, I didn't know that. I guess that's what I get for shooting my mouth off and for not reading more Bill James.

I still don't see that that qualifies him for the Hall, but whatever, I've already shown enough ignorance today.
_Magpie - Wednesday, December 08 2004 @ 08:53 PM EST (#10847) #
I still don't see that that qualifies him for the Hall, but whatever, I've already shown enough ignorance today.

Oh, I don't think anyone really thinks Rizzuto is in the HoF because of his DP skills.

It's the poetry!

Hero or the Goat

All right, this is it,
The whole season coming down
To just one ball game,
And every mistake will be magnified,
And every great play will be magnified,
And it's a tough night for the players,
I'll tell ya.
I know last night, being in the same situation many times
With the great Yankee teams of the past,
you stay awake,
And you dream,
And you think of what might be,
If you are the hero or the goat.
_tangotiger - Thursday, December 09 2004 @ 10:07 AM EST (#10848) #
Ok, I looked at the 2004 Win Shares. SS got 16% of all fielding win shares. Fielding WS were one-sixth of all Win Shares.

So, that gives us:
total WS = 81 * 3 = 243
fielding WS = 243 / 6 = 40.5
SS fielding WS = 40.5 * .16 = 6.5

If Ozzie has played 15 seasons of 162 games of 9 innings, then an average SS would have 6.5 * 15 = 97.5 fielding Win Shares.

Ozzie, according to Mike's numbers above, had 141.5 fielding WS. That makes him +44 career fielding WS above average. That's about +15 wins, or +150 runs saved.

That's +150 runs over 15 full seasons (or an average of 10 runs saved per 162 games). I think it's fair to say that Ozzie, as well as all other great fielders, are severely shortchanged by Win Shares.

My analysis says that great fielders are worth +20 to +30 runs per season above average. How can I tell? When you compare the great SS to the average SS, the great SS makes +.05 more outs per play than the average one. That is, if the ZR for an average SS is .85, the great SS is .90, etc. This is essentially true at all positions.

Anyway, a SS has 5 plays to make a game, or about 800 a season. At +.05 plays, the great SS makes +40 more outs per 162 games. As discussed, +40 more outs is equal to +30 more runs. That's essentially the top limit of a great fielder. Even if you argue that Ozzie was only at the top of his game for 5 years (say +25), near the top for another 5 (say +15), and ordinary for the last 5 (say +5), that's still an average of +15 runs per season. And that's a conservative estimate.

If as Mike is saying Ozzie made 500 more assists than the average SS, that's +33 plays per 162 GP, or +25 runs per season.

For WS to give Ozzie credit for being +10 runs per season above average, when in all likelihood he was double that, should force more analsysts to really dig into the validity of Win Shares.
Mike D - Thursday, December 09 2004 @ 10:59 AM EST (#10849) #
Interesting, Tango -- and compelling. Where do you stand on the various (admittedly fledgling) defensive metrics?
_tangotiger - Thursday, December 09 2004 @ 11:56 AM EST (#10850) #
Any metric that uses actual data, as opposed to estimated data, has to be superior.

Play-by-play data knows exactly how many lefty batters Jeter has faced, exactly the groundball tendency of the pitcher/batter combination, exactly if there was a runner on 1B,2B,3B and outs, and if a bunt is the play. Davenport, Saeger, Palmer, Humpheries et al all have to try to estimate these things. So, that's a huge knock on them.

Play-by-play data relies on observers to establish if a ball in the air was a pop, liner, or fly. The non-PBP systems try to estimate this. I'd score this one for the observer.

Play-by-play relies on observers to establish if the ball was hard hit, medium hit, or soft hit. The non-PBP systems don't even bother to establish this explicitly, though they may try to work with this implicitly. If the observers don't go a good job, this can make the play-by-play less reliable. MGL tells me that the impact of this parameter is only a couple of runs anyway. So, at worst, this is a push.

Play-by-play will tell you exactly what parks the player played in. Non-PBP estimate this as well.

Play-by-play also relies on observers to establish in what grid location on the field the ball passes through or lands. Same as above, non-PBP systems try to figure this out in some form or other. They need to know how many balls are hit "in the area of responsibility". PBP, even if they can't tell you exactly what zone, will at least tell you if it's anywhere close to the player, which is all the non-PBP systems can hope for anyway.

Do you realize that a non-PBP system also can't even tell you how many batted balls a fielder cleanly fielded? You might think "Assists for infielders, putouts for outfielders", and you'd only be mostly right. Relay throws with an out on base gets an assist for your middle infielder. So, assists included batted balls and thrown balls.

A PBP system will also tell you if the fielder was part of a relay out, but will NOT tell you if he was part of a relay safe. However, a non-PBP system won't know that either.

In this day and age, there is zero reason to rely on a non-PBP system for any player from 1989 to today.

If you want to construct a non-PBP system for 1988 and earlier, then fine. But, it's completely wrong to continue to use that same system because you want to use one system. It does NOT make the comparisons fairer. If you do believe that it makaes the comps fairer, then you are saying that the bias in the non-PBP system gets to be carried over in pre-89 and 89-and-post. If there is a systematic bias like that, why in the world would you want to continue using it? If it's a random bias, then why not simply go with the PBP system?

When you look at UZR or Pinto's system, the problem is the reliance on sophisticated programming and mathematical techniques to parse through the PBP data. Some of the results of these systems are just incredulous.

If Astros fans are telling me that Jeff Kent is atrocious, as is Moises Alou and others, and UZR is saying that they are not, then something's gotta give. Ichiro is far more highly regarded by fans than is Wynn, which is the opposite of UZR.

Essentially, is the issue simply one of sample size, and anything can happen. Say, Brady Anderson is not truly a 50-HR hitter, and even though we credit him with 50 HR, no one believes he'll do it again. So, did Wynn really luck into saving 10 runs over an average CF, and Ichiro really unluck into costing 5 or whatever runs compared to an average RF?

Or, is it that the quality of the play, even after all the above parameters, still cannot be ascertained well enough? That is, say that the fans can honestly say that Ichiro was 20 runs better than Wynn in 2004 performance and in 2004 talent, then there is a huge hole that UZR is not covering. I don't know what the answer is. Yet.
Mike D - Thursday, December 09 2004 @ 01:09 PM EST (#10851) #
Certainly, figuring all of this out is a rolling process. But with careful and inquisitive efforts such as yours, I think we'll be in good hands.

Thanks, Tango.
Mike Green - Thursday, December 09 2004 @ 03:53 PM EST (#10852) #
Robert Dudek's study in the Hardball Times of hang-time and fly ball out conversion rates is, in my view, a seminal advance. Robert found 2 things of relevance to this discussion, as it relates to outfield defence: fly-balls with hang-time over 6 seconds were caught over 97 per cent of the time, and hang-time was more significant than rough zone for out conversion rate.

Over time, I expect that we will be able to analyze very well outfield defensive performance using hang-time and zone rating combined for range and other measures for arm.
_tangotiger - Thursday, December 09 2004 @ 04:21 PM EST (#10853) #
I agree that Robert's study was outstanding, and I was remiss in not mentioning it. I've been clamoring for hang time for a long time.

I suspect that once you get some sort of GPS tracking on players (and preferably balls, like the FoxTrack puck), whether by having lasers or videos focused on the players/balls, or whatnot, then we'll have all we'll need, and all we ever wanted: four-dimensional (length, width, height, time) data of balls and players.
Mike Green - Friday, December 10 2004 @ 11:30 AM EST (#10854) #
http://www.mcfarlandpub.com/book-2.php?isbn=0-7864-1978-4
I'm not alone in my view of Bill Dahlen. Lyle Spatz, SABR's Records Committee chair, has just written a book on the topic. It sounds like an interesting read. COMN.
Hall Watch 2004-The Shortstops-The Standards | 21 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.