When is this guy going to start to age? He doesn't look that much different than he did 25 years ago.

Is there explanation for Randolph's relative lack of inseason durability? He only has one season with more than 147 games played. That strikes me as one of the bigger warts in his case.

I would assume that out at the extreme edges of offensive shape, with OBP higher than SLG, that crude measures like OPS+ are severely inaccurate representations of his value and that even more comprehensive offensive metrics may be challenged.

Recipe for HoM 2Bman:
1. take one Nellie Fox
2. add a pile of walks
3. bake for a few extra years
4. serve Willie Randolph in first year eligible

As Bill James has spent his life saying:

"If, then" is not a good way to analyze the situation. The question is whether Randolph is really the next best candidate.

agree, sunny - I was merely enamored by the irony of Willie showing up right when Nellie was welcomed in.

bb-ref translated stats for Randolph (top) and Fox, two gold glove 2Bmen:

.. G .. AB .... R ... H .. 2B . 3B HR RBI . BB . SO .SB .Avg .OBP SLG OPS
2257 8346 1380 2379 342 .69 .55 .757 1345 689 296 .285 .384 .362 .746
2445 9554 1335 2769 370 116 35 .824 ..746 225 ..76 .290 .349 .364 .713

If Nellie were still eligible, Randolph would be higher on my ballot as of this morning. This surprises my gut feel; I never thot of Randolph as all that special. I think some of that is that by missing a few games per year, he never scored 100 runs in a season. But scoring 80 runs/yr while playing only 130 games and not making a ton of outs is pretty durn valuable to your team.

Don't forget that Randolph deserves managerial credit too ;-)

When is this guy going to start to age? He doesn't look that much different than he did 25 years ago.

There was a side-by-side shot of him and Franco in the dugout the other day. Franco (who's four years younger) looked at least ten years older than Randolph. Maybe fifteen.

Of course that is if you believe that Franco is actually four years younger. There is always the cahnce that he is actually pulling a satchel page on us and is like 55.

Wow, that's great, a photo of Randolph and Franco! My favorite player from the 1980s besided my favorite player from the 1970s!

bb-ref translated stats for Randolph (top) and Fox, two gold glove 2Bmen:

I love bb-ref's new toy. Let me add a third 2nd baseman to the list

(my apologies in advance if this doesn't line up right)

.. G .. AB .... R ... H .. 2B . 3B HR RBI . BB . SO .SB .Avg .OBP SLG OPS
2257 8346 1380 2379 342 .69 55 .757 .1345 689 296 .285 .384 .362 .746
2445 9554 1335 2769 370 116 35 .824 ..746 225 .76 .290 .349 .364 .713
1872 7082 1167 2174 344 140 83 .960 ..716 290 337 .308 .376 .431 .807

The bottom player is one of my best friends, Larry Doyle. Is it possible that some voters mark down Doyle for the runs he gave away in the field without crediting him for the runs he produced at the plate?

One item we ought to remember about Randolph, and guys like him, is that leadoff hitters get many more PA per game, but each PA is worth slightly less in run-scoring opportunity (fewer men on).

If you rate players per game played, this is moot.

If you rate players by WS/WARP/RCAP (actual PA), a player who batted leadoff his entire career should have his offense value lowered by about 5%.

Except that Randolph wasn't THAT much a leadoff hitter. He was primarily a leadoff hitter from 1980 through 1984, with a fair amout of leadoff in 1977 and 1979. Once Henderson joined the Yankees in 1985, Randolph batted some 2nd and some lower end of the order, and for the rest of his career he batted 2nd (including stints with the Dodgers with the leadoff hitter being either Alfredo Griffin or Juan Samuel), with some lower in the order.

Of course, Randolph was never destined to have all that many RBI, no matter where you batted him. Not with his TB/PA.

A quick check of Randolph's gamelogs on bb-ref indicates that he batted primarily in the leadoff spot 1980-84. In other seasons he was in the #2 spot or in the #7-#9 spots.

Andrew M, at least in my case, it's also that when Doyle played the depth of the 2B position was comparable to the depth of corner outfielders. 2B around 1910 were absolutely a dime a dozen. In value terms, Doyle would be comparable to Frank Howard or Greg Luzinski, except that he was a significantly worse hitter.

2B around 1910 were absolutely a dime a dozen.

Hey, just like SS in the 1970s! ; )

Frank Howard or Greg Luzinski? There's two guys with similar characteristics, except that Frank had a lot more of whatever it was. Frank's peak 38-34-30 is more like Al Simmons or Tim Raines or Willie Stargell or Charlie Keller. Luzinski's 30-28-27 is more like Jose Cruz or George Foster or Don Buford or Topsy Hartsel. Frank's "5" of 153 is more like, well, Simmons or Joe Jackson (better than Shoeless), Kiner or Sherry Magee or Rickey Henderson. Luzinski's 121, well, that's from a different planet, the one that Kirk Gibson and Buford and Kevin McReynolds and Lefty O'Doul and Tom Tresh are from (except Tresh is better).

In short, Frank Howard was Al Simmons. Luzinski was Tom Tresh. Other than that they're pretty similar.

Or IOW, Frank Howard is to Greg Luzinski as Willie Randolph is to Horace Clarke or Johnny Ray or Juan Samuel, at least in terms of peak.

The point is that they were both sluggers who were so bad in the field that they wound up being not That valuable. WS will grossly overrate both of them (not sure about their relative merits) since it doesn't have Loss Shares to capture their impact on defense. That said, I do agree that Howard was about 50% more valuable than Luzinski overall.

Dan R exaggerates on the level of 2B play. Where are the numbers to back that up? Doyle has more RCAP than either of the two OF-ers you mention despite playing in a lower run context.

I'm still one of Doyle's biggest fans. What's clear to me, though, is that the electorate is particularly dismissive of glove-men who have mediocre gloves. You really need a WS letter grade of B or above to get any traction as a 2B or SS candidate here. It doesn't help that WARP's wacko fielding replacement really hammers down these guys as well. Of course, understanding why they vote that way doesn't mean I don't wish the electorate would change their mind. :)

David Foss, you think I'd make an assertion without having the numbers to back it up? You should know me better than that by now. :) I did get a bit confused, I meant first basemen--corner outfielders were still hitting a bit better then.

Just as a simple measure, 2B and 1B in both leagues had the exact same positional EqA of .272 from 1906-15. Using more sophisticated data, the worst three starting 2B in both leagues averaged 1.0 wins below overall league average (by my measure) from 1906-15, while the worst three starting 1B in both leagues averaged 0.7 wins below overall league average. Nate Silver's Freely Available Talent levels are 0.5 wins per season below the worst three regulars average for 1B, and just 0.1 wins per season below the worst three regulars average for 2B. So my calculated replacement level for 1B in 1910 is 0.7 + 0.5 = 1.2 wins below average per season, while my calculated replacement level for 2B in 1910 is 1.0 + 0.1 = 1.1 wins below average per season. 2B and 1B were most definitely equal around 1910, any way you look at it. Corner outfielders were slightly stronger, with a .278 EqA and a replacement level of 0.9 wins below average.

DavidFoss, not true about poor fielders at key positions--I vote for Toby Harrah! Derek Jeter will get in one day, and so will Mike Piazza. I don't think Kent's good enough.

You should know me better than that by now. :)

I do. You've made the 2B-OF comparison before. I know you have the numbers, I was just asking to see them. :)

Well, the 1B comparison doesn't surprise me. 2B really is even with 1B in this era (especially between 1909 and 1914). I'm not sure if this is due to 2B being easier or just a down era for 1B. The 1B gap between Anson/Brouthers/Connor and Sisler has been discussed here before. The best MLB 1B of this era were Konetchy & Daubert (versus Collins, Lajoie & Doyle at 2B).

And my comment about poor fielders at key positions was directed more to the electorate as a whole and not specifically to you. I like Toby Harrah. Not enough to vote for him, but I like him. But there was a guy here a while back who tried to put down Ezra Sutton by accusing him of being the Toby Harrah of his era. There were other flaws in his argument (he didn't account for the season-length changes back then), but the point was that he was using Toby Harrah to invite ridicule. Who else is getting short shrift? I'd guess Lombardi & Lazzeri. I'm not sure we need more 1930s guys, but both of these guys lower on people's radar screens then their hitting levels would warrant.

Doyle's a big one though. I had him above Childs. He hit better and for longer, but fielding concerns put him 25-30 places below Childs in the balloting.

Being a mediocre fielder also didn't seem to hurt Rogers Hornsby. :)

I know that Doyle has taken a number of knocks for playing the position he did in the league and time he did. I like middle IF defense as much as the next voter--I thought Fox was a worthy candidate, I suspect I'll vote for Randolph. I voted for Bancroft and Rizzuto. I might even vote for Davie C. someday, and I'll give them all the credit in the world for saving more runs in the field than Doyle or the mediocre fielding middle IF of your choice. But, at the same time, guys like Doyle and Lazzeri and Jeff Kent (who is a pretty good comp for Doyle in everything except personality) ought to get credit for being significantly better at generating runs at the plate than the guys mentioned above. Whether you'd rather have Larry Doyle or Nellie Fox, Vern Stephens or Phil Rizzuto, Jeff Kent or Willie Randolph on your team seems more a matter of personal preference than anything else.

David Foss, OK, but 1B, 2B, and OF are *all* pretty close then--corner OF is stronger, but only by 0.2 wins per season. They're virtually identical. Point being that in Doyle's day, 2B was a hitting position, even more so than 3B is today.

I don't know why you keep saying that it's *only* Doyle's poor fielding that holds him back. It's the fielding *plus* the fact that he was playing an offense-first position. If 2B = 1B then, would we elect a stone-gloved 1B with his hitting numbers? No. Thus, we don't elect Doyle either.

Andrew M, I think Rizzuto's fielding advantage was simply worth far more runs and wins than Stephens' hitting advantage.

Eddie Collins and Napoleon Lajoie Posted: April 18, 2007 at 01:52 AM (#2337904)

2B around 1910 were absolutely a dime a dozen.

Hi

I think I misused the phrase. What I meant to say is that the league was overflowing with good-hitting 2B, it wasn't hard to find one at all. Collins and Nap's hitting should be assessed as if they were 1B or OF, not 2B as the defense-first position it evolved into. That said, Collins and Nap were such good hitters and fielders that they'd be inner circle even as 1B or OF.

my home computer has "issues".

can anyone (KOJK?) with Sinin's bb encyc look up Randolph's (& D Concepcion's) RCAP for me? Thanks.

I still don't quite see why a preponderance of the best hitters at a certain position means that they weren't individually the best hitters and therefore among the best players.

Let's say that all the best-hitting shortstops of the 1990s and early 2000s had all for some strange reason been moved to second base instead. The best-hitting shortstop of the era becomes Royce Clayton, while second basemen routinely lead the league in all kinds of individual and aggregate hitting categories. Does that make Royce Clayton a HOMer?

crucial Q, Bob D Cri -
and the answer depends on how fungible (overlap, easily movable from one to another) these middle infielders are.
we all agree that LF-RF is probably 90% fungible. CF to corner OF, less but still some there, LF to 1B a bit, 2B-SS some, 2B/SS to 3B less, C to anywhere virtually 0%.

A player who is best in MLB at his position is very valuable, since if you're tyring to win a pennant, he is the one you want to trade for. UNLESS there are 4 guys who are playing some other position that could easily be moved over if they had to, to fill in a gap.

Per the idea that 2B are a dime a dozen in 1910, here's just a quick eyeball test. I once upon a time picked all-star teams, 3 deep. Here are the best NON-all-star 2B and LF of 1910.

AL 2B--Larry Gardner 2-36-.283 in 113 games
AL LF--Patsy Dougherty 1-43-.248

NL 2B--John Hummel 5-74-.244
NL LF--Your choice Fred Clarke or Zack Wheat, probably Wheat who played 150+ games versus Clarke's 120+ at age 37

Let's try 1911.

AL 2B--Amby McConnell was the only non-all-star 2B to play 100 games
AL LF--Nixey Callahan 3-60-.281

NL 2B--Dots Miller 6-78-.268 and Hummel is still around
NL LF--Clarke and Wheat still in the mix though Josh Devore may have been better

This of course doesn't prove anything, just anecdotal, but it seems that journeyman quality LF were at least a bit easier to come by for these couple of years.

Bob Dernier Cri, well, my evaluations of positional value are based on the freely available talent level, not the position average. So moving A-Rod etc. to 2B wouldn't change Clayton's value in my system at all.

TomH, any SS could play 2B, but even elite fielding 2B would be bad SS. Most 2B are failed SS, after all. I know, Nellie Fox already got elected, I know.

sunnyday2, my data are based on a 54-player sample (worst three starters at 2B in each league for 9 years), so I'd like to think I've gone a bit beyond just picking eight middle of the pack players and eyeballing them. That said, I agree that LF were *slightly* deeper than 2B for those years (0.2 wins per full season). 1B were not. If you're interested I'm happy to send you the list of players and their stats.

As a lurker to all the HOM threads, I was very excited to see my absolute favorite fielder from my pee-wee and little-league baseball days come up for consideration. I absolutely wanted to be Willie Randolph, once it was revealed to me I was too damned short to be Tom Seaver.

Ummmm.... how much longer until you guys actually start talking about Willie?

Collins and Nap's hitting should be assessed as if they were 1B or OF, not 2B as the defense-first position it evolved into. That said, Collins and Nap were such good hitters and fielders that they'd be inner circle even as 1B or OF.

I'll disagree with this. Shoo-in HOM-ers, yes, inner circle? How much better would Collins be than Sam Crawford if not for the positional adjustment? Better Just because there's a couple of stars at a harder position and a shortage of stars at an easier position doesn't necessarily make them equivalent. Does anyone think McInnis & Collins had equal defensive skill? Does anyone think 2B couldn't shift to 1B as they got older if they hit well enough? Does anyone think a 1B could switch to 2B? These "spectrum inversions" happen from time to time but I don't see them as anything more than historical curiosities. When they needed a place in the field to try and hide Jack Fournier, those chose 1B.

I'm on board with 2B and 3B swapping places in the spectrum, that's well documented. But I'm not going to be grouping them in with OF & 1B.

DavidFoss, I have Collins with 98.5 standard deviation-adjusted wins above overall league average (which is (BRAA + FRAA)/9 for BP users), and Crawford with 59.5. The positional adjustment does extend Collins' lead, particularly since he played well into the 1920s when 2B got progressively more difficult, but if Collins were an OF he would have taken Crawford or anyone of his ilk to school.

Remember that the number of stars at a position has zero impact on my positional ratings. I look at the bottom of the distribution to determine positional value, not the top. The question you have to ask is, if playing 2B was much harder than playing 1B back then, why were there so many guys like Dick Egan (not counting his All-Star 1909)? He was among the worst regulars at his position in the majors, but still hit close to the league average *and* played excellent defense. At shortstop, that production would be the best at the position in MLB; at 2B it was among the worst.

TomH

Your rCAPs, sir.

Randloph 235
Concepcion 126

Well, actually, when somebody says that 2B grew on trees in 1910, it really only takes an anecdote to prove that it ain't so. There were some pretty bad actors out there.

any SS could play 2B, but even elite fielding 2B would be bad SS. Most 2B are failed SS
this is a gross overgeneralization. Most of us, given a hour of spare time, could come up with elite 2Bmen who could also play short. (Pokey Reese is the first one that comes to my mind). And, the particular skill at turning the duece would eliminate some SS from playing 2B (since not all have been tried there, it's kind of hard to assert "any shortstop could..."). When SS who start losing range, are they moved to 2B? Wouldn't that make most sense,, if 2Bmen were fialed SS? But we often see SS move to 3B or OF instead. There's probably a god large study there for someone to make....

I suppose replacement level for 2B also is affected by GM/managers decisions not to let a great hitter play there for fear of injury.

Is the tendency also for 2Bmen to miss more time with injuries? If so, this would put a premium on 2B playing time, or at least would again affect 'replacement level' in that we'd have to make sure we're not merely measuring replacement as worst three starters, if indeed each team's backups get more time at 2B. But it's only a postulation I once more make without the data to back it up - man, some of us laggards oughta get cracking the data instead of posting what-ifs.....

TomH
One item we ought to remember about Randolph, and guys like him, is that leadoff hitters get many more PA per game, but each PA is worth slightly less in run-scoring opportunity (fewer men on).

If you rate players per game played, this is moot.

If you rate players by WS/WARP/RCAP (actual PA), a player who batted leadoff his entire career should have his offense value lowered by about 5%.

Do you mean that users should deflate his offense value measured by those statistics?


quoting David Foss twice:
I'm still one of Doyle's biggest fans. What's clear to me, though, is that the electorate is particularly dismissive of glove-men who have mediocre gloves. You really need a WS letter grade of B or above to get any traction as a 2B or SS candidate here. It doesn't help that WARP's wacko fielding replacement really hammers down these guys as well.

WARP and Win Shares frequently disagree sharply about fielding value, so we consider them two independent judgments ;-)

Who else is getting short shrift? I'd guess Lombardi & Lazzeri. I'm not sure we need more 1930s guys, but both of these guys lower on people's radar screens then their hitting levels would warrant.

Er, they should be lower than their batting would warrant.


TomH
it's only a postulation I once more make without the data to back it up - man, some of us laggards oughta get cracking the data instead of posting what-ifs.....

need stable computing platform. Don't fix rotten planks; use them as excuses.

When SS who start losing range, are they moved to 2B? Wouldn't that make most sense,, if 2Bmen were fialed SS? But we often see SS move to 3B or OF instead. There's probably a god large study there for someone to make....

SS are usually not moved to 2B because they lack "range" in the sense of moving left or right, it's usually because they don't have the arm to make an effective throw to 1B after moving to their right.

Many players are drafted at SS- if they have the range but not the arm they move to 2b, the arm but not the range to 3B, if they are not cut out for the infield to the OF. The vast majority of these moves are made at the minor league level not in the majors.

veteran SS are moved to 3b or the OF if they can hit enough to play those positions but can no longer field SS effectively.

If you rate players by WS/WARP/RCAP (actual PA), a player who batted leadoff his entire career should have his offense value lowered by about 5%.

Do you mean that users should deflate his offense value measured by those statistics?

yes

example:
from one previous stretch of MLB data, leadoff batters had .505 runners on base on avg when they came to bat.

cleanup hitters averaged .74; it is the most prodcutive spot per PA.

At least one previous study published by SABR showed that the relative importance of an atbat varies almost exactly proportionately with the number of runners on plus the batter (bases juiced is 4 times as crucial as no one on).

So, typical cleanup hitter's PAs are (batter=1.00 + .74 runners) 1.74/1.505 = 16% more important than #1 hitters.

Of course, leadoff men come to the plate about 8% more often.

WS, WARP, RCAP, take into account more often, but not the 'less runners on' part.

There probably are other nuances, such as the increased importance of getting on when the guy following you can slug which this does not measure.

Someone with a lineup simulator (gee, me again if I got my home computer fixed..) should be able to estimate the amount of this effect.

I have to say that I disagree with you there Tom. I am not one to parse the different values of AB's, that begins to track too closely to WPA, which I am not a fan of at all. To me a single is a single, a double a double, no matter how many people are on base.

This doesn't mean of course that the only value given to a hit should be the bases reached. Obviously a single is more important than a walk because it is more likely to advance and/or score a baserunner, etc.. But I don't like to give credit to players for having the privilege of hitting with more runners on base. The only people I like to leverage are those that are leverage by their managers relief pitchers, and I guess pinch hitters (though none of the latter are HOM candidates).

Seriously, isn't seperating leadoff hitters and cleanup hitters because of this the same as seperating hitter from good and bad teams or hitters from good and bad run environments?

I guess what I am syaing is this:

1. leadoff hitters get more PAs.
2. leadoff hitters' PAs have less leverage.

you can account for both, or neither, but don't give one without the other.

WS/WARP/RCAP, I believe, are guilty of giving credit for (1) and not penalizing for (2).

TomH, Pokey *played* plenty of SS, you can't characterize him as just a 2B. Of course, this is a bit tautological--since SS is the toughest position on the diamond, if you are capable of playing SS, you almost always do! The only exception I can think of is the A-Rod/Jeter imbroglio.

Replacement level is a rate--playing time has no part in it. If your guy goes down, you plug in a replacement. If your replacement goes down, you plug in another replacement. It's the freely available level, and while not *infinite*, given the shape of the talent distribution there are presumably at least 10-20 freely available guys capable of playing around rep level at any given time at a position.

I have no idea if you're right about 2B being hurt more often. But it's certainly true that catchers have less PT, and that doesn't decrease the replacement level at all, at least in my system--remember that, at heart, my replacement levels are Nate Silver's Freely Available Talent levels, which is the average production of players over age 27 making less than twice the league minimum salary. It just means that if you don't subjectively boost catchers' value over replacement--by multiplying (increasing their playing time) rather than by adding (lowering their replacement level)--you will wind up with a HoM nearly entirely devoid of catchers.

> The only exception I can think of is the A-Rod/Jeter imbroglio

Pesky/Stephens

Good one. Point being, the only reason a team wouldn't put a SS-capable player at SS is if they already have a SS on their team. Interestingly, Pesky is going to shoot up my 1998 ballot even though he spent half his time at 3B. Pesky seems to have been a much better fielder than Vern; why didn't they slide over Stephens?

yeah, Pesky/Stephens, I was about to say
Wright/Barnes
Williamson/Burns
teammates, teammates, teammates, supporting DanR's general point

Au contraire,
One view of Earl Weaver and Cal Ripken (is it only Weaver's own view?) is that numerous big men were capable of playing shortstop, and didn't because everyone followed a social norm. Compare African-American men playing quarterback until about the same time.

One view of Honus Wagner through 1902 is similar. This may be Bill James view: Wagner was a superior shortstop bud he didn't look like a shortstop, so years passed before he became a regular there.


Lassus, try the search feature to find some discussion of Willie Randolph.

Mark Shirk, whose surname is hard to believe,
I agree with you on the big point that it feels wrong to discount linear weights and other measures of run production by batting position, with a heavy discount for batting first because those batters' plate appearances have especially low relative value or leverage.

But batting position is the result of a management decision analogous to that which uses some relief pitchers in high leverage and others in low leverage situations. and at the same time analogous to that which uses starting pitchers for 200 innings, some relief pitchers for 100 innings, and some for 50 innings. Management knows that the fourth batter will appear with more runners on base, and will appear fewer times, than will the first batter.

Pesky seems to have been a much better fielder than Vern; why didn't they slide over Stephens?

Because Joe McCarthy was getting pretty senile has always been my theory.

I think the idea that replacement level talent is freely available and almost infinite (ok, 20) is overstated because teams can't always make the deal to get the guy in. The Twins lost Cirillo and now Punto and Luis Rodriguez remains. Yes, that Luis Rodriguez. Hit about .230 last year with 1 HR. So yesterday Ron Gardenhire says that, well, he's not ideal as a replacement at 3B because he doesn't have the range. He hits great, Gardy said, but he's not much with the glove. (.230 with 1 HR and he hits great; how bad must his fielding be?) There's the 3B of the future in AAA but he's not ready and they'd rather not have him struggle. So they're talking about bringing up 1 of the 2 guys who don't even start at Rochester, guys who fill in at 1B and 3B and DH. Replacement level? By definition, but they're not as good (overall) as Luis Rodriguez who, until Cirillo and Punto got hurt, pretty much looked like the definition of replacement himself. Now, there's probably 20 guys out there who are better than any of them, but the Twins don't own 'em and they're not gonna lay out a contract for a 10-20 day fill in. So replacement can be a bit lower down the food chain than we sometimes think.

Are MLB player contracts always for a year? The NBA has 10-day contracts...

Tom and Paul,

I have no problem with the scenario (adjusting for # of AB's without adjusting for leverage), though I could also be convinced of adjusting for neither. I also always thought that WS did the latter. While it splits up WS by games, doens't it divide within the team by RC which would be sensitive to PA's? How does WARP adjust for the extra PA effect?

Also, I am not necessarily sold on the leverage argument that Paul makes. If I read Tom correct, and maybe I didn't, leadoff hitters have th least amount of leverage because ideally 7/8/9 are the worst hitters. However, a manager does not choose his worst player to hit in thsi position for obvious reasons. In fact, he can't because the guy batting after the worst hitters will always have the lowest leverage no matter the batting position no? So there are only a few position where a player can be properly leveraged.

Also, if you are getting leverage do leadoff hitters get any bonus points for the role in getting on base in order to leverage the #3 or 4 hitter? Shouldn't some of that leverage go to the guy who actually creates the situation by getting on base?

Dan,

I am pretty sure that MLB contracts are for one year, but where do you find players mid-season for a short contract? Most of the plaeyrs are already signed and playing at a AAA club.

Paul,

I will have you know that I have earned that surname through 25 years of laziness and procrastination!

Why don't teams just stock their AAA rosters with freely available replacement players in the offseason?

not that this is where your going but Eddie Gadell and the Tigers replacment had 1 dat contracts

Pesky seems to have been a much better fielder than Vern; why didn't they slide over Stephens?

There were starting to toy with the idea of shifting Pesky to 3B already in 1947 -- swapping him with Pellagrini for a couple of dozen games. Pesky's fielding numbers look to be down that year. Baseballlibrary.com says that Pesky got married in the offseason and gained 30 lbs. I'm guessing that's one of the reasons they went out and got Stephens in the first place... they had already decided to move Pesky.

According to BP, Stephens' fielding numbers in 1946 and especially 1947 are fairly solid. It could be that Stephens was the better option in the spring of 1948.

This issue is famous. Both David Halberstram and Bill James have weighed in on it.

I'll talk about Willie. Here's the problem. Win Shares.

Yes he has 312, more than any other eligible 2B. And he had that year with 31. But here's the problem.

31-23-23.

I mean, even 31 is not that much. Doyle, Myer and Bobby Avila had better years and, hey, Steve Sax had a 31, and Lazzeri and Stanky had 30s. Still, ok, 31 is a nice year.

But 23 as your 2nd and 3rd best? And 54 for your top 2? Here are some of the guys who did better on all 3 of those numbers. Doyle, McAuliffe, Myer, Evers, Gilliam, Avila, Sax.

If Willie Randolph, why not Bobby Avila.

Randolph 31-23-23/114
Avila 34-24-24/124

Randolph .276/.373/.351/105
Avila .281/.359/.388/104

Randolph B+
Avila A-

Yeah, I know that Avila only had about 5200 ML PAs but he probably played as much ball as Randolph did over the course of his whole career. I understand you'd want to see the MLEs, but my gut is that they are pretty comparable for career while Avila was clearly better at his peak.

If Randolph, why not Evers?

Bobby Avila MLEs. BTW the Links to Latino Stars is not working.

YEAR LG AGE PO AVG OBP SLG G PA AB H TB BB ops+ sfws
----------------------------------------------------------------------
1944 NL 20 2b .276 .353 .348 154 612 547 151 190 65 98 18.5
1945 NL 21 2b .263 .337 .312 114 450 404 106 126 46 81 11.1
1946 NL 22 2b .311 .392 .408 146 589 520 162 212 70 127 24.5
1947 NL 23 2b .302 .382 .403 122 491 435 131 175 56 109 19.6
1948 NL 24 2b .205 .270 .222 59 227 208 43 46 18 34 2.3
1949 AL 25 2b .300 .390 .404 140 563 497 151 201 66 109 22.6
1950 AL 26 2b .299 .387 .383 80 230 201 60 77 29 98 7.0
1951 AL 27 2b .304 .374 .410 141 602 542 165 222 60 112 24.0
1952 AL 28 2b .300 .370 .415 150 664 597 179 248 67 118 24.0
1953 AL 29 2b .286 .353 .379 141 617 559 160 212 58 98 22.0
1954 AL 30 2b .341 .404 .477 143 614 555 189 265 59 142 34.0
1955 AL 31 2b .272 .368 .400 141 619 537 146 215 82 107 20.0
1956 AL 32 2b .224 .317 .318 138 583 513 115 163 70 69 14.0
1957 AL 33 2b .268 .334 .289 129 509 463 124 134 46 73 13.0
1958 AL 34 2b .253 .349 .365 113 430 375 95 137 55 98 12.0
1959 2lg 35 2b .227 .314 .322 93 298 264 60 85 34 74 5.0
======================================================================
.282 .360 .375 2004 8097 7217 2037 2709 881 101 273.6

If you use 162adjWS, his career total is 287.

Avila 287/34-24-24-24-23-22-20-20-18-14-13-12-11
Randolph 312/31-23-23-22-22-21-20-20-20-19-18-17-15-13-11

Avila is as good or better every year through #8, then Randolph gets back the marginal WS in years #9-10-11. Then their last 2 double-digit WS seasons are about equivalent. That leaves Randolph's career advantage as 2 seasons of 17-15, his 12th and 13th best seasons. Those are the only seasons that his career advantage relies upon. Granted, Avila was only "as good" the rest of the way, unless you're a peak voter, in which case his top 6 years are consistently if slightly better than Willie's.

Avila 101*/142-27-18-12-9-9-7 (?100 in 7 of 14 full years)
Randolph 105/133-27-22-12-8-7-5-4-3-2-2-0-0 (? 100 in 13 of 14 years)

*104 in actual ML action

Pretty comp. Avila is better by an average of 1.5 points through year #7, after which Avila had no more above average years. But during Avila's below average years (in order from high to low, not chronologically) Randolph was in the 100-105 range. Avila clearly had some inferior seasons during that time but also had 5 98s. So Randolph's advantage here is pretty slight.

So offensively Randolph has a slight edge, defensively Avila has an edge. During 7 full-time ML seasons Avila averaged about 137.5 games per season, not great, but Randolph was at about 133.5. For their top 5 seasons in games played Avila was at 150-143-141-141-141, Randolph was at 153-147-144-143-142. Very slight advantage for Randolph.

In summary, Avila seems clearly to have been the better player at his peak. Randolph has the better career even looking at Avila's MLEs. But it also seems likely that adjusting to league and culture changes couldn't have helped Avila much. I can see the argument for Randolph to be ahead and I can see a peak voter preferring Avila.

Bottom line: If we are really committed to MLEs, Avila is a 287 WS 2B, the 2nd best total eligible today, and with a better peak than the top career guy. Where does he fit into y'alls ballots?

Baseballlibrary.com says that Pesky got married in the offseason and gained 30 lbs. I'm guessing that's one of the reasons they went out and got Stephens in the first place... they had already decided to move Pesky.

I don't buy that at all. Whatever Pesky was doing had nothing to do with getting Stephens. Stephens was a stud star, an obvious perfect Fenway fit. The only question was exactly how big of a package did the Red Sox need to present to the Browns. The Red Sox's acquisition of Stephens had nothing to do with any sort of problem with Pesky; whether Pesky was doing great or terrible had no bearing on the deal.

BTW the Links to Latino Stars is not working.

It was closed, for some reason. Fortunately, I have the key. :-)

Evers: shorter career, weaker league (NL<AL in 1910)

Couple o' lineup sims I did:

created a typical lineup that scored 706.4 runs per 162 game season.

added to the leadoff hitter 10 singles, 10 walks, 5 doubles, 5 HR, 30 fewer outs per 665 PA (+45 OBA, +73 SLG). Then did the same with the cleanup and 9th place hitters.

new lineup with better hitters scored extra

1st batter: 32.8 runs per year. Ran 10000 seasons (1.62 million games)
4th hitter: 33.3 runs/yr
last batter: 28.0 runs/yr

I estimate the std dev of runs/yr at 0.3 for 10K trials.

Leadoff man got a few extra PAs, but it washed out with the 4th hitters' leverage.

Obviously, the 9th hitter comes up less often, and you get less benefit.

-
Somewhat related question:
Pitchers bat 9th, AND they likely hit with fewer men on; because in some crucial (men on, close game) situations, they are pinch-hit for. I wonder if when I adjust for pitchers' hitting, I'm overdoing it somewhat?

in terms of career shape, Randolph to me looks most like.. Jake Beckley!

long career peakelss wonders, as seen by WARP1:

age 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
W.R 00 75 78 70 74 87 40 69 59 81 66 55 73 44 60 39 75 24 00 00
J.B. 44 72 77 71 65 81 73 66 43 42 41 80 77 74 57 65 74 53 16 -6

5-yr moving window:
age 22 23 24 25 26 27 .28 .29 .30 31 32 33 34 35 36 37
W.R 59 77 70 68 66 67 .67 .70 .71 68 64 54 58 48 40 28
J.B. 66 73 73 71 66 61 .53 .54 .57 63 66 71 69 65 53 40
diff +7 -4 +3 +3 00 -6 -14 -16 -14 -5 +2 17 11 17 13 12
scary similar, ain't it?

all of the standard WARP1 caveats apply, of course....

Sunny said:

I'll talk about Willie. Here's the problem. Win Shares.

Yes he has 312, more than any other eligible 2B. And he had that year with 31. But here's the problem.

31-23-23.

Wow that is a pretty good arguement, especially since Jake Beckley might be his biggest competitor - indirectly.

And here is another one. He was primarily a leadoff and #2 man - which if using the anti-Brock arguement means he had lots of opportunities to rack up those win shares.
From Retrosheet below:

I cam of age as a Yankee fan in the Randolph era, but I'm firmly in the anti-Randolph camp. He's somewhere around 40 or 50 for me. Here are a few comparisons:

(1) Randolph v. Fox: I agree that they are very similar, but I had Fox 41st last year. I just don't think these guys hit well enough. Further, I think Fox is better because (1) he has a few genuine peaky seasons and (2) he was the best 2B of his era.

(2) Randolph v. Avila/Evers/Lazzeri and a few others: I think Randolph is very comparable to a bunch of guys who we rejected long ago. For the most part, his career totals are higher than these guys but (1) some of that is the longer seasons and (2) I don't see how the difference between first-ballot HoMer and no votes for 2 guys of equal offensive and defensive abilities can possibly be one or two 15 WS seasons. Basically, the if Fox, then Randolph argument is exactly the kind of argument that (by ignoring the other similar excluded players) drove the HoF down its long spiral to mediocrity.

(3) Randolph v. Beckley: Not a bad comp, but (1) Beckley put up substantially more career value when you adjust for season length and (2) Beckley was the only successful long career 1B of his era while Rnadolph had a set of peers.

(4) Randolph v. Nettles: Teammates. Nettles ranks higher all time at his position, was a better hitter, and was an historically great fielder at his only slightly easier position. I have Nettles in the 20s and can't see putting Randolph higher.

(5) Randolph (and Fox) v. Bancroft (and Rizzuto): Let's just say I agree with the camp that has the .265-.275ish good-fielding SS's ahead of the offensively similar 2B.

(6) Randolph v. Doyle and Monroe: The cases for these 3 are so different that your milage may vary, but my methods have these two as the best available 2B by a decent margin. They are in the low 20s and edging closer to the ballot.

When I say .265-.275ish good fielding SS's above, I meant .265-.275ish EQA.

How much of the argument for Randolph centers on his low standard deviation era and the low replacement level of his ear?

I currently have him just above Fox at #27

Carter and Blyleven are #'s 1 and 2 while Stieb is at #8 and could be higher (Walters is third among my backloggers)

How much of a competitor to Randolph will Stieb be?

I looked over Avila and I don't think he measures up. I have him down near the bottom of the spreadsheet (always good to add to it) above Schoendienst and around where I had Nellie Fox. WARP is much kinder to Randolph's fielding than it is to Avila's fielding. Randolph also has a couple of extra years career advantage, not to mention the DH boost.

I agree with the Nettles and Beckley comparisons, they're all in the middle of my ballot.

(4) Randolph v. Nettles: Teammates. Nettles ranks higher all time at his position, was a better hitter, and was an historically great fielder at his only slightly easier position. I have Nettles in the 20s and can't see putting Randolph higher.

I agree that these are similar players, but not that Nettles was a better hitter.

By WARP1, Nettles has a .273 career EQA, Randolph .275.
By WARP2, Nettles has a .275 career EQA, Randolph .284. WARP2 is relevant here because it adjusts for the DH.

Nettles is pulled down a little bit by some bad late seasons, but if you peel off his last 700 PA in his last three seasons, he's going to come out similar to Randolph in both PA and WARP1 EQA, I think.

If they are equal offensively, the question comes down to whether a very good to excellent defensive second baseman is more valuable than an excellent to historically great defensive third baseman. WARP and WS have opinions about this: WARP gives the edge to Randolph: 549 FRAR to 435 FRAR. I don't have WS at work, but that's a comparison that also should be made.

While on the subject of Randolph's offense, it might be worth comparing Randolph to the other well-known leadoff hitter we are considering: Lou Brock.

By WARP1, Brock has an offensive edge on Randolph, .282 to .275, but by WARP2, this is reversed, Randolph leading .284 to .281. I think if Lou Brock had been a very good defensive second baseman instead of an average to below average defensive left fielder, we would have elected him easily. Brock obviously has the durability chit that Randolph doesn't, so Randolph is not simply Lou Brock + second base defense, but I think it's worth noting.

As Randolph's lack of durability may be partly due to his being a second baseman rather than a left fielder, it would be good to learn more about Randolph's history: why did he miss so many games? It would also be good to have some comparative statistics on games played per year, IF vs. OF.

Randolph's seasonal win share totals obviously suffer partly because of the durability issue, as do those of others of the infielders in the high or middle backlog. Are we properly assessing them on durability? (It's here, of course, that the issue of replacement level becomes so important -- the lower the replacment level, the more durability is rewarded.)

Randolph B+ 97 dWS
Nettles A- 93 dWS

Not much to choose.

Randolph 97 dWS in 2153 fielding games
Nettles 93 dWS in 2504 fielding games

much to choose!

second base has a higher intrinsic weight than third. in other words, there are more win shares available for second basemen than thirdbasemen - i think. i'm at work and don't keep the win shares book with me.

It is possible, maybe even probable, that an A-, 93 WS 3B provides more of an edge over the competition (not to mention over replacement) than a B+, 97 WS 2B.

Well, since win shares defines "replacement level" as zero, Randolph has, by definition, more value above replacement as win shares sees it.

If you want to find "fielding win shares above average" for Randolph and Nettles and you have the win shares book, find the ws/1000 def innings value that is midway between B and C in the letter grade listings for the two positions, and multiply them by the defensive innings listed for the two players in the defensive innings tables, and that will give you the value of an average fielder in those innings.

I expect that Nettles will have more fws above average than Randolph, as he has more FRAA, but you never know. I'd do it myself, but I won't have a chance to look at the book until Friday evening . . .

I use Fielding Win Shares as part of my defensive analysis. The methodology is:

1. Calculate how many FWS an average fielder at the player's position would have accumulated in his playing time, and subtract. This is FWS above average.
2. Multiply by the league's runs per win (a function of its R/G) and divide by 3. This is WS-FRAA.
3. Multiply this figure by the ratio of the standard deviation of Chris Dial's ZR data to the standard deviation of WS-FRAA for the given position. This expands or contracts the scale of WS-FRAA so it is in line with play-by-play estimates.

By this measure, I have Randolph with 85 career WS-FRAA (outstanding) and Nettles with 150 (among the greatest of all time, particularly since he was below average in the latter part of his career).

Chris Cobb:
By WARP1, Brock has an offensive edge on Randolph, .282 to .275, but by WARP2, this is reversed, Randolph leading .284 to .281. I think if Lou Brock had been a very good defensive second baseman instead of an average to below average defensive left fielder, we would have elected him easily.

Certainly. I suspect that Lou Brock would have been elected if he were Willie Wilson on defense --merely a good CF.


mulder is Kelly?
second base has a higher intrinsic weight than third. in other words, there are more win shares available for second basemen than thirdbasemen

This is not something to correct for unless you disagree radically with Win Shares. That is, the implied correction would be a fundamental revision whereas most corrections people make or imagine are relatively superficial.
(And I am with Win Shares on this. One shouldn't walk down the road toward thinking of "available Win Shares" as a denominator except in the case jimd(?) has noted: the ceiling for fielding value on the team level, at 33% of team defense.)


Mark Shirk
Paul,
I will have you know that I have earned that surname through 25 years of laziness and procrastination!

I had to check your response, having taken so many cracks about my surname over the years, despite the misspelling. Your response is very good to historically great!

My point about the batting order as a management decision encompassing frequency (how many times will he bat) and leverage (with what importance will he bat) should be understood at the highest level --encompassing the frequency and leverage broadly defined and jointly.
I agree with you that the first batter has more leverage than the fourth insofar as higher quality batters follow the first one than the fourth one. But the fourth batter has more leverage, net, I presume. (TomH may mean leverage more narrowly.)
By the way, while its true that the 7-8-9 batters are relatively weak, a major factor that decreases leverage for the first batter is the first inning. The first batter's first time at bat is the only one in the game that is guaranteed to have value only in the reaching base aspect, not at all in the advancing runners aspect. And those appearances constitute almost 25% of the season for an everyday leadoff man.

DanR & yest
52. David Concepcion de la Desviacion Estandar (Dan R) Posted: April 18, 2007 at 11:08 PM (#2338929)
Why don't teams just stock their AAA rosters with freely available replacement players in the offseason?
53. yest Posted: April 18, 2007 at 11:18 PM (#2338939)
not that this is where your going but Eddie Gadell and the Tigers replacment had 1 dat contracts

The role of player contract has changed a lot in 100 years not to mention 150. Reading the May 1912 coverage in the Detroit News-Tribune (a while ago) it seems to me that Detroit already needed approval of the League Pres or National Commission to employ the replacement players. I don't recall whether they had written contracts and I'm not sure yest means that; written or spoken, they did have approved agreements to play one day for $50.

My reading of the Burleigh Grimes transaction in spring 1934 (trying to date it precisely a while ago) is that Grimes and manager met on the road and made verbal agreement about his employment. He did not sign until the team returned home to St Louis? er, maybe it was return to New York after a road trip to St Louis.

During Willie Randolph's time the control of player employment and compensation is more advanced, formalized, rule-governed. I am not sure this difference in conditions means that the concept of freely available talent must be used differently for Randolph and, say Buddy Myer and Larry Doyle, if one's focus is on those players (2Bmen) for the Hall of Merit! By the way, what is the timespan covered by FAT in the hands of its author?

Paul, I agree. I was pointing out why Randolph would have more DWS in fewer games than Nettles despite a lower grade. Randolph can earn more win shares because second basemen make more plays. That's why the steps Dan R. takes are so important.

Nate's FAT study was from 1985-2005. 1985 is the first year bb-ref has salary data.

Yes, mulder is a pseudonym for Kelly from South Dakota ;-) (Here in the midwest SD = South Dakota.)

mulder is Kelly?

You have learned the truth, Paul.

mulder is Kelly?

You have learned the truth, Paul.

Are you sure?

Maybe Scully is Kelly?

Maybe Scully is Kelly?

You have learned the rest of the truth, Jim.

Unless Scully isn't Dana but rather Vin!

Now you're just being ridiculous, Eric. ;-)

TomH wrote:

Evers: shorter career, weaker league (NL<AL in 1910)

Couple o' lineup sims I did:

created a typical lineup that scored 706.4 runs per 162 game season.

added to the leadoff hitter 10 singles, 10 walks, 5 doubles, 5 HR, 30 fewer outs per 665 PA (+45 OBA, +73 SLG). Then did the same with the cleanup and 9th place hitters.

new lineup with better hitters scored extra

1st batter: 32.8 runs per year. Ran 10000 seasons (1.62 million games)
4th hitter: 33.3 runs/yr
last batter: 28.0 runs/yr

I estimate the std dev of runs/yr at 0.3 for 10K trials.

Leadoff man got a few extra PAs, but it washed out with the 4th hitters' leverage.

In a sense this result is not surprising -- if 4th hitters PAs weren't leveraged relative to leadoff hitters, it would be optimal to lead off with your best hitter. Since no one seriously considers leading off with a Ruth, Williams, or Pujols, we obviously believe that the extra PAs are generally offset by lack of leverage.

Where I'm hesitant about applying your result, however, is that the best leadoff hitters aren't just better hitters, they're players whose skills are optimized for the leadoff role. Some events, e.g., walks and stolen bases, are worth more leading off an inning and other events, e.g., extra base hits, are worth more with runners on base. Suppose you ran your simulations adding 30 walks, 30 SB, and 3 CS. My guess is that adding those skills may be worth more for the leadoff hitter than for the 4th hitter. And if my hypothesis is correct? -- well, wouldn't that suggest that leverage isn't just a constant, but a variable that depends on the batters particular mix of skills?

1. Calculate how many FWS an average fielder at the player's position would have accumulated in his playing time,

Over what timespan is the average FWS measured? Eg 5- or 11- or 21-year moving average.

Randolph and peak: another angle

When I look at peak, I look at it in two ways. One is total value above average, for which I subtract the value of an average player (in the WARP1 and WS systems) from the player’s value for all the seasons he is above average, then sum the differences remaining. This measure is significantly influenced by seasonal durability.

The other is peak rate. For this, I find the player’s highest rate of performance per 162 games during a period of at least five consecutive seasons. I use five seasons generally, but if the player, due to a single off year in the middle of a longer run of good years, has a better rate over six, seven, or eight consecutive seasons than over five, I use the longer period. I do this for WARP1 and WS also, adjusting for the DH by adding .4 wins in WARP1 or WS to the player’s total per 162 games played. This is a way of looking at quality of play, irrespective of playing time (although I generally do not include seasons before a player became a full time player in the time frames I examine.)

Anyway, I thought others might find it interesting to see how Randolph’s peak rate looks in relation to the “glove” candidates who are his contemporaries, so I compiled peak rate lists in WARP1 and WS, including all of the serious C-2B-3B-SS candidates whose peaks and careers fell mainly in the 1970s and 1980s.

I have put two dividing lines into the lists. One falls at 10 wins/162 games line, which I think nearly marks the line above which the candidates are nearly all obvious HoMers. The other falls below Darrell Evans, who has the lowest rate of any infielder from these decades that we have yet elected.

For ease of comparison, I have denominated the win shares list in wins. I have provided dates for the peaks in the WARP1 table. The WS peaks are usually but not always in the same set of seasons. I did not have the WS peak seasons data handy, so I omitted it. I have included with the WS list the variance of the WS peak rate from the WARP1 rate for each player.

A similar list could be compiled for Dan R.’s WAR, but I haven’t done so yet.


A few notes. Randolph falls pretty squarely in the middle of what each system defines as basically the borderline group, .75-.80 wins/162 games below the level of the players who will be obvious HoMers.

His peak rate is almost identical in WARP and WS, which suggests that he is a balanced candidate (bat heavy candidates do better in WS, fielding heavy candidates do better in WARP1).

This table doesn’t unearth conclusive evidence that Randolph is a hidden GIANT of his era, but it does suggest that, during his peak, he was a high-impact player, highly comparable in quality to some players whom we have elected.

attempt at summing up discussion so far:

Willie vs other 2B candidates: tough, since top 2 reutrnees are Doyle and Monroe. Doyle much better bat, but not as good a glove and 2B was a LOT different in 1910. Monroe's case involves reasonable speculation. Compared to Nellie Fox, better stick, almost as good a glove, longer prime, not as durable.

Willie vs other modern infielders: best comps are Nettles and Concepcion - all good gloves, decent bats. Randolph leads in 'batting vs position' if you use RCAP, but Concepcion may be clsoe if you use replacement level, particularly the poorer shortstops of his time. Nettles likely th ebest glove.

--

Randolph's case seems to be that for 16 years, he was above average - typical Willie year is 16 runs better than your JoeAvg 2Bman with the bat&legs;, 7 more with the glove. I take 2 runs away from the batting mark for the leadoff 'more at bats but lower leverage' thing, YMMV. Even so, Willie has enough for me that he climbs to the top of the backlog. I've been voting Beckley at the backlog's top, and Willie looks like him, but league strength makes me choose Randolph.

Willie vs other 2B candidates: tough, since top 2 reutrnees are Doyle and Monroe. Doyle much better bat, but not as good a glove and 2B was a LOT different in 1910. Monroe's case involves reasonable speculation. Compared to Nellie Fox, better stick, almost as good a glove, longer prime, not as durable.

TomH - you pretty much summed up exactly what I considered when evaluating Randolph and you came to more or less the same result. Randolph makes my ballot and I only need to decide whether I want to PHOM him or Ken Boyer, who are practically equivalent for me.

"Well, since win shares defines "replacement level" as zero,"

WS most definitely does not define replacement level as zero. Replacement level in WS is around .200 WPct.

WS defines replacement level as the 'marginal' level, which is .5 of league runs scored or 1.5x league runs allowed.

In a league where an average team scores 750 runs in 162 games, that's a .111 overall WPct if a team has replacement level offense and defense. A 'replacement' level offense with an 'average' pitching staff/defense would post a .212 WPct.

Well, since win shares defines "replacement level" as zero

WS most definitely does not define replacement level as zero

WS DOES define replacement level as zero. It's creator, Bill James, says so.

Read "Win Shares and Replacement Level (Or ... Pandora, shut your trap)" in the Win Shares book.
Quote from page 107: "There is no replacement level contemplated within this system."

Then again, just because he says it's so doesn't make it so. The definition of WS explicitly ignores the Pythagorean (i.e. non-linear) effects of runs on W-L. Taking those effects into account causes there to be an "effective replacement level", which is what Joe is talking about. The effective offensive replacement level is at .068 (11-151, assuming a defense that is just as execrable). The effective defensive replacement level is at .158 (25-136). A team with both components at "effective replacement level" would be at .105 (17-145); maybe if I used PythagenPat instead of an exponent of 2, it would be .111? Close enough.

In a league where an average team scores 750 runs in 162 games

The "effective offense replacement" team would score 52% of that (390 runs) and the corresponding bad defense would give up 1442 runs (1/.52 or 192% of league average).

The "effective defense replacement" team would give up 152% of that (1140 runs) and the corresponding bad offense would score 493 runs (1/1.52 or 66% of league average).

Win Shares has different effective replacement levels for offense and defense (due to its assumption of a linear conversion from runs to wins). The effective replacement level for offense is considerably lower than that for defense which is why Win Shares is so highly biased in favor of offense. Hitters get more "free" Win Shares that pitchers and fielders do not get ("free" meaning no value content).

Hm. Being quoted out of context creates confusion . . .

When I wrote above that in win shares replacment level equals zero, I was describing how to interpret the meaning of a player's fielding win shares, not the nature of the win shares zero point. If you are trying to compare the defensive values of two players at different positions as the win shares system sees it, all you have to do is look at their total fielding win shares: that is equivalent to looking at FRAR in BP's WARP, which is adjusted for position.

If you want to know how far a player was above average as a fielder at his position in win shares, to find an equivalent in the win share system for FRAA in BP's WARP, that requires some more complex calculations.

That is all I was claiming in the post above. The Randolph thread is probably the wrong place for analysis of the zero-point in the win shares system, although jimd's point about the discrepancy in replacement level between offense and defense in win shares cannot be reiterated too often, at least until more voters start to notice. . . .

jimd
The effective offensive replacement level is at .068 (11-151, assuming a defense that is just as execrable). The effective defensive replacement level is at .158 (25-136). A team with both components at "effective replacement level" would be at .105 (17-145); maybe if I used PythagenPat instead of an exponent of 2, it would be .111? Close enough.

> In a league where an average team scores 750 runs in 162 games

The "effective offense replacement" team would score 52% of that (390 runs) and the corresponding bad defense would give up 1442 runs (1/.52 or 192% of league average).

The "effective defense replacement" team would give up 152% of that (1140 runs) and the corresponding bad offense would score 493 runs (1/1.52 or 66% of league average).

What is the reasoning that specifies the "just as execrable" or "corresponding bad" defense (or offense, given the defense)? "Corresponding" in the sense that when paired with average offense (or defense) the team win would win the same number of games? Is that the criterion --founded on the Pythagen theory that runs ratio determines wins ratio, ie (W/L) a function of (RS/RA)?

Going over his career, Willie is close, but no cigar. Almost makes my ballot.

With Jim Gantner on the ballot, I decided to run my offensive system for him. The results:

Over his career, Willie Randolph was about 20 wins better than an average offensive player, and about 47 wins better than 75% of average.
Over his career, Jim Gantner was about 8 wins worse than an average offensive player, and about 13 wins better than 75% of average.
Over his career, Frank White was about 17 wins worse than an average offensive player, and about 9 wins better than 75% of average.

What does that mean for Randolph? You can see how much better on offense he was than two near-contemporaries who kept their jobs forever.

Some people who rank a little ahead of Randolph, offense only, in my system: Childs, Evers, Lazzeri, Herman, Gordon.
Some people who rank very close to Randolph, offense only, in my system: Doerr, Lopes.
Some people who rank a little behind Randolph, offense only, in my system: Myer, McAuliffe, Huggins.

What is the reasoning that specifies the "just as execrable" or "corresponding bad" defense (or offense, given the defense)? "Corresponding" in the sense that when paired with average offense (or defense) the team win would win the same number of games? Is that the criterion --founded on the Pythagen theory that runs ratio determines wins ratio, ie (W/L) a function of (RS/RA)?

It's similar to the "equivalence" of 125 ERA+ to 125 OPS+. A team that scored 25% more runs than league average would be 125 RS+. A team that gave up 80% of the league average would be 125 RA+ (1.00/.80).

"Corresponding" in the sense that the other half of the team is just as bad. The .52 RA+ offense is paired with a .52 RS+ defense. Or the .66 RS+ defense is paired with a .66 RA+ offense. When someone says "a team of replacement players", I don't think of a replacement level offense that can field at an average level for average pitchers (though maybe others do). I think of a team ALL of whose players are of little value to any other team.

Pairing a "replacement level" offense with an average defense (or vice-versa) to determine the location of the WS effective replacement level doesn't capture how low it really is.

It's similar to the "equivalence" of 125 ERA+ to 125 OPS+. A team that scored 25% more runs than league average would be 125 RS+. A team that gave up 80% of the league average would be 125 RA+ (1.00/.80).

"Corresponding" in the sense that the other half of the team is just as bad. The .52 RA+ offense is paired with a .52 RS+ defense.

So I guessed. That is one analytical sense. We don't know that such equally bad offense and defense measured in runs is equally common, or equally easy to acquire on short notice, etc. Your point is a troubling asymmetry, and we may presume Bill James didn't ground in empirical work the "unequally bad" offense and defense that play analogous roles in Win Share (I rely on jimd #90). But it may be that replacement level offense and defense are unequally bad measured in runs, if defined in another way such as by reference to the market for players.

The main point of #90, I think, is to simply to quantify the badness of the Bill James replacement level offense and defense, not to argue that they should be equally bad.

We don't know that such equally bad offense and defense measured in runs is equally common, or equally easy to acquire on short notice, etc.

True. But they are equally nasty in their impact on W-L, if you take Pythagoras' word for it.

True. But they are equally nasty in their impact on W-L, if you take Pythagoras' word for it.

Wow. He has been dead for 2,500 years, yet he has become a leader in the sabermetric movement. ;-)