I've read that thread, and it has made me look forward to Concepcion's debates. Since we're not gonna elect Aparicio, it'll be nice to have one Venezuelan in the Hall of Merit ;-)

To me, his meritousness depends on two questions:

- Was he really that good with the glove? I wasn't there for it, but the evidence suggests that he might be, and anyway, we reject citizenship to poor-fielding shortstops ;-)

- More importantly, what was behind the drought of shortstops during that time (and unless I'm mistaken, shortstops who could hit were few and far between from Wagner until the 80's)? If it was simply close-mindedness by managers and GMs, then he´s probably not a good choice. If, by pure chance, no capable-hitting shortstops were born in the 40´s-50´s, then maybe. If there was something else behind it, then he has enough Merit IMO.

That thread is worth bookmarking and saving for when we discuss David.

As for the methodology, I´ll have to read it more carefully later, as I´m a bit tired now, but to me it sounds like the end product is something like VORP+Dial, which has been accepted (as far as I can tell).

Juan,

Those are possibilities, as is the possibility that there's a theoretical issue in Dan's findngs or an assumption that's not quite right.

A couple months ago I had a nice discussion with a guy who seemed to really know his stuff on replacement. He said that generally speaking the replacement level is around -17 RAA, where RAA is FRAA + BRAA. And you adjust positionally from there. (Don't know if that -17 is accurate for Concepcion's era or McCovey's or if it's a more contemporary figure). The gentleman's point was that replacement is function of RAA, not a function of replacement, and that you have to start with RAA to get to replacement. His second point was that replacement is about sum contribution on O and D, suggesting (to me) that there is no defensive replacement level, only an adjustment for positional offensive levels, but that either way, you can't double count. I might be incorrectly interpreting his comments, and if that's true I apologize to him (if he's reading here) and to the group.

So with this conversation in mind, I feel like Dan's making a double adjustment. His policy is to use the three worst regulars at each position to come up with replacement on hitting and something of the sort for defense. I'm not really savvy enough to defend or attack that system with any zeal or with specifics drawn from his numbers, so I'm just reporting what my gut tells me (and my gut is ample!). It don't know which is best or better or what, but it feels like making two adjustments (one on O and one on D) must be double counting since the offense adjustment already implies the absence-of-offense-is-proxy-for-defensive-difficulty assumption. Therefore a subsequent defensive adjustment will boost a Concepcion or Ozzie Smith into the range that a big hitter will be in and lead to the result that Concepcion and McCovey are similar HOF/HOM candidates and that Davey is actually better.

Someone please tell me if I'm getting this wrong because I'm still trying to master replacement theory, and I don't want to slander someone else's system if I've got my ideas all jumbled. Thanks.

Thanks very much for your interest in my work.

I am very sensitive to the double-counting issue, and I am most definitely not doing so. My approach is to use the combined hitting and fielding of the worst three regulars at a position, measured in terms of total wins per season below a positionless player hitting and fielding at the league average. I'll re-post my two posts on methodology from the Concepción thread here.

1. Compile a list of starters at a given league-position-season (led team in PA among players whose primary position was the one in question).
2. Take the first player, let's say he has 650 PA. Take a theoretical team, league-average in both runs scored and runs allowed, and replace 650 PA of league-average offense with the player's stats. Distribute his extra outs saved/created evenly across the rest of the team. Calculate how many runs that team scores.
3. Take the player's FRAA (calculated by averaging BP FRAA, Fielding Win Shares converted into FRAA, and Dial for post-1986, after adjusting them so they all have the same standard deviation) and subtract it from the theoretical team's league-average runs allowed. Use Pythagenport with these RS/RA to see how many games this league-average team plus the player in question would win.
4. Repeat this for all starters.
5. Convert this wins above/below average number to a rate stat (e.g., wins above average per 650 PA).
6. Take a straight average of this wins above/below average rate for the three lowest rates in each year.
7. Average this three-worst-regulars rate for every season from 1985-2005. Compare this to Nate Silver's empirically determined Freely Available Talent (FAT) level from 1985-2005 to get the gap between the three-worst-regulars rate and the "true" replacement level (e.g., the worst three starting AL SS average from 1985-2005 was 3.5 wins below a positionless player hitting and fielding at the league average per season. Silver's FAT shortstops in the AL are 3.3 wins below a positionless player hitting and fielding at the league average. Thus, the gap is 0.2 wins per season).
8. For each season, take a nine-year moving average of the worst-three regulars average (the season in question plus four years on either side). Add on the worst regulars-FAT gap to his average. This is the replacement level. (e.g., the worst three regular SS in the AL averaged 4.6 wins below average per season from 1976-1984. Add on the 0.2 gap, and replacement level for 1980 AL SS is 4.4 wins below average per season).
9. Multiply the replacement level by the player's fraction of a season (e.g., if an AL SS had 325 PA in 1980, a replacement SS would have been 4.4*325/650 = 2.2 wins below average). Subtract this number from the player's wins above/below average to get his WARP.

====
Here are two examples, to illustrate. XR = Extrapolated Runs, PF = Park Factor, FWS = Fielding Win Shares, PPLA = a positionless player hitting and fielding at the league average, FAT = Freely Available Talent.

Dave Concepción, 1974. He hit .281/.335/.397 with 41 SB and 6 CS in 653 PA with a 99 PF, which is 84.2 XR. The average NL team in 1974 scored 670 XR in 4,161 batting outs, or .161 XR/Out. Subtracting Concepción's 436 batting outs (including his CS) from 4,161 leaves 3,725 batting outs for the rest of the team, which at .161 XR/Out is 600 XR, plus Concepción's 84.2 XR is 684 runs scored. BP has him with 19 FRAA, which I regress to 15 since BP's FRAA has a higher standard deviation for SS than Chris Dial's numbers. He had 10.0 FWS which for a shortstop with 653 PA is equivalent to 15 FRAA as well (the equation, using Chris Dial's standard deviations for SS, is 4.9*FWS- .051*PA). Averaging 15 and 15 gives you 15 FRAA. Subtract 15 runs from the league-average 670 XR and you get 655. 684 RS and 655 RA = 84.2 wins, 3.1 wins better than the league-average team. The worst three starting NL SS between 1985 and 2005 averaged 2.9 wins below PPLA per season, exactly the same as Silver's FAT SS, so the three-worst-regulars average is equal to the replacement level. The worst three starting NL SS between 1970 and 1978 averaged 3.5 wins below PPLA per season. The 1974 NL averaged 633 PA per lineup spot, and Concepción had 653 PA, so a replacement SS would have been 3.5*653/633 = 3.6 wins below PPLA in his PA. Concepción was 3.1 wins above average, so 3.1 + 3.6 = 6.7 WARP.

Willie McCovey, 1970. He hit .289/.444/.612 in 638 PA with a 99 PF which is 131.6 XR. The average NL team in 1970 scored 722 runs in 4,153 batting outs, which is .174 XR/Out. Subtracting McCovey's 355 batting outs from 4,153 leaves 3,798 for the rest of the team, which at .174 XR/Out is 661 runs, plus McCovey's 131.6 is 792 runs scored. BP has him with 3 FRAA, which I regress to 2 since BP's FRAA has a higher standard deviation for 1B than Chris Dial's numbers. He had 2.1 FWS which for a 1B with 638 PA is equivalent to -1 FRAA (the equation, using Chris Dial's standard deviations for 1B, is 8.01*FWS - .0284*PA). Averaging 2 and -1 gives 0.5 FRAA (call it 0 for convenience's sake). 792 runs scored and 722 runs allowed is 88.1 wins, 7.1 wins above a league-average team. The worst three starting NL 1B between 1985 and 2005 averaged 0.3 wins below PPLA, and Silver's FAT 1B averaged 0.2 wins below PPLA, so the gap between the three-worst-regulars average and the replacement level is 0.1 wins. The worst three starting NL 1B between 1966 and 1974 averaged 0.2 wins above PPLA, plus the 0.1 gap to the FAT level, is 0.3 wins above average. The 1970 NL averaged 632 PA per lineup spot, and McCovey had 638 PA, so a replacement 1B would have been 0.3*638/632 = 0.3 wins above PPLA in his PA. McCovey was 7.1 wins above average, so 7.1 - 0.3 = 6.8 WARP.

Concepción's 1974 and McCovey's 1970 were, thus, equally valuable seasons.

My approach is to use the combined hitting and fielding of the worst three regulars at a position, measured in terms of total wins per season below a positionless player hitting and fielding at the league average.

Dan, thanks so much for explaining!

Again, I'm not all that savvy to the ins and outs of replacement, so I just want to query this sentence to make sure I get what you're doing.

If you use FRAA for fielding, then the positionless fielder is a 0 FRAA fielder. But if you take the worst three players at a position, you may get FRAA values below 0. But getting a FRAA value below zero for the worst three guys seems like it would contradict the logic behind making a positional-hitting adjustment. When you make the hitting adjustment for position, it's, in essence, acknowledging that SS is worth more on defense than other positions due to its intrinsic difficulty, but we can't exactly say how much that intrinsic value is, so we're using the position's absence-of-offense as a proxy for its intrinsic defensive value. Therefore, to my limited understanding, a FRAA value less than zero seems incompatible with the replacement-positional-hitting adjustment. Wouldn't you need to take the zero FRAA as your replacement baseline for fielding, instead of whatever defensive value the three worst SS have, in order to make it compatible with the hitting replacement adjustment?

Or is that what you are doing, and I'm totally missing it? That's totally possible!!! ; )

Thanks again for patiently answering my questions.

And here are the careers of Concepción, Smith, and McCovey, again reposted from the other thread. The methodology is simply to calculate total wins, combining offense and defense, as measured against a positionless player hitting and fielding at the league average (otherwise known as BRAA + FRAA), for both the player in question and for a replacement player in his plate appearances, and then subtract the second from the first.

So compare the charts for '74 Concepción and '70 McCovey. '74 Concepción had 14 BRAA (above league, not positional average), worth 1.6 wins, and 15 FRAA, worth 1.7 wins, making him 3.3 wins better than a positionless player hitting and fielding at the league average in his 644 plate appearances. A league-average shortstop in 1974 (with 0 FRAA, by definition, and positional-average offense) was 16 runs/1.8 wins worse than a positionless player hitting and fielding at the league average over 644 PA. A replacement shortstop (defined as above) was 15 runs/1.7 wins worse than a league-average shortstop over 644 PA, counting both offense and defense. Adding Concepción's production above league average (3.3 wins), league average production above the SS average (1.8 wins), and SS average production above SS replacement level (1.7 wins), we get 6.9 wins above replacement. Look, ma--no double-counting! :)

Similarly, '70 McCovey had 7.1 wins above a league (not positional) average hitter, and was an exactly league-average defensive first baseman, making him 7.1 wins above a positionless player hitting and fielding at the league average in his 638 plate appearances. A league-average first baseman in 1970 (with 0 FRAA, by definition, and positional-average offense) was 2.4 wins better than a positionless player hitting and fielding at the league average over 638 PA (1970 was a particularly good year for NL first basemen, not that it matters). A replacement first baseman (defined as above) was 2.0 wins worse than a league-average first baseman over 638 PA, counting both offense and defense--which is to say, a replacement first baseman was 0.4 wins better than a positionless player hitting and fielding at the league average over 638 PA. Yes, that means I'm saying that above-league-average offense was freely available in 1970 if you could stick the guy at first base. (This is no longer true, largely thanks to the DH creating 14 new jobs for good hit/no field guys who would previously have been in the minors). Adding McCovey's production above league average (7.1 wins), league average production below the 1B average (-2.4 wins), and 1B average production above 1B replacement level (2.0 wins), we get 6.7 wins above replacement. Once again, no double-counting.

Here is the full breakdown for Concepción (not counting his two years at 2B), along with his most comparable player (Ozzie Smith) and his polar opposite (Willie McCovey). I hope this chart provides enough detailed data for lots of people to draw interesting distinctions and conclusions. The Hall of Fame cutoff line is around 60 career WARP. Yes, I think Dave Concepción had a more valuable career than Willie McCovey, and here's my evidence.

OWAA = offensive wins above the overall league (not positional) average for that year.
DWAA = defensive wins above the positional average for that year.
Av-Pos = wins gap between the overall positionless league average and the positional average in the player's plate appearances for that year.
Pos-Rep = wins gap between the positional average and the positional replacement level (summing offense and defense) in the player's plate appearances for that year.
TWAA = wins above the overall positionless league average (OWAA + DWAA).
PWAA = wins above the positional average (OWAA + DWAA + Av-Pos).
WARP = wins above the positional replacement level (OWAA + DWAA + Av-Pos + Pos-Rep).

Numbers may vary by fractions from what I've posted previously due to sac hits and sac flies, sorry. 1960, 61, 72, 81, 94, and 95 are straight-line adjusted for season length.



Dave Concepción

SeasonRank Year   PA OWAA DWAA Av-Pos Pos-Rep TWAA PWAA WARP
Best       1981  455  2.2  1.4    1.8     1.9  3.6  5.4  7.3
2nd Best   1974  644  1.6  1.7    1.8     1.7  3.3  5.1  6.9
3rd Best   1979  655  1.7  1.5    1.3     2.1  3.2  4.6  6.7
4th        1976  623  0.9  2.2    1.3     2.1  3.1  4.5  6.6
5th        1978  619  1.5  0.8    1.1     2.3  2.4  3.5  5.8
6th        1975  546  0.3  2.1    1.7     1.3  2.3  4.0  5.3
7th        1977  615 -0.2  2.1    2.1     1.2  1.9  4.0  5.2
8th        1982  615  0.0  1.6    1.1     2.1  1.7  2.8  4.9
9th        1973  348  1.2  0.7    1.3     0.6  2.0  3.3  3.9
10th       1980  657 -1.0  0.6    1.8     1.7 -0.4  1.4  3.0
11th       1986  335 -0.4 -0.1    0.7     0.9 -0.5  0.3  1.2
12th       1984  579 -1.1 -0.9    1.7     1.4 -2.0 -0.2  1.1
13th       1972  416 -1.9  0.5    1.2     1.1 -1.4 -0.2  1.0
14th       1970  290 -0.5 -0.2    0.8     0.8 -0.7  0.1  0.9
15th       1985  616 -1.6 -0.8    1.2     2.0 -2.3 -1.1  0.9
16th       1983  581 -2.6  0.0    1.0     2.1 -2.6 -1.6  0.5
17th       1971  345 -2.4 -0.2    1.2     0.8 -2.6 -1.3 -0.6
TOTAL           8939 -2.1 13.1   23.3    26.1 11.0 34.4 60.4

Ozzie Smith

SeasonRank Year    PA OWAA DWAA Av-Pos Pos-Rep TWAA PWAA WARP
Best       1985   603  1.9  2.4    1.2     2.0  4.3  5.5  7.4
2nd Best   1987   687  2.3  1.5    2.0     1.3  3.8  5.8  7.1
3rd Best   1991   633  2.6  0.9    0.4     2.6  3.5  3.8  6.4
4th        1992   576  2.1  1.7    0.6     2.0  3.8  4.4  6.3
5th        1988   646  1.4  1.6    1.0     2.0  3.0  4.0  6.1
6th        1986   598  1.9  1.0    1.3     1.7  2.9  4.2  5.9
7th        1989   648  1.5  1.1    0.9     2.0  2.6  3.5  5.5
8th        1982   555  0.1  2.3    1.0     1.9  2.4  3.4  5.3
9th        1984   468  1.0  1.4    1.4     1.1  2.4  3.8  4.9
10th       1978   636 -0.8  1.9    1.1     2.4  1.1  2.2  4.6
11th       1983   616 -0.2  1.4    1.0     2.2  1.2  2.3  4.5
12th       1980   683 -1.4  1.9    1.8     1.7  0.5  2.3  4.0
13th       1993   585 -0.4  1.5    0.4     2.2  1.0  1.4  3.6
14th       1990   570 -0.5  0.6    1.1     1.5  0.1  1.2  2.8
15th       1981   483 -3.2  1.8    1.7     2.2 -1.4  0.4  2.5
16th       1996   254  0.2  0.5    0.3     0.8  0.7  1.0  1.8
17th       1994   417 -1.0 -0.4    0.8     1.8 -1.5 -0.7  1.1
18th       1979   630 -3.1  0.7    1.3     2.0 -2.5 -1.2  0.8
19th       1995   175 -1.3  0.3    0.3     0.5 -1.0 -0.6 -0.2
TOTAL           10466  2.9 24.1   19.7    33.8 27.0 46.7 80.6

Willie McCovey

SeasonRank Year    PA OWAA DWAA Av-Pos Pos-Rep TWAA PWAA WARP
Best       1969   612  9.1 -0.4   -2.4     2.1  8.7  6.3  8.4
2nd Best   1970   638  7.1  0.0   -2.4     2.0  7.1  4.7  6.7
3rd Best   1968   595  6.7 -0.8   -2.0     1.7  5.9  3.9  5.6
4th        1963   625  5.6 -1.1   -2.2     2.7  4.5  2.3  5.0
5th        1966   581  5.3 -0.5   -1.8     1.7  4.8  3.0  4.7
6th        1965   628  5.1 -0.2   -1.5     1.3  4.9  3.3  4.6
7th        1967   527  4.7 -0.3   -1.3     1.0  4.4  3.1  4.1
8th        1973   488  4.3 -0.6   -1.2     1.0  3.7  2.5  3.5
9th        1974   440  4.0 -0.5   -1.1     0.9  3.5  2.4  3.3
10th       1962   253  2.4  0.2   -0.8     1.0  2.5  1.7  2.7
11th       1959   216  2.9  0.0   -0.4     0.3  2.8  2.4  2.7
12th       1961   365  2.1  0.5   -1.3     1.0  2.6  1.3  2.3
13th       1971   393  3.1 -0.9   -1.6     1.5  2.2  0.6  2.1
14th       1977   545  2.5 -0.8   -1.6     1.7  1.7  0.1  1.8
15th       1975   470  1.9 -0.2   -1.5     1.4  1.8  0.3  1.6
16th       1960   305  1.5 -0.4   -0.8     0.6  1.1  0.3  0.9
17th       1964   425  1.1 -1.0   -1.2     1.6  0.1 -1.1  0.5
18th       1979   389  0.4 -0.6   -1.0     1.0 -0.2 -1.1 -0.1
19th       1978   387 -0.3 -0.2   -0.7     0.8 -0.4 -1.2 -0.4
20th       1976   250 -0.5 -0.1   -0.7     0.7 -0.6 -1.2 -0.5
21st       1980   126 -0.5 -0.1   -0.3     0.3 -0.6 -1.0 -0.6
22nd       1972   302 -0.1 -0.5   -1.1     1.0 -0.5 -1.6 -0.7
TOTAL            9559 68.4 -8.5  -29.1    27.2 59.9 30.8 58.0

I've read that thread, and it has made me look forward to Concepcion's debates. Since we're not gonna elect Aparicio, it'll be nice to have one Venezuelan in the Hall of Merit

Well if not Davey, then maybe Vizquel,
and if Omar can't make it in, then surely the guy that is 3rd in lifetime winsfor a Venezuelan will make it.

Or maybe that guy who once got traded for Kevin Stocker....

Eric, my motto is the following: There is no such thing as replacement statistics. There are only replacement players. Again, replacement level isn't some fuzzy concept like the meaning of life. It is a real, observable, and empirically measurable, and Nate Silver has measured it. I think you are confusing yourself far too much by trying to separate out offensive and defensive replacement levels, which is misguided and pointless. All we want to do is compare the total contribution above/below league average of the player in question to the total contribution above/below league average of a replacement player in his playing time.

Now, I'm happy to get into discussions about the nitty-gritty of my approach to extrapolating Nate's empirically derived replacement levels back in time; I'm sure it can be improved. But for convenience's sake, let's just keep this simple.

In an average year from 1985-2005, the average production of the worst three regulars at a position will be equal to the freely available replacement level. So let's take a hypothetically average year (10 runs = 1 win), and say that we've identified the following three players as the worst regular SS in the league:

1. Full season, 35 runs below overall league average hitting, league average fielding.
2. 3/4 season, 34 runs below overall league average hitting, 11 runs above average fielding.
3. Half season, 4 runs below overall league average hitting, 16 runs below average fielding.

The first step is to convert these to rates. Guy 1 is 35 total runs (35 + 0) below a positionless player hitting and fielding at the league average per season. Guy 2 is 23 runs below average (-34 with the bat, +11 with the glove), and did so in 3/4 of a season, so 23*4/3 = 30 runs below average per season. Guy 3 is 20 runs below average (4 with the bat, 16 with the glove), and did so in half a season, so 20*2 = 40 runs below average per season. Average those three together (I intentionally equal-weight these), and we get a replacement level for SS of 35 total runs = 3.5 wins below average per season.

From there, the rest is simple. Guy A is 15 runs above positionless league average with the bat and 7 above average with the glove in 4/5 of a season playing shortstop. He's 22 total runs/2.2 wins above positionless league average. A replacement SS would have been 3.5 * 4/5 = 2.8 wins below positionless league average in his playing time. 2.2 + 2.8 = 5.0 wins above replacement, which is a good All-Star level. Guy B puts up 10 runs below positionless league average with the bat, and is a -9 fielder, in half a season. He's 19 total runs/1.9 wins below positionless league average. A replacement player would have only been 17.5 total runs/1.75 wins below positionless league average in his playing time, so he is 0.15 wins below replacement, and deserved to lose his job.

I really don't think this is very complicated. What's trickier are the mechanics of actually calculating said replacement level.

"In an average year from 1985-2005, the average production of the worst three regulars at a position will be equal to the freely available replacement level."

Dan, I still thinks this assumes an optimal GM selection of talent - and that's where some of us disagree.
You're capturing some value of owning Davey instead of the stiffs that the other teams played - but that's not necessarily fully, or even substantially, a conclusion of Davey's merit.

If the other GMs are dopes - and I think they were - that doesn't mean I have to elevate someone else in the process.

If basestealers were swiping 200-300 bags a year in that era, and the only way to stop it was a rabbit at SS, and Concepcion matched the other rabbits defensively and outhit them by a lot, then yes, DC is a HOF/HOM guy.
But if not, then he may just be a one-eyed man in the valley of the blind - and I'd rather elect two-eyed guys from all eras, with all due respect to the visually challenged community.

Juan V, everything I've seen, including Michael Humphreys' Defensive Regression Analysis, suggests Concepción was a superlative, Ozzie-level defensive shortstop for about six years. And as for the drought, the real drought was in the AL centered around 1980, not the NL. In the NL, SS replacement levels in the 70s were only 0.5 wins lower than they were in the 90s, and comparable to AL SS replacement levels in the 90s (thank you, Pat Meares and friends). My theory is that the gap between the #16 and #26 SS in the world is much bigger than the gap between the #16 and #26 1B, and that that showed up due to the expansion in the 60s and 70s.

Right, Howie, we've hit this point before, and I don't necessarily disagree with you. I do think that the answer lies in attempting to derive defensive translations, a project which I hope to get to this week.

Shouldn't replacement level depend more on a percentage of an average player than what the three worst players were? TO me it seems that by measuring what the three worst players are each year you woudl get distortions if there is only 1 or at least 5 really bad players in a league.

And if Concepcion is really better than McCovey, then what is the dissonance between WARP, WS and your method? I guess to me it is that you are giving Concepcion way too much credit for playing in an era with historically bad SS's.

KJOK, you got your RCAP machine cranked up? How does Davey C come out? Should closely mirror what Dan R has done here.

We're back up! :-D

Tom,

Not to speak for KJ, but the Sinis 'cyclopedia shows the following
RCAA: -165
RCAP: +126

Mark S., I strongly believe that using league average as the basis for replacement level, while a widely accepted approach, is borderline ludicrous. The presence of a few superstars at a league-position (AL SS in the early 80s and late 90s come to mind) pulls up the league average, but does not make an AAAA player any better. Whether you want to define replacement level in relation to backup players or worst X regulars or some mixture of the two is up to you, but I simply don't see why the performance of the best players in the league should have any bearing on how we calculate the performance of the worst players in the league. To calculate replacement level, you need to look at replacement players, not average ones. If I can make this a mainstream view I will have done a service to the analysis community.

Again, my actual methodology is to use a nine-year moving average of the average wins-below-league-average-per-season rate of the worst three starters per year, adjusted for the gap between that level and Nate Silver's empirically derived Freely Available Talent levels from 1985-2005. I would hope that using a sample size of 54 players would smooth out the random variations you describe in any given year.

The dissonance between WS and my method is that WS doesn't have a replacement level. Let's check my same example of '74 Concepción versus '70 McCovey. By my measure, the worst three starting NL SS in 1974 were Frank Taveras, Enzo Hernández, and Craig Robinson. Concepción had 25.4 WS in 653 PA. Taveras had 5.6 WS in 367 PA (10.0 per 653). Hernández had 9.3 WS in 563 PA (10.8 per 653). Robinson had 6.0 WS in 506 PA (7.7 per 653). Average them and you have 9.5 WS in Concepción's playing time, making Concepción 15.9 WS above replacement. Similarly, the worst three starting NL 1B in 1970 were Deron Johnson, Bob Watson, and Lee May (McCovey, Jim Hickman, Wes Parker, Bob Robertson, Ron Fairly, Dick Allen, Donn Clendenon, Orlando Cepeda, and Nate Colbert were all better). McCovey had 33.5 WS in 638 PA. Johnson had 18.5 WS in 650 PA (18.2 per 638). Watson had 8.9 WS in 362 PA (15.7 per 638). May had 13.6 WS in 649 PA (13.4 per 638). Average those and you have 15.8 WS in McCovey's playing time, making McCovey 17.7 WS above replacement. OK, so Win Shares and I disagree by 17.7-15.9 = 1.8 WS = 0.6 wins here. We're hardly on different planets.

As for BP's stats, WARP1 has McCovey with a 1.4 win edge over Concepción in '74, 11.3 to 9.9 (again, hardly enormous), but I don't think that takes into account the varying replacement levels over time. WARP3 agrees with me completely: 10.7 WARP3 for '74 Concepción, 10.6 for '70 McCovey, so I imagine WARP3 is properly capturing the adjustment. A similar dynamic takes place at the career level--McCovey somewhat bests Concepción in WARP1, 112-100, but Concepción wins slightly on WARP3, 110-107, again exactly in agreement with my measure. Clearly the BP WARP1-WARP3 adjustment is capturing the same phenomenon that I've found.

Also, to reiterate--I don't necessarily think Concepción is a Hall of Famer. I just think he's very comparable to Willie McCovey, who (sacrilege!) would not have been near the top of my Hall of Merit ballot if I were voting when he was elected (nor Stargell). Before the DH, big-hitting 1B were a dime a dozen; you had to really be amazing to earn your keep, and McCovey only had two years at a superstar level, a few more as an All-Star. His lack of durability definitely hurts him, and he should have retired years before he did--was he just hanging around to get to 500 HR?

As for those of you questioning my use of the worst three regulars for replacement level (remember, what I did above is just an example, in fact, I use 54 players (not 3) to determine the replacement level), you need to make one of two arguments: 1. Nate Silver's FAT levels don't accurately represent replacement level or 2. the relationship between the worst-three-regulars average and the FAT level must have changed enormously between the modern game and the 1970's. The former seems outlandish to me. The latter may well be true, but I don't know how to study it for the reserve clause era.

Can someone fix Page 1 of this thread, or is it just me?

I just read Nate's FAT piece. And, interestingly enough it matches with my general idea of replacement in that if you take the FRAA of the replacement levels his research suggests, and create a team of just replacement guys on offense and pythagpat out their win% and then you balance that with a pitching staff that achieves the same win%, you get a 47 or 48 win team. Which is about where I've always imagined a team of replacements would end up. See, I can get at least a little something about replacment sort of right!!!! Or better phrased: it matches my idea, so IT MUST BE RIGHT! ; )

Dan,

I'd love to know how your system sees the candidates on, say, the 1975-1992 ballots. Just to get a sense of how it compares to our results. I mean if you have that handy.

Unfortunately, my system is very much a work in progress, as it is phenomenally time-consuming. I have NL SS from 1893-2005, NL 1B and 3B from 1947-2005, and AL SS from 1960-2005. I also have NL OF from 1893-2005, but I'm not satisfied with them yet because until 1980, the worst three CF average was actually better than the worst three corner OF average, so guys like Mays and Snider are effectively penalized (by some 0.2 wins) for playing in center. (any idea how to fix this would be greatly appreciated.) Moreover, guys from the 50s all come out badly because obviously replacement level was closer to average pre-expansion. The post-1980 data for NL OF should be good, though. I'm working on NL 2B now. If you want I can send you a spreadsheet with my data so far.

Is there a way to back-of-the envelope your system? Take a couple short cuts where the calculation gives you the least bang for the buck and come up with initial estimates. Perform the in-depth calculations only where you need the granularity.

Well, let's see. BP's BRAA is very well-calculated, it almost always is a dead match for mine. Defense above average is slightly trickier since you simply can't eyeball Win Shares without also looking at playing time--a 7 fielding win share season by a SS is below average if he plays 162 games; a 4.5 fielding WS season by a right fielder could be +18 if he only played 2/3 of a season. So maybe just regress BP FRAA by 1/3? The replacement levels are the trickiest because what I feel I am adding here, beyond what BP WARP gives you, is contextual replacement level, how good freely available players actually were in a given time period, as opposed to some shortcut. Until I've done the research on a given league-position--which requires having wins-above/below-average numbers for every starter in the league over the entire time period I'm studying--I don't know what I'll find. Let me look at the data I have so far and see if any patterns leap out at me that I might be able to extrapolate.

my assessment of Dan R's method:

- comprehensive
- accurately captures value above real-world replacement level

issues:
- should both leagues be used as baseline or just the one the player is in?
- 'worst 3 regulars' may need to be adjusted for 8 team leagues vs 16 team leagues when comparing players across eras
- we've had the "value" vs "ability" debate for a while. Joe Sewell would come out lookin' great by this method, but many voters were not persuaded that just because many SS didn't hit much in 1925 was not a reason to HoM Mr. Sewell.

TomH--one reason to keep the leagues separate is that wins above average are always lower in the AL due to the presence of the DH. If you take two 5-WARP shortstops with identical playing time in the AL and NL in a given year, the AL SS will be, say, 1.5 wins above average with a -3.5 win replacement level, while the NL SS will be 2 wins above average with a -3 win replacement level.

Worst 3 regulars most certainly does need to be adjusted for expansion. I just tested this last night for the NL, and found the following adjustments:
8-team league: add 0.9 WARP per full season (eg, if a guy has 325 PA add 0.45 WARP)
10-team league: add 0.5 per full season
12- and 14-team leagues: add 0.3 per full season

I think the biggest problem with my system at the moment, by far, is that yes it does assume optimal GM allocation of talent across positions, and it can't handle situations where one position may have the most talent, period, all the way down to the replacement level. This may or may not be such a big issue for 70's/80's SS. But it is a huge issue, with, say, 1890's outfielders, since centerfielders were hitting the crap out of right fielders. I am at a loss as to how to address this, and any help anyone could provide would be greatly appreciated.

If during a certain period, most all of the best OFers were playing CF (like in the 19th century and in the NeLs), you could use to bottom nine OFers for replacement, or maybe the combined total of the bottom three for each position. This would give you an OF replacement level, which in some ways would work since the best OFers are mostly playing the same position.

Dan and Tom,

The last point about using two leagues is an interesting question. On one hand, all the teams are in competition for the same pool of players. On the other hand, the leagues will each have different replacement levels and win values, especially once the DH arrives. Must be a compromise somewhere, right?

An additional wrinkle is that the flow of talent once it's in the big leagues has changed. If I recall correctly, and I may not please correct me, until quite recently the waiver system went bottom-to-top within league THEN bottom-to-top in the other league. So FAT wouldn't necessarily be available to the other league if were already in a major league. And ditto for trades where until recently a player had to clear waivers in his own league before being traded to the other league, regardless of the time of year. I've seen this explained as the reason for the relative paucity of blockbuster interleague trades until like the 1970s or so. (Again, I might have this a little bit wrong, I bet Treder or cblau will have the story more accurate than I do.)

Mark S., but how do I adjust their FRAA so that they are all on an equal footing? Clearly +5 in CF is not the same as +5 in RF...

Can someone fix Page 1 of this thread, or is it just me?

Its Post #90. Cblau's code-block is extremely (and unnecessarily) wide and stretching out the entire page.

I think if John/Joe reformatted that table a bit, it would fix the page.

Dan,

Good question. I guess my answer would be to leave fielding as is and only play with the offensive replacement levels. If a guy is truly not good enough defensively to play CF then he will not play CF, even if all of the better players (generally the better hitters in this context) are playing in CF. Therefore, I guess I would say keep fielding as is and only play with the offensive side of things.

These things happen because there are simply not enough good athletes to fill all the outfield positions so the best hitter tends to be the best fielder. Think slow pitch softball where the best hitters are almost always playing SS or CF because they are the best athletes. Also, you may want to note that CFers in the 19th century did not get many balls because there were fewer fly balls to be had. Some (karl) even argue that 1B may have been above CF in the defensive spectrum during this time, which I guess could be defensible. In fact we have bandied about the idea that CF didnt' become a really important defensive position until after WWII, which may in part explain why so few great hitters played there after Mays/Mantle/Snider. It's pretty bare between them and Puckett/Griffey.

Actually now that I think about this, I guess it isn't too big a deal to just punt defensive differences (thinking that it may be more accurate to ignore them than to try and artificially create them) since they were small and the best athletes tended to be the best hitters as well (of course I could be wrong). I guess after this useless post I will have to think more about this.

I once tried to come up with defensive translations for 1893-1906, but no matter which way I sliced it, guys who played in both CF and LF/RF invariably had a higher BP defensive Rate in CF. D'oh.

Its Post #90. Cblau's code-block is extremely (and unnecessarily) wide and stretching out the entire page.

I think if John/Joe reformatted that table a bit, it would fix the page.

I amended Cliff's post using the "pre" tags instead of the "code" ones, but to no avail. I don't know where the problem lies, David.

Tangotiger has found that the defensive gap in the modern game between CF and LF is 8 runs. Can anyone make any educated guesses as to how that gap might have evolved over time? Would it have been bigger or smaller in the 1890's? Smaller, I imagine, due to fewer BIP to the outfield in general...

I'd like to know more about Ellis Kinder. His minor league career is a little strange. According to the Professional Baseball Players Database, he pitched briefly in the KITTY League in 1936 and again in 1938 before really starting his pro career in 1939 at age 24. He pitched there most of the next 3 years before moving up to the Eastern League. His stats are generally quite good, but he apparently missed all of 1943 due to a suspension, then missed 1945 due to the war. In 1944, he was 19-6 with a 2.80 ERA in the American Association.

Any idea what the suspension was about? And 1945 was the only year he was in the service or whatever (war related)?

sunny,

You needed to ask? Third strike for PEDs. Duh.

I amended Cliff's post using the "pre" tags instead of the "code" ones, but to no avail. I don't know where the problem lies, David.

I think its just that the table was way too wide. Extra whitespace that was probably from copying it from somewhere else. When extremely wide lines are code- or pre- tagged it will only wrap the lines as a last resort... and it will stretch the width of the entire page as a result. Its not a *huge* deal but I think it was the source of the problems. Post #139 above is a reformatted version of his table.

I can't find details, but my guess is that Kinder was placed on the ineligible list because he chose to work in a defense job rather than playing organized baseball. The papers of the time suggest that it was a fairly common practice for minor leaguers to leave their teams in order to take jobs in shipyards and the like (probably because they paid better, and because of concerns about whether the minors would actually survive due to wartime travel restrictions), and there are several instances where Judge Branham, the minor league czar, placed such players on the ineligible list.

-- MWE

Dan R
TomH--one reason to keep the leagues separate is that wins above average are always lower in the AL due to the presence of the DH. If you take two 5-WARP shortstops with identical playing time in the AL and NL in a given year, the AL SS will be, say, 1.5 wins above average with a -3.5 win replacement level, while the NL SS will be 2 wins above average with a -3 win replacement level.

Why not use all contemporary major leagues to estimate the mean difference between positions?

Dan R
Tangotiger has found that the defensive gap in the modern game between CF and LF is 8 runs. Can anyone make any educated guesses as to how that gap might have evolved over time? Would it have been bigger or smaller in the 1890's? Smaller, I imagine, due to fewer BIP to the outfield in general...

For some time, including the deadball era I guess, LF must rank above RF. But that is a secondary question.

How did Tom Tangotiger estimate the CF-LF defensive gap for the "modern game"? Does that mean Retrosheet era?

Bill James is famous here for overrating CFs, especially around the turn of the century, but he does rank contemporaries Jimmy Sheckard and Fred Clarke the alltime great LFs (which isn't silly, by reputation), with ratings that would make many the mother of a CF proud. What drives those findings?

Paul Wendt, I could do that, I'd just have to be careful to properly factor in the DH effect so as not to wrongly penalize AL players.

The "modern game" is the 1999-2003 period for which there is publicly available UZR data. He looked at guys who had played multiple positions, controlled for their experience at the position, and saw how many fielding runs per 150 games players gained/lost when they switched positions in the aggregate.

I really have two problems now. The first is estimating the size of the gap between the outfield positions (according to BP, LF was almost as tough as CF in the 1890's, with RF waay behind). But once I've done that, I have another issue, which is figuring replacement level for 1890's outfielders. Let's just say, hypothetically, that in the 1890's, CF is 4 runs above LF and 12 above RF. RF were also usually really bad hitters, much worse than the other two outfield positions. Thus, if I add 8 runs per season to all LF FRAA scores, 12 runs per season to all CF FRAA scores, and then measure all OF on an equal footing, I'm going to get some extraordinarily high WARP scores for 1890's outfielders.

For example, I have 1895 Ed Delahanty at 9.0 wins above a positionless player hitting and fielding at the league average (PPLA). The 20th century average replacement level for corner OF is about 0.8 wins below PPLA, so Delahanty would be between 9.5 and 10.1 WARP in any other era. But in 1895, when guys like Bill Hassamaer were ringing up 69 OPS+'s with bad defense and still started in the outfield, it looks different. The three-worst-regulars average for outfielders from 1893-99 was 2.6 wins below PPLA. Moreover, those guys were all right fielders, and presumably if they had played left they would have given up 8 more runs on defense. So 9.0 wins above PPLA, plus the 2.6 between the worst 3 starting outfielders and PPLA, plus 0.8 for playing LF, is a monster 12.4 WARP for Delahanty, which is almost getting into Barry Bonds/Honus Wagner territory.

I have no doubt that in 1895, Ed Delahanty was 12.4 wins (season length adjusted) better than the three worst starting outfielders that year. My question is, do those three worst starting outfielders properly represent the freely available talent (FAT) level? If they do, then outfielders were just very scarce in those days--you could find two pretty easily, but getting a third was exceedingly difficult, and if your team could put a decent guy in RF you'd have a big advantage over the other teams, just like AL shortstops around 1980. In that case, Delahanty really was 12.4 WARP, 1890's outfielders are dramatically underrated, and guys like Van Haltren and Jimmy Ryan are no-brainer HoM'ers.

To avoid that conclusion, we have to argue that NL teams consistently and repeatedly gave full-time jobs to players who were worse than the FAT level in the 1890's. Is that true? Why would they have done so? And if they did, how should I re-calculate replacement level for outfielders from that era? Remember, I don't want to use a percentage of league or positional average, because my entire point is that the relationship between replacement level and average is not stable over time. I want to use real replacement players and average their rate stats. If the Farmer Weavers and Bill Hassamaers of the world were, in fact, below the FAT level, who wasat the FAT level?

In short, IF you believe that Ellis Kinder was a ML pitcher in 1944 (19-6 in the American Association), and IF you believe he proved himself to be a ML caliber pitcher in 1940 (20-6 in the Southern League), then there is a fair amount of MLE credit worth at least talking about.

1943 and 1945--out of action due to the war

1944--coulda/woulda/shoulda been in the weakened MLs, except that he was inactive in 1943 so he was sort of forgotten

And maybe even 1941-42 you might argue that he was held back unreasonably, though I agree that's a stretch.

But I could see up to 3 full years of MLE credit. And he was a starter at that time. Unfortunately his ERA+ his first 3 years of ML play (established value as best [as only] we have) was 112-86-117. He began to step forward a bit, then, at 130-15 (as a starter) and 175-153-227 etc. as a reliever. So those would be 3 years as an average ML pitcher (average OPS+ of 105 first 3 years). Given his age I would say he was the same pitcher in 1943-44-45 that he was in 1946-47-48, not an up and coming young guy with developing skills. He threw 450 IP his first 3 full ML seasons. Had he been in the MLs those 3 allegedly MLE years, he probably would have thrown 450 those 3 years, then more like the 600+ that he threw in his second, third and fourth (and fifth) ML seasons (that is, 600+ in any rolling 3 year period). So I would add 600 IP at 105 for a career total of 2100 IP, but with an ERA+ that would drop from 125 to maybe 119.

Then figure that he had that relief pitcher peak of ERA+ 175-153-227 with 45 saves in over 300 IP. Make hima relief pitcher peak candidate who also has 2000 IP at 119. Like I said before, he is now looking a lot like Rollie Fingers.

Dan R - regarding the difference between the 3 OF positions in the 1890's. I recently did a study on Hugh Duffy's play throughout the OF and how the results showed him to be the primary defensive outfielder for his team, even in some years where he was not the CF. It seems that sometimes back then, the bright managers optimized their outfielders by placing the greatest one in Left insted of center. You can see my finding in the Duffy/GVH/Jimmy Ryan thread.

link? can't find the thread.

http://www.baseballthinkfactory.org/files/hall_of_merit/discussion/23522/P100/

Piecing together bits from the Sporting News:

In 1940 and 1941, Kinder was pitching in the Kitty League (Class D), not the Southern Association. He went to Memphis in 1942; the Southern Association was class A-1 (equivalent to AA today). In mid-season, Kinder was sent down to Jackson, in the Class B Southeastern League. In the 1942-1943 offseason, Kinder apparently took some sort of defense job (the first reference I have in 1943 mentions Kinder as a player who "had a tryout with Memphis last year" and who was "planning to return to baseball" - from where, the article didn't say), and the reason I conclude that he stayed in a defense job is that the next reference I have is to his reinstatement from the ineligible list. In 1944, Kinder was back with Memphis, which was still in the Southern Association (not the American Association). Memphis sold him to the Browns in the 1944-1945 offseason.

Based on that, I don't think there's any justification for war credit for Kinder.

-- MWE

What fraction of Dave Kingman's value comes from hitting home runs? Can you think of anyone else quite like him?

Some quick thoughts, on the "bat" side:

Rose: Any system that first classifies players by position will have trouble with him. I went ahead and sorted him into the corner outfielders pile to see what comps that turned up - and got that he's a fair match (less peak value, more career value) on offensive value to Yastrzemski. Carew is another pretty good match - again, Carew has a better top year or two, but Rose has more of them. Even Rose's "hang around" years at the end weren't all that bad. Under the specialty "leadoff hitter," who was the best of the late 60's - early-to-mid 70's? Rose ahead of Brock, for sure; not so sure about classifying Bonds, but Rose outlasted him.

Grich: his offensive value in my system is a little above Frisch, Childs, Herman, Gordon, and Doerr (with Evers and Lazzeri in the same mix). To get from there to an "elect me" spot, there will have to be a defensive side to his argument - so how good was his glove?

Perez: I've got him with more peak than either Vernon or Beckley - not that that constitutes high praise. I've got a freak-show peak/career balancing stat that oddly enough puts Perez into a tie with Sisler. Since I didn't vote for Sisler, that's no argument either. I could see using defense and career length to put him ahead of Boog Powell - but I'm not voting for Powell. So far, I'm striking out on finding reasons to put him on my ballot.

I haven't worked up Cedeno yet, or the shorter-career sluggers: Foster, Baker, Kingman.

Thx, Mike, especially for clarifying what leagues Memphis and Jackson are in. Still, a guy who sat out 1943 and 1945 due to the war--surely he would have been playing ball if not for the war--and was in the majors in 1946 makes me wonder "what might have been." But it wouldn't have been the 3-4 additional years he needed to get into consideration here.

OCF,

I think the consensus is that Grich was a great defensive second baseman. By win shares, he's an A, with a better career rate of WS/1000 defensive innings than anybody with over 12,000 innings at second except Mazeroski.

Here's the top 10, by rate, in over 12000 innings:

1. Mazeroski, 6.18 (18,301 innings)
2. Grich, 5.68 (15098)
3. Schoendienst, 5.61 (15782)
4. Frank White, 5.58 (17852.7)
5. Nellie Fox, 5.54 (20127)
6. Frankie Frisch, 5.42 (15483)
7. Hughie Critz, 5.41 (12847)
8. Manny Trillo, 5.30 (12796.7)
9. Bid McPhee, 5.25 (18789)
10. Willie Randolph, 5.23 (18648)
Doerr, 5.23 (16377)

WARP doesn't put him among the top defensive second basemen of all time, but it still sees him as far above average: 110 FRAA for his career, as opposed to 169 for Doerr and 166 for Frisch. I suspect the truth is somewhere in between: WARP clearly shows that the Baltimore defenses of the early 1970s were great (Palmer being their beneficiary), but it doesn't show either Belanger or Grich as extraodinary defenders. Well, maybe Belanger. Win shares I suspect overrates Grich's defense slightly because of "great team" effects. Nevertheless, Grich is clearly in Doerr/Frisch territory with the glove, which is all he needs to be a lock for the HoM.

On Perez: My system sees him as similar to Rusty Staub: a little worse with the bat (this was a surprise to me), more valuable with the glove. What does your system see with Perez if you compare him to corner outfielders as well as first basemen?

A couple more: Cedeño comes out OK, especially considering his defensive reputation, but nowhere near Wynn's peak. I've got him behind Van Haltren and Duffy, and probably behind Ryan, and that's not in my top 30. (But we'll always have September of 1985!)

Kingman in an RCAA-based system? Not enough RC, way too many outs. His case adds up to "not a candidate." Jeff Burroughs looks much better in this system, and Burroughs is not a candidate.

OCF,

The player eligible in 1992 that I'm very curious to see put through your system is George Foster. How does he come out?

Chris: presented without further commentary.



Vernon's stinko -24 year was 1948, right in the middle of his career.

Foster? Eh. Tommy Henrich (without war credit). Roger Maris. His top few years in the system clock in at 58 46 45 40 37. He's got about a 7-year run as a good player and drops off sharply on both sides. Now the system may very well wind up preferring him to Jim Rice, but then it REALLY doesn't like Rice.

Still, a guy who sat out 1943 and 1945 due to the war--surely he would have been playing ball if not for the war--and was in the majors in 1946 makes me wonder "what might have been."

I seriously doubt that it would have been a lot different, quite frankly. In fact, one might be able to make an argument that Kinder's progress through the minors was accelerated rather than hindered by the war. The minor leagues were hit much harder by the loss of able-bodied young men to the war than were the majors - by 1943, there were only nine minor leagues in operation - and this may very well have created an opportunity for Kinder in the upper minors in 1942 that would not have existed with a full complement of players on hand.

-- MWE

Dan R
I really have two problems now. The first is estimating the size of the gap between the outfield positions (according to BP, LF was almost as tough as CF in the 1890's, with RF waay behind).

plausible to mee ;-)
Is this deduced from Davenport Translation webpages?

I have no doubt that in 1895, Ed Delahanty was 12.4 wins (season length adjusted) better than the three worst starting outfielders that year. My question is, do those three worst starting outfielders properly represent the freely available talent (FAT) level? If they do, then outfielders were just very scarce in those days--you could find two pretty easily, but getting a third was exceedingly difficult, and if your team could put a decent guy in RF you'd have a big advantage over the other teams, just like AL shortstops around 1980. . . . [Else] we have to argue that NL teams consistently and repeatedly gave full-time jobs to players who were worse than the FAT level in the 1890's. Is that true? Why would they have done so? And if they did, how should I re-calculate replacement level for outfielders from that era?

I think of Farmer Weaver as a catcher which doesn't fit his career record. (He played less in the OF, more at C, as he proved himself a poor batter). May it be a clue?
How many players traveled, with clubs saving expenses by leaving someone under contract at home? (cblau has focused on roster size but he may know this)

When was the economic depression most severe, and did it vary much by city? (karlmagnus may know. jimd?)
You mention 1895. Is the problem for your FAT estimate significantly worse than, say, 1897?
Does the distribution of inferred-FAT players across teams vary much, being unusually concentrated in the period where your problem is most severe?
What I have in mind is that particular clubs maybe struggled unusually for a time around 1895 to field a competent team. Louisville and Washington are obvious candidates.

I don't have great hopes for these ideas, since you don't mention any 1880s problem and the NL imbalance was severe in the second half-decade --on the field, we know, and in the pocketbook, I suppose.

Dan R:
Mark S., I strongly believe that using league average as the basis for replacement level, while a widely accepted approach, is borderline ludicrous. The presence of a few superstars at a league-position (AL SS in the early 80s and late 90s come to mind) pulls up the league average, but does not make an AAAA player any better.

Right. This is one reason I have wondered, sometimes aloud, about the 21-year moving average as a measure of average batting by fielding position.
1900-1920
1901-1921
etc

That is, the 21-year moving averages of batting by fielding position
as the basis for estimating relative fielding demands at a time.

Mark S., I strongly believe that using league average as the basis for replacement level, while a widely accepted approach, is borderline ludicrous. The presence of a few superstars at a league-position (AL SS in the early 80s and late 90s come to mind) pulls up the league average, but does not make an AAAA player any better.

That's why you need to also figure out the standard deviation, which corrects that problem that you point out, IMO.

The depression was as severe as the Great Depression, but didn't last as long because economic policymakers were less loony (that is a bipartisan remark; both Cleveland and McKinley were streets ahead of either Hoover or FDR in their basic economic IQ -- far fewer mistakes, though not none.) 1895 was the bottom, though things hadn't improved much by the summer of '96. By '97, things were sharply improved, which is why Sam Leever that year decided it was safe to abandon teaching for baseball.

Going into it, '92 was distinctly less good than '91, and danger signs were apparent (e.g. Homestead strike; also probably why Harrison lost that November.) '93 was a sharp lurch downward, and '94 was really bad, close to '95. So '94-96 were an extended trough, if you like.

It did vary between cities, but I don't know in detail how. Probably the difference between viability of franchises made more difference than the intensity of the depression (i.e. Giants/Brooklyn OK anyway, Louisville and probably in those days Washington (I think Cleveland cut federal jobs, unlike Hovver/FDR) would have been in trouble.) Pittsburgh and Cleveland (the place) would have been solider than one thinks; both were boom towns around that era.

Paul, yes, I just looked at FRAA-FRAR for the outfield positions.

My worst-three-regulars rate is a moving average. The actual worst-three-regulars averages for NL corner outfielders are the following: (WBA = wins below average per 162 games)

Year WBA
1893 2.8
1894 2.4
1895 2.7
1896 2.0
1897 2.8
1898 2.6
1899 3.2
1900 1.1
1901 2.3
1902 2.1
1903 2.6
1904 3.1
1905 0.6
1906 1.1
1907 1.4
1908 1.5
1909 0.9
1910 0.6
1911 1.5
1912 0.3
1913 1.0

Clearly there's a major discontinuity between 1904 and 1905. The average from 1905-1913 is 1.0 wins below average, which is exactly the average for the rest of the 20th century as well. From 1893-1904 it is 2.5 wins below average, nearly as bad as shortstops (3.1 wins below average from 1893-1904).

I don't mention the 1880's because I haven't looked at any pre-1893 seasons.

Let's see. From 1893-99, we have 21 worst-regular outfielders. Five were from St. Louis (1896 Joe Sullivan and Klondike Douglass, 1897 Dan Lally, 1898 Dick Harley and Tommy Dowd). Four were from Cleveland (1894 Buck Ewing, 1896 Harry Blake, 1899 Dick Harley and Charlie Hemphill). Three were from Washington (1895 Bill Hassamaer, 1898 Jake Gettman, 1899 Jack O'Brien). Two were from Brooklyn (1893 Dave Foutz and 1895 George Treadway). Two were from Louisville (1894 Fred Clarke and Larry Twitchell; I dropped Farmer Weaver because he was so horrific he'd skew the average all by himself). The rest came from Cincinnati (1893 Jim Canavan), Baltimore (1893 Jim Long), New York (1895 Oyster Burns), Chicago (1897 George Decker), and Philadelphia (1897 Tommy Dowd).

Can anyone draw any conclusions from this? Or have any suggestions about what the freely available level really was back then?

John Murphy, but using SD's implies that the distribution is normal, while the distribution I have is anything but normal. Replacement players are, by definition, at the very bottom of the MLB talent distribution, which is why I think my approach of "anchoring" replacement level to the worst three regulars rather than to the league average is much more conceptually sound. However, it seems clear to me that the 1890's will require some tweaking.

I updated the credit I'm giving Charlie Keller and Gavy Cravath. For Keller it makes a small difference (I'm essentially doubling 1946 and 1947) and for Cravath it makes a big difference - using the updated numbers from his thread, no credit for 1906. I was underestimating Cravath's numbers before, he jumps onto the ballot. Keller edges closer. No minor league credit for Keller because I'm not discounting his war performance at all.

Revised Prelim:
1) Seaver
2) Rose
3) Grich
4) Bob Johnson
5) Luis Tiant
6) Norm Cash
7) Quincy Trouppe
8) Jake Beckley
9) Rusty Staub
10) Reggie Smith
11) Tony Perez - very good glove, Beckley is Perez after smoothing
12) Tommy Bridges
13) Virgil Trucks - significant relief innings pitched, if you're evaluating based on some in-season durablity criteria his relief work will mess that up.
14) Gavy Cravath - very similar to Cepeda
15) Jim Wynn (assuming Fox is elected)
16-20) Orlando Cepeda, Edd Roush, Bus Clarkson, Charlie Keller, Dutch Leonard
21-25) Bob Elliott, Jack Quinn, Luke Easter, Dick Redding, Vic Willis
26-30) Dave Bancroft, Urban Shocker, Jerry Koosman, Frank Howard, Bobby Bonds
31-36) Hilton Smith, Alejandro Oms, Johnny Evers, Dizzy Trout, Ben Taylor, Ken Singleton

Appropriate to send this back to the top of the thread list - Fox's near-miss puts him head-to-head with Grich, which is fun. Also Rose vs Grich, Rose controversy, and Perez and Cedeno to bat around.
Not bad for our last pre-Christmas pontificating....

Fox's near-miss puts him head-to-head with Grich, which is fun. Also Rose vs Grich, Rose controversy,

For me, no boycott. I don't think Rose actually threw games, or participated in conspiracies to throw games. And that's where my line is. That puts Rose on the ballot.

Rose in the #2 slot. Rose well ahead of Grich. Grich well ahead of Fox. I don't have Fox in my top 15; I will put Grich there, maybe pretty high up.

My preliminary placement of the newly eligibles:

1. Seaver -- best right-handed pitcher since WWII as of 1992.
2. Rose -- lots of baggage, but I don't feel a need to boycott. No satisfaction when he is elected, though.
3. Grich -- exceptional offense, great defense, big mistake by the BBWAA. Much satisfaction when he is elected.


18. Perez -- tough to place. Could go as high as about 10, or as low as 22. I'm still working on how to judge careers like his and Staub's.

51 & 52. Foster & Cedeno. Foster did much better than I expected. WARP thinks he was an outstanding outfielder during his peak, which was news to me. WS sees him as very good for a corner outfielder during his peak, but not quite so outstanding. Not that it matters a whole lot, since he is nowhere near election, but I hope to hear some eye-witness commentary on his defensive performance when he was in Cincinnati. I also wonder why he was a bit player with the Reds so long. Was he injured? Held back because there was no room for him? If the latter, I might rank him a bit higher.

Rose moving from OF to 3B in 1975 certainly created an opportunity for an outfielder (while at the same time sending Dan Driessen to the bench.) I'd have to look in more detail to see whether it was Foster or Griffey that was the main beneficiary.

My preliminary placements...

1. Seaver - Duh, maybe best pitcher since WWII, I like either he or Clemens
(2.) Rose - I will most likely be boycotting, but if I don't he will be here. Interesting to see if he still gets in. Part of me wants him to go in this year so that I never have to vote for him.
2 or 3. Keller
3-6 Grich - My system puts him just above Gordon and Doerr, but for some reason I think he was better than that. I have always thought that I have been underrating exceptional D. This is why I lifted Boyer up onto my ballot and why I always look at Bancroft every decade. Speaking of which...

40-60. Perez - I have to say pass. I don't like his peak much (though the time at 3rd helps) and his prime, while long, wasn't spectacular. Career candidate and I am nto much of a fan of those, generaly. Better than Staub though.

Foster and Cedeno drop out of consideration set

Blue is intersting, but won't make my top 60.

It was Foster, who they famously wanted to get into the lineup, leading Rose to pick up a new position from scratch.

On a note only interesting to me, I'd somehow skipped over Mike Marshall in my re-evaluation leading up to voting for the first time in a while (largely because I don't think he got a vote, and of course as a RP he didn't stand out on the newly eligible chart). This was a mistake...I'm not voting for him, but he's right on the edge of the top 50. His peak is outstanding for a reliever; if he had one more midrange season, he'd be evenw ith Fingers for me, 2 more midrange seasons and he'd be on my ballot.

BTW, the election will not end on the 25th. It will be the 26th,

For me, no boycott. I don't think Rose actually threw games, or participated in conspiracies to throw games. And that's where my line is. That puts Rose on the ballot.

Rose in the #2 slot. Rose well ahead of Grich. Grich well ahead of Fox. I don't have Fox in my top 15; I will put Grich there, maybe pretty high up.

My feelings mirror OCF - I will be placing him #2 on my ballot, though I need to playoff Rose vs Seaver just to make sure.

I want to triple-check these at Christmas break, but for now, what I have (finally adding Mathews). I gave Moore a "90" bullshirt dump at SS, about as high as I'd give a Negro Leaguer of that era, as everyone seemed to play some games elsewhere. Suggestions for adjustments on ANY Negro Leaguers are welcome - I do all in 5 and 10 pct increments, as we're just guessing anyway.
We just filled 3 spots with zero OF or P credit, though Carew added another half a 1B.

C-2B-3B-SS = 55.83
1B-OF-DH...= 71.12
P......... = 49.64


HOM by pct at position, thru 1991

HOM batters by percentage of games played at position (min. 10 pct at a position, otherwise it's not listed and not tallied)

C (11.79) - Cochrane 100, Dickey 100, Hartnett 98, Gibson 95, Campanella 95, Freehan 90, Bennett 88, Berra 87, Mackey 80, Bench 78, Santop 75, Ewing 47, Torre 41, Kelly 36, McVey 30, White 28, O'Rourke 11

1B (18.77) - Start 100, Gehrig 100, Mize 100, Terry 99, Brouthers 98, Sisler 97, Leonard 95, Connor 88, McCovey 88, Foxx 87, Anson 83, Greenberg 83, Suttles 70, Banks 51, Carew 50, Allen 47, Wilson 45, Killebrew 40, Stargell 40, Stovey 37, Torre 36, Charleston 35, Musial 35, McVey 31, Jennings 26, Lloyd 25, Yastrzemski 23, Heilmann 22, Ewing 19, Kelley 16, Delahanty 15, Hines 12, Lajoie 12, Mantle 11, FRobinson 11, Spalding 10, O'Rourke 10, Dihigo 10, JRobinson 10, Irvin 10

2B (14.57) - McPhee 100, Doerr 100, Childs 100, Gehringer 99, Morgan 99, E Collins 98, Gordon 98, Herman 95, Lajoie 83, Frisch 77, Hornsby 72, Grant 70, Barnes 69, JRobinson 65, Carew 47, Richardson 43, HR Johnson 25, Ward 24, Groh 20, Hill 20, Pike 18, Dihigo 15, Wright 10, Wilson 10

3B (12.27) - Baker 100, BRobinson 99, J Collins 98, Hack 98, Santo 95, Mathews 93, Boyer 90, Groh 79, Sutton 69, White 51, Beckwith 50, Wilson 40, Allen 38, Sewell 34, Killebrew 33, Torre 23, Davis 22, Frisch 20, Wallace 17, Dihigo 15, JRobinson 15, McVey 14, Richardson 13, Vaughan 11, Ott 10

SS (17.20) - Pearce 96, Boudreau 95, Reese 95, Glasscock 94, Appling 94, Cronin 92, Wells 90, Moore 90, GWright 89, Dahlen 88, Vaughan 85, Wallace 74, Jennings 70, HR Johnson 70, Lloyd 70, Wagner 68, Sewell 65, Davis 58, Banks 45, Ward 39, Beckwith 35, Barnes 28, Grant 20, Sutton 19, Hornsby 16, Dihigo 15, Irvin 10, WBrown 10

OF (51.91) - Carey 100, Clarke 100, Hamilton 100, Thompson 100, Wheat 100, Goslin 100, DiMaggio 100, Averill 100, Doby 100, Slaughter 100, TWilliams 100, Ashburn 100, Snider 100, Clemente 100, Simmons 99, Burkett 99, Cobb 99, Flick 99, Gore 99, Sheckard 99, Speaker 99, Medwick 99, Jackson 98, Stearnes 98, Keeler 97, PWaner 97, Mays 97, Kiner 96, CP Bell 95, Crawford 94, Minoso 93, Magee 91, Ott 90, Kaline 89, Mantle 88, Aaron 86, BWilliams 86, WBrown 85, Hines 82, Torriente 80, Kelley 79, Ruth 79, Heilmann 77, FRobinson 77, Irvin 75, Pike 73, Delahanty 72, Hill 70, O'Rourke 69, Rogan 65, Musial 65, Stovey 63, Yastrzemski 63, Charleston 60, Stargell 60, Kelly 47, Richardson 40, Caruthers 33, Suttles 30, Killebrew 20, Santop 20, Dihigo 20, McVey 18, Ewing 17, Greenberg 17, Allen 15, Davis 13, Wagner 13, Berra 13, McCovey 12, Spalding 11, Ward 10, White 10, JRobinson 10

DH (0.34) - Yastrzemski 13, FRobinson 11, BWilliams 10

P (49.64) - Alexander 100, Covaleski 100, Faber 100, Plank 100, Vance 100, Grove 100, Hubbell 100, Lyons 100, Newhouser 100, Feller 100, Ruffing 100, Rixey 100, Wynn 100, Spahn 100, Roberts 100, Koufax 100, W Ford 100, Drysdale 100, Bunning 100, Wilhelm 100, Marichal 100, Gibson 100, Waddell 100, Pierce 100, GPerry 100, Palmer 100, Jenkins 100, R Foster 99, Brown 99, Mathewson 99, Walsh 99, SJ Williams 99, Young 99, B Foster 99, Paige 99, W Johnson 98, McGinnity 98, WFerrell 97, Lemon 97, Keefe 96, Nichols 96, Rusie 95, RBrown 95, Griffith 95, Clarkson 94, Galvin 92, Mendez 90, Radbourn 78, Spalding 80, Caruthers 66, Rogan 35, Dihigo 25, Ward 25, Ruth 20

Caveats: Totals treat all careers as equal. A little off on players like McVey and Sutton due to changing schedule length. Guesstimates on Negro Leaguers. Hybrid P-hitters such as Ward, Ruth, Caruthers, Spalding have estimates that attempt to reflect their respective roles.

BTW, the election will not end on the 25th. It will be the 26th

In past years there has been a holiday hiatus for two weeks (and mock balloting WRT the current HOF ballot). If somthing similar were to happen this year, I would have assumed that the next election would close on Jan 8th.

No holiday hiatus?

Prelim

1/2. Rose/Seaver - not sure which order yet. Seaver ranks higher among pitchers than Rose does among hitters, so I may go with Seaver.

3. Grich
4. Perez
5. Walters
6. Trouppe
7. Staub
8. Trout
9. Wynn
10. Ryan
11. Johnson
12. Fox
13. Beckley
14. Cravath

15. It's crazy how close Singleton, Cedeno, Bonds, and Pinson rank in my system. I had Singleton ahead last ballot, but I'm going to look closer at all 4 of these guys.

Harrah and Foster both come in the top 50. Pretty good year for newbies.

Rose playing time percentages

No holiday hiatus?

Other than the HOF election in place of the HoM's, I don't believe we ever had a two week hiatus.

do we hold our own HOF election? Is the 92 ballot running in the same scheduled pace as the others? ie - does voting for 92 start next Monday?

That's a good question though, should we do a HoF ballot?

I could never vote for Ripken and Gwynn until they prove they didn't use steroids.

I asked, becasue I think it would be fun to examine. I recently placed all of this year's candidates through my system in order to see if my gut picks are above my in/out line. I thought a forum here would be enlightening, if not ultimately conducive to the HOM

do we hold our own HOF election? Is the 92 ballot running in the same scheduled pace as the others? ie - does voting for 92 start next Monday?

The HOF discussion thread will be posted on the 26th and the election will end on January 8.

Voting for '92 starts next Monday.

Anyone else think the upcoming election of Rose is the perfect time to get additional publicity for the Hall of Merit? We could highlight the differences and the similarities and give our short list of Hall of Fame mistakes (no ballot support HoFers).

Prelim


PHoM: Pete Rose, Tom Seaver, Tony Perez


Ballot:


1. Pete Rose
2. Tom Seaver
3. Tony Perez
4. Bobby Grich
5. Rusty Staub
6. Jake Beckley
7. George Van Haltren
8. Nellie Fox
9. Jimmy Wynn
10. Edd Roush
11. Quincy Trouppe
12. Tommy Leach
13. Lou Brock
14. Mickey Welch
15. Orlando Cepeda

16-20. Duffy, Bonds, Cash, R.Smith, Singleton
21-25. Ryan, Johnson, Browning, Cedeno, Rice
26-30. Grimes, Redding, Streeter, Willis, McCormick

Rusty - 1992 is an elect 2 year.

not much luv for pitchers, eh Rusty?

Bobby Bonds / Reggie Smith / Jimmy Wynn

common items:
excellent RFers / decent CFers
mid 60s thru 1980
low batting avg, good at everything else

uncommon:
Wynn on verge of being elected, other two with small support.

career comparisons using uberstats, dropping off any poor seasons at beginning or end of careers
player Years WS WARP OWP Gms
Wynn. 65-76 290 90.7 .652 1717
Bonds 68-79 295 90.9 .641 1718
Smith. 67-82 325 91.6 .654 1983
Smith*67-80 309 87.1 .654 1836 (2 more end yrs chopped off)

Bonds - Wynn, could you find two more comparable peas in a pod? Wow, they are dead ringers

best year comparisons, using RCAA (since same era, run distortiosn and league qual distortions should be small)

Wynn. 61 45 45 41 37 35 38 28 05 04
Bonds 49 43 41 38 34 30 24 23 21 17
Smith 65 46 39 35 30 28 27 26 23 20

Reggie = very close Bonds except Smith had one monster year.
Bonds vs Wynn -- Jimmy trades three good Bobby years for 2 blah seasons and one killer one.
Wynn vs Smith -- durn close, ain't it?

RCAA does not measure defense, but these guys are pretty close with the glove. I think RCAA underrates SB a bit, since SB occur in more close-game situations, whereas RCAA measures Runs and assumes they are all equal. This would give a small edge to Bonds.

How can you tell these guys apart? I have them as 13-15-18 on next week's provisional ballot. What makes Wynn the one to elect? The Big Year? Reggie has that. Reggie has him on career as well, and it sure is tough to look at winning teams following Mr Smith around and not at least consider a teeny bonus there. Bobby beats Wynn by microscopic amounts in both WARP and Win Shares.

All in all, given how many OFers we've already honored in the HoM, I'd like for guys high on my ballot to actually, you know, stand out against their brethren. These three are all very good players, and I would not be firmly against electing any or all of them. But right now... sure seems to me like I can find more than a few others that can cut in front of these guys in line.

error in RCAA table above
Wynn. 61 45 45 41 37 35 31 28 05 04

I have them Reggie2, Wynn, Bonds. I think Wynn is getting too much credit for playing CF.

I am definitely driving the Jimmy Wynn is overrated bus.

WS (top 7)

Wynn: 36,32,32,31,28,28,27 (305)
Smith: 29,29,26,25,25,25,24 (325)

Wynn has 10 years over 15 WS, Smith has twelve years

WARP1

Wynn: 11.8,11.4,10.8,10.7, 9.2, 8.6, 8.3 (99.7)
Smith:10.2, 8.9, 7.2, 7.1, 7.0, 7.0, 6.7 (98.9)

Smith has 11 seaons over 5 WARP, Wynn has 10.

Now, I know that I will incur the wrath of Howie here, but these two do not show two players who are anywhere close. Smith's best season in WARP would be Wynn's 5, his second best would be Wynn's 6th. Smith's best season in WS would be Wynn's 5th and 6th, his second best would be Wynn's 8th. Their career numbers are similar and Smith has an extra seasons or two above average. Does that overcome Wynn's very distinct peak advantage?

Using only offense, BPro says that Wynn has five seasons better than Smith's best (70,56,51,49,48 vs. 47) as well as another season better than Smith's second best (46 v. 45). So it looks like the difference isn't in fielding so much as a difference of opinion between BRAA and RCAA. How does RCAA have Smith's best seasons as worth 19 runs above average than BPro?

Also, Wynn did play 1182 games in CF as opposed to Smith's 808. Equal offense is worthmroe in CF than RF (where Smith actually played more games, 879 to 808).

Ditto.

Elect 2? Oops, Tony will have to wait then.


No love for pitchers? I have a pitcher at #2. The only reason he isn't at #1 is that he became eligible in the same year that one of the greatest players ever joined the ballot.


And since it is an Elect 2 year, the only possibility of a debate about who gets in is if people use that (imo) silly one year boycott clause.

Prelim ballot

PHOM - Seaver and Rose

1. Seaver - Meaningless otherwise, but in my discussion on the Seaver thread, I was missing that his '77 was divided between two teams, so he gets some more peak/prime credit.
2. Rose - Seaver clearly better on the pure untouched record. Rose enough better than Grich that it's not worth deciding whether to dock him some for his managerial insertions of himself into the lineup, as it wouldn't matter. Too peakless to be inner circle like Seaver.
3. Grich - Big gap before and after him. Great glove, very good bad, consistent as all hell. Completely embarassing HOF whiff.

4. Dunlap
5. Elliot
6. Johnson
7. Rizzuto
8. Trouppe
9. Maranville
10. Walters
11. Redding
12. Shocker
13. Grimes
14. Bancroft
15. Monroe

20. Perez - Likely to PHOM one day, doubtful on making my ballot, as no one from 4-15 other than maybe Trouppe is likely to be elected. Like his career, if he'd stuck at 3B for a few more years (and it feels like he should have, he was good there), he'd be an easy induction. Better than Staub, as he was familiar with the concept of a glove, and was a slightly better hitter (his rate is held down by many years he was finding his way into lineups he shouldn't have been in).

39. Foster - Excellent prime, just nothing outside of it. Too bad, especially as I'm a Met fan, and the end of his prime coincided with his arrival in Shea.

52. Blue
56. Harrah
57. Cedeno

Nearly identical in my system to Staub (53); Foster's just a touch ahead. Blue had some nice years, but needed 2-3 more. Harrah demonstrates the Doyle principle that a nice bat at a glove position isn't enough when you stink at that position. Cedeno is Foster minus one extra prime season.

Using only offense, BPro says that Wynn has five seasons better than Smith's best (70,56,51,49,48 vs. 47) .... How does RCAA have Smith's best seasons as worth 19 runs above average than BPro?

BPro has Smith's best at 59, not 47.

I am definitely driving the Jimmy Wynn is overrated bus.

Looks like you're starting to weave a little bit there, Ryan. ;-)

The HOF discussion thread will be posted on the 26th and the election will end on January 8.

Voting for '92 starts next Monday.

Holiday hiatus in prior years:

1916: 12/22/03
1917: 01/12/04 (one week)

1941: 12/20/04
1942: 01/10/05 (one week)

1967: 12/26/05
1968: 01/23/06 (two weeks)

1992: 12/26/06
1993: 01/08/07 (no weeks)
1994: 01/22/07

I take it that you'd prefer not to have one, which is OK by me.

>one of the greatest players ever

Whoa.

You are right Tom. Still Smith would only have one of the top six seasons between, still no question in my mind who was better.

About Wynn:

If a guy isn't playing a specialty defensive position (catcher, ss, 2b, pre-war 3b or a true center fielder), shouldn't he have some black/grey ink? An outfielder who really can't play centerfield and who can only occasionally crack the top 10 in BA, RBI, HR, Runs, OBA, SLG, 2b, SB, Hits or 3b (and never finishes first in any category) doesn't sound like a candidate for enshrinement. He regularly got grey ink in the category of walks. (My black/grey ink analysis is park-adjusted).

When you tear away all the micro-analysis we do for these candidates, does anyone's gut have Wynn as one of the best players who ever lived? Do the people who saw him play, or played with him, think he is one of the best who ever lived? I recognize he was a good player -- I'm not denying that -- but I have NEVER heard anyone, anywhere (other than perhaps Joe Morgan during a slow Sunday night telecast, who also thinks Cesar Geronimo ought to be in) suggest that Wynn is in that category.

Even with a short career, I bet a lot more people in Keller's era turned to each other and said "That guy is awesome."

but I have NEVER heard anyone, anywhere ... suggest that Wynn is in that category

But don't forget that people's perceptions are colored by how the candidates were portrayed during their careers. This comment implies rating people by some kind of "buzz" factor. Steve Garvey has a much higher buzz factor than Jimmie Wynn - and a lot of that is batting average. Wynn's power was noticed when he was active, and so was his very low batting average, as was the fact that his RBI were a little on the low side for someone with that many HR. But I never heard, "Omigod, he just drew 148 walks in a season. Do you know how rare that is?"

I'm not denying that his candidacy is flawed. Every time we touch that backlog (and it will be a few years until we're back there again), we're dealing with balancing among many flawed candidates. My own Wynn comment says something like "a HoMer shouln't have a year like Wynn's 1971 right in the middle of his prime." But I am rating him as the best of the Smith/Bond/Singleton grouping and I am voting for him, not because of 1971 but becuase his good years were so very good.