OK: I know the outs are too high, I suspect because the "27" isn't really 27. But it's consistent with the way the league averages are computed. The handbook uses different RC formulas for different eras, and I think something like that is needed for cross-era comparisons. Accuracy as compared to other run estimators? I don't know, I haven't really investigated that. I've stayed with the same source to maintain internal consistency. (I will start to have a problem with players whose career extend beyond 1999.) And because it's convenient for me.

It's funny - there was a time when I felt like I was one of the guys arguing against George Van Haltren, and now he may be in my top 5. I have Roush in my top 15 as well, but definitely behind GVH. But I wanted to remind everybody that Van Haltren gets an excessive boost in his WS for his pitching, which was no more than league average at best.

Wayyyyyy back in 19 and 30 summin' when I joined this here project, we used to walk uphill both ways to vote, in the snow, without boots, over unpaved roads littered with tetanus-covered rusty nails (and priskes). And when I got to the votin' booth, you better believe George Van Haltren was at the top of my ballot.

But now I just think he's a little below the in/out line.

We're ships crossing in the night.

The handbook uses different RC formulas for different eras, and I think something like that is needed for cross-era comparisons.

Disclaimer: I'm not a runs estimator expert, I've just done a little hunting on the subject, and I'll see if I can summarize what I've learned.

I've been intending to investigate base runs at greater length. I've read a little, and I think it supposedly mitigates the problem of different eras through minor adjustments to weighted variables. Tango has a site where you can plug the variables in and figure out the weights for any run environment. Supposedly it also is very reliable even into the extreme run environs where other estimators break down.

As you say, OCF, RC adjusts by having different, empirically derived (trial/error?) formulae. As Diz noted, its most recent iteration includes an adjustment that removes the player from his own team context, which supposedly helps it a little for more extreme Ruth/Bonds type players.

Dan mentioned XR, and I think it's supposed to be very good for post-War players, but isn't supposed to be used for pre-War players, IIRC. I think the reason is that it was derived by multivariate regression, and many of the stats it regressed on weren't kept or noted in the pre war era. Actually, you know, it might be since like 1955 or so not just post-war. Can't quite recollect now. Jim Furtado runs the place, maybe he could clarify if he's happening into our neck of the woods!

I don't know about EQR. Does anyone?

The one-line summary? Pick the one that works best and most accurately for what you're doing.

Otherwise, why does it matter what Lynn's raw OPS, or OPS+, or VORP was? OPS+ and VORP (which is based on the MlVR formula, which uses BA/OBP/SLG) are simply not as accurate measures of offensive value as the best run estimators available to us, as measured by RMSE or any other such check. If the best tools we have say those years were offensively equal, I don't see why we'd cast unnecessary doubt on them by comparing them to inferior ones--it's like saying there must be something wrong with UZR when it disagrees with fielding percentage (albeit much less extreme). I don't mean to sound condescending, nor to toot my own horn; I didn't come up with the run estimators. But unless you're saying that what they're telling us is wrong and OPS or OPS+ or VORP are right--and if so, please do tell me how!--I'm not sure what the point in bringing them into the discussion is.

I'm hoping Dan will comment on the math I posted in #98 when he sees it--but the reason for bringing them up is as sort of a "smell test." OPS and OPS+ and VORP are flawed measures--but they're not *that* flawed. And the difference between Wynn and Lynn on those measures are enormous, well outside a reasonable error margin. If you have a system that is telling you that Jimmy Wynn was just as valuable a hitter in 1965 as Fred Lynn was in 1979, then that system needs to be rethought.

Actually, I did find one small glitch in those numbers in #98--Lynn should be at +59.7 runs, not +60.2. This, of course, changes the point not at all.

HOM by pct at position, thru 1995

Big year for 3B, as Schmidt-Evans-Trouppe add 1.66 to their score.

If 65 pct is your cutoff, then we have elected 13 Cs, 13 1Bs, 15 2Bs, 10 3Bs, 17 SSs, 52 OFs, 54 Ps.
If 50 pct is your cutoff, then we have elected 13 Cs, 15 1Bs, 15 2Bs, 13 3Bs, 18 SSs, 56 OFs, 54 Ps.

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

C (13.21) - Cochrane 100, Dickey 100, Hartnett 98, JGibson 95, Campanella 95, Freehan 90, Bennett 88, Berra 87, Mackey 80, Bench 78, TSimmons 77, Santop 75, Trouppe 65, Ewing 47, Torre 41, Kelly 36, McVey 30, White 28, O'Rourke 11

1B (19.36) - 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, DaEvans 32, McVey 31, Rose 27, 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 (15.61) - McPhee 100, Doerr 100, Childs 100, Gehringer 99, Morgan 99, E Collins 98, Gordon 98, Herman 95, Grich 86, 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, Rose 18, Dihigo 15, Wright 10, Wilson 10

3B (14.11) - Baker 100, BRobinson 99, J Collins 98, Hack 98, Santo 95, Mathews 93, Schmidt 92, Boyer 90, Groh 79, Sutton 69, DaEvans 54, White 51, Beckwith 50, Wilson 40, Allen 38, Sewell 34, Killebrew 33, Torre 23, Davis 22, Frisch 20, Trouppe 20, Rose 18, 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 (53.21) - 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, SJJackson 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, RJackson 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, Rose 38, Caruthers 33, Suttles 30, Killebrew 20, Santop 20, Dihigo 20, McVey 18, Ewing 17, Greenberg 17, Allen 15, Trouppe 15, Davis 13, Wagner 13, Berra 13, McCovey 12, Spalding 11, Ward 10, White 10, JRobinson 10

DH (0.79) - RJackson 23, Yastrzemski 13, TSimmons 12, FRobinson 11, BWilliams 10, DaEvans 10

P (53.64) - 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, Seaver 100, Carlton 100, Niekro 100, Sutton 100, R Foster 99, MBrown 99, Mathewson 99, Walsh 99, SJ Williams 99, Young 99, B Foster 99, Paige 99, WJohnson 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.

Dr. Chaleeko--you are indeed entirely correct. XR is deadly accurate for post-integration seasons, and not very good at all beforehand. That's why I use BaseRuns for 1946 and before. BaseRuns doesn't exactly make "minor adjustments" by era--it's just extraordinarily flexible, because of its design--runs scored equals non-HR baserunners times % of non-HR baserunners who score plus HR. By allowing the second term--the % of non-HR baserunners who score--to float for each league-season so that predicted league runs match actual league runs, it gives me *extremely* good results predicting team runs scored all the way back to 1893. (It doesn't work for the 1870's and 80's though, I tried--too many reached on errors). I could just use BaseRuns all the way up to the present, but it's much more cumbersome to use than XR and no more accurate for post-1947 seasons.

EQR gets no press, but it's terrific--it consistently matches my XR and BaseRuns estimates to within a run or two. Just take a look at any Baseball Prospectus DT team page, and compare the UEQR (the run estimation) to actual runs scored--it's always right there, even in the 1890's (with the exception of those funky Boston teams that always outperformed their component stats). In terms of accuracy, it's probably just as good as BaseRuns for the 1893-1910 period that XR totally whiffs on, but I prefer BaseRuns' conceptual elegance. I doubt EQR can really handle the 1870's and 1880's, but you never know.

OCF--I'm not aware of any RC formula that actually matches EQR, BaseRuns, or XR for accuracy, but the most advanced RC formulas (not sure if Stats uses them) are probably insignificantly worse. That said, try adding up the RC from your Stats book for any one team from the 1890's. If the sum isn't within a few percent of the team runs scored total, your RCAA numbers will NOT be reliable for that era.

Does Stats RC count sac flies, net double plays, or non-SB baserunning runs? If not, then that 134 in 1979 looks suspiciously high to me--I have him at 127.5 (non-park-adjusted) without those factors, and I'm quite confident in that estimate given that EQR, BaseRuns, and XR all agree to within a fraction of a run. What are your RC estimates for the 1979 Red Sox team total and 1979 AL league total? Are you sure they aren't overshooting?

DCW3--I COMPLETELY disagree with the following statement: "If you have a system that is telling you that Jimmy Wynn was just as valuable a hitter in 1965 as Fred Lynn was in 1979, then that system needs to be rethought." Not necessarily! It means that one system or the other is wrong, and you should investigate the cause/source of the discrepancy. More often than not, when you add up all the little factors that OPS+ or VORP don't catch (double plays, stolen bases, sac flies, OBP-heaviness etc.), you can account for all of the gap. Sometimes you just have a typo in your formula, which is why smell tests are useful--they highlight unexpected results, and force you to figure out what caused them. But most of the time, you're able to explain why you've gotten a different result, and therefore why the advanced metric is right and the simple one wrong.

Now to address your math in #98. The Lynn RCAA figure is exactly right. I have him at 60 runs/5.9 wins above average. (What Pythagorean theorem are you using to get 60 runs as only 5.67 wins in the 1979 AL?) 104.4 is right for Wynn as well if you don't count his NetDP and sac flies.

But I *definitely* don't agree with using those baseball-reference park-adjusted league average numbers as the basis for comparison if they don't include pitchers. If you subtract pitchers from the NL, the other hitters in the lineup FAR outproduce the American League average, by a full 7%--I can show you the numbers. If that's what baseball-reference's OPS+ numbers are based on, then they will *substantially* overrate post-1973 AL hitters in relation to NL and pre-1973 AL ones by 7%, which is a damn big difference. I didn't know that (I don't use OPS+, since it's simply much less accurate than using run estimators). But it's something that everyone who does use OPS+ should take into account. (Conversely, BP's WARP1 do NOT account for the DH, penalizing post-1973 AL players, and people who use those should correct for that).

It seems quite clear to me that the proper way to account for the DH is to compare both the players in question and replacement players to the league average. You've got two guys, both of whom produce 100 runs in the same number of outs, one in the AL, one in the NL. Let's say the average AL player produces 70 runs in those outs, and the average NL player (including pitchers) 65. Then indeed, the AL player is only 30 runs above average, and the NL player 35. But when we compare them to replacement, the difference vanishes. A replacement player will produce 54 runs in that number of outs, which is 16 below the AL average and just 11 below the NL average. So, 30 + 16 = 46, as does 35 + 11. When you simply subtract a replacement player's performance from the player in question's performance, the entire DH issue magically goes away. Either way, it's 100 minus 54.

I have determined empirically (again, I can show you the math if you're interested) that a replacement non-DH player in the AL is 0.6 wins per 685 plate appearances worse than a replacement non-DH player in the NL, when compared to league averages. (This is a big gap, and I mistakenly used my old 0.4 figure in my initial Wynn vs. Lynn comparison. However, it is *not* as big as the gap suggested by baseball-reference's park-adjusted league average numbers, which I would like to repeat are simply W-r-o-n-g.). So, again, I would do the players this way, without counting fielding (although I would use floating rather than fixed replacement levels, that's another issue entirely):

Murcer: 124 park-adjusted XR in 623 PA and 357 outs. Average 1971 AL team: 623 runs in 4145 outs for .150 R/Out. Average 1971 AL team with Murcer: 124 + ((4145-357)*.15) = 693 runs. 693 RS and 623 RA in 162 games is 88.9 wins. Replacement pre-1973 AL CF: -1.4 wins per 676 PA. 88.9 - 81 + (1.4*(623/676)) = 9.2 WARP.

Lynn: 125 park-adjusted XR in 622 PA and 358 outs. Average 1979 AL team: 752 runs in 4116 outs for .183 R/Out. Average 1979 AL team with Lynn: 125 + ((4116-358)*.183) = 812 runs. 812 RS and 752 RA in 162 games is 86.9 wins. Replacement post-1973 AL CF: -2.0 wins per 676 PA. 86.9 - 81 + (2.0*(622/676)) = 7.7 WARP.

Cedeño: 113 park-adjusted XR in 624 PA and 407 outs. Average 1972 NL team: 605 runs in 4040 outs for .150 R/Out. Average 1972 NL team with Cedeño: 113 + ((4040-407)*.15) = 657 runs. 657 RS and 605 RA in 156 games (there was a brief strike) is a .537 winning percentage, or 87.0 wins per 162 games. Replacement NL CF: -1.4 straight-line-adjusted wins per 648 PA. 87.0 - 81 + (1.4*(624/648)) = 7.3 WARP.

Wynn: 117 park-adjusted XR in 658 PA and 411 outs. Average 1965 NL team: 656 runs in 4238 outs for .155 R/Out. Average 1965 NL team with Wynn: 117 + ((4238-411)*.155) = 709 runs. 709 RS and 656 RA in 162 games is 86.8 wins. Replacement NL CF: -1.4 wins per 676 PA. 86.8 - 81 + (1.4*(658/676)) = 7.2 WARP.

So Lynn's '79 was worth 0.5 wins more than Wynn's '65, on pure offensive value. The "missing" 0.6 wins--DCW3, you calculated the gap at 1.1--are primarily due (I imagine) to baseball-reference's overstating the league average for the 1965 NL by subtracting pitchers (which means the NL average they use will on average be 7% higher than the AL average they use, given equivalent levels of run scoring).

"BP's WARP1 do NOT account for the DH, penalizing post-1973 AL players, and people who use those should correct for that."

How? Is there a certain number of BRAR and BRAA I need to add per plate appearance?

If 65 pct is your cutoff, then we have elected 13 Cs, 13 1Bs, 15 2Bs, 10 3Bs, 17 SSs, 52 OFs, 54 Ps.
If 50 pct is your cutoff, then we have elected 13 Cs, 15 1Bs, 15 2Bs, 13 3Bs, 18 SSs, 56 OFs, 54 Ps.

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

Has anyone calculated HOM batters by % of Win Shares (or WARP, etc.) at position? It seems like this might provide a better analysis of the electorate's consideration of each spot on the field. Ernie Banks jumps out at me as someone who is categorized at a different position from the one in which he established most of his value, and Buck Ewing doesn't even make the 50% cutoff for catcher even though the vast majority of his production came there.

Also, Dan R, where can we find James Click's EqBR numbers?

Hmm. I'm going to run some more numbers when I get home from work tonight, but I have one more question, Dan: you're saying that using OPS+ will overrate post-1973 AL players in relation to NL players, or vice versa. Now, Wynn has some offensive value that doesn't show up in his OPS+, with his stolen bases, but he was still only 8th in the NL in OPS+ in 1965 (behind a lot of Hall of Famers, true) and was not in the top ten in OBP, SLG or raw OPS, for what little that's worth. Now, not only was Lynn's raw OPS in 1979 the highest in the AL between 1962 and 1993, but his OPS+ was the fifth-highest in the AL in the 1970s--after the league switched to the DH, it was the second-highest from 1973 to the end of the decade, and the third-highest from 1973 to 1984. Could the entire league in that era really have been that spectacularly weak?

DL from MN--Most definitely! I have it as 0.6 wins per 675 PA, so that would be 0.6*9 = 5.4 BRAA/BRAR per 675 PA. BP is quite explicit in their glossary in saying themselves that they don't adjust for the DH and that that disadvantages post-1973 AL players.

Dandy Little Glove Man, I've got the spreadsheet, I can send it to whoever wants it. Rod Carew and Robin Yount are the best; Mark McGwire cost the Cardinals nearly ten runs in 1999! Hard to do.

DCW3--You are my new best friend! It has nothing to do with league strength or weakness, and everything to do with the standard deviation. As I've posted 8 zillion times, the 1972-1992 period had the lowest standard deviations in MLB history by a pretty large margin, due to low run scoring, only one expansion, and integration. That is precisely the league adjustment I use to convert WARP1 to WARP2, and it's why my ballot is full of 1970's and 80's players. Although Lynn's performance in 1979 was only 0.5 raw wins more valuable than Wynn's in 1965, Lynn's league was most definitely "harder to dominate"--he was more standard deviations away from the mean. Thus the gap between him and Wynn would be bigger in WARP2 than in WARP1. Take a look at the "StDevs and Rep Levels" spreadsheet available at www.mindspring.com/~cooberp/WARP.zip to see the evolution of standard deviations in the NL from 1893-2005.

You are my new best friend!

Well! In that case, do you mind answering a couple more questions (if your spreadsheet is set up to do these kinds of sorts)?

1) Where does Wynn's 1965 rank among the top offensive seasons in the NL that year?
2) Where does Lynn's 1979 rank among the top offensive seasons of the 1970s AL--or, say, the AL between 1973 and 1984?

> 5.4 BRAA/BRAR per 675 PA

That's huge. Lots of boosts all across my ballot if that's true.

If 1965 was not Wynn's best offensive season why are we comparing Lynn's 1979 with Wynn's 1965?

Thus the gap between him and Wynn would be bigger in WARP2 than in WARP1.

Actually, BPro lists them with identical WARP2 numbers--11.0.

Using fixed rather than floating replacement levels for now:

1. Wynn ranks second in the 1965 NL with 6.7 WARP2. Mays is well ahead at 8.1. If you take fielding into account, Santo and Aaron pass Wynn as well.

2. 10th. Here's a chart:

SFrac: Percentage of the league average plate appearances per lineup slot
BWAA: Batting Wins Above Average
BRWAA: Baserunning Wins Above Average
Rep: Difference between league average and replacement performance in the player's plate appearances
WARP1: Wins above replacement player
LgAdj: Ratio of the 2005 NL standard deviation to the projected standard deviation of the league-season in question
WARP2: Standard deviation-adjusted wins above replacement player

1981 is straight-line adjusted. BWAA + BRWAA + Rep will not always equal WARP1 because Rep is calculated using the 2005 NL standard deviation.

Player       Year Pos  SFrac BWAA BRWAA  Rep WARP1 LgAdj WARP2
Robin Yount  1982   6   1.02  5.8   0.6  3.7  10.1  .984   9.9
Robin Yount  1983   6    .97  4.8   0.9  3.5   9.2  .990   9.2
Dwight Evans 1981   9   1.12  6.3   0.8  1.8   9.0  .938   8.5
Rod Carew    1977   3   1.02  7.4   0.7  0.9   9.0  .935   8.4
George Brett 1980   5    .75  6.7   0.3  1.5   8.6  .967   8.3
Cal Ripken   1984   6   1.05  4.1   0.1  3.8   8.0 1.014   8.1
Toby Harrah  1975   6    .92  4.4   0.0  3.3   7.8  .976   7.7
Rod Carew    1975   4    .90  5.0   0.6  2.0   7.7  .976   7.5
Cal Ripken   1983   6   1.06  3.8  -0.1  3.8   7.6  .990   7.5
Fred Lynn    1979   8    .92  5.8   0.1  1.8   7.9  .942   7.4

A few notes:

1. There is no greater commodity in baseball than a shortstop who can hit. (again, see Nate Silver's piece on Freely Available Talent which explains why SS replacement level is so much lower than that of any other position).
2. Yount's superlative non-SB baserunning is critical to his placement. Contemporary subjective accounts support James Click's assessment of his baserunning.
3. If Evans' 1981 looks high, that's because it is. This is due to three key factors: a. He had an exorbitant number of plate appearances: 12% more than the league average PA per lineup spot. He played in every game and was the #2 hitter for a team that led the league in OBP by 10 points. b. It was a strike season, and I'm not regressing it. c. He was, according to James Click, an absolutely outstanding baserunner--5.7 EqBR in just 108 games is one of the highest rates I've seen.
4. On a rate basis, Brett's 1980 is about 50% better than any other season in the sample.
5. Lynn's 1979 AL was the highest-scoring league in the time period you specified, and it was just two years after an expansion. That means it was one of the "easiest-to-dominate" years of the period (although not as easy as Wynn's 1965 NL).
6. Once you add on defense, the Ripken seasons jump to the top.
7. Carew's 1975 would rank higher if I were using a floating replacement level (as I do in my actual WARP)--2B replacement level was much lower in the 1970s than the 1985-2005 average.

Oh, sorry, I guess I was unclear--I was wondering where those season ranks solely in terms of offensive performance, without adjusting for position.

DL from MN--make those adjustments. From BP's own Web site (http://www.baseballprospectus.com/glossary/index.php?mode=viewstat&stat=7):

"Statistics that have been adjusted for a single season are the best stats to use when you are only interested in that one season. In these, adjustments have been made to account for the home park and for the offensive level of the league as a whole. Hitters have an adjustment for not having to face their own team's pitchers; pitchers have a similar adjustment for not having to face their own hitters. Hitters in the AL since 1973 have a disadvantage in these statistics, since the league average is artificially inflated by the use of the DH and no adjustment is made for that."

Again, my research (which I'm happy to show to anyone who's interested) shows that a replacement player will be 0.6 wins closer to average per 675 plate appearances in a non-DH league than in a DH one.

DCW3--I meant *my* WARP2, the standard deviation-adjusted version, rather than BP's WARP2, the league difficulty-adjusted (but not season length-adjusted) one. I know, I really need to come up with a new name for my system so as to avoid confusion with BP.

Arghhhh my preformatted text didn't work! I HATE it when that happens. Well, for anyone who's interested, I'll post it again and hope it formats properly.

Player       Year Pos SFrac BWAA BRWAA Rep WARP1 LgAdj WARP2
Robin Yount  1982 6    1.02  5.8   0.6 3.7  10.1  .984   9.9
Robin Yount  1983 6     .97  4.8   0.9 3.5   9.2  .990   9.2
Dwight Evans 1981 9    1.12  6.3   0.8 1.8   9.0  .938   8.5
Rod Carew    1977 3    1.02  7.4   0.7 0.9   9.0  .935   8.4
George Brett 1980 5     .75  6.7   0.3 1.5   8.6  .967   8.3
Cal Ripken   1984 6    1.05  4.1   0.1 3.8   8.0 1.014   8.1
Toby Harrah  1975 6     .92  4.4   0.0 3.3   7.8  .976   7.7
Rod Carew    1975 4     .90  5.0   0.6 2.0   7.7  .976   7.5
Cal Ripken   1983 6    1.06  3.8  -0.1 3.8   7.6  .990   7.5
Fred Lynn    1979 8     .92  5.8   0.1 1.8   7.9  .942   7.4

DCW3--I can tell you where it ranks in relation to average, but not to replacement, since there's no such thing as a positionless replacement player! Comparisons to average are BAD and MISLEADING as far as I am concerned, because they will overstate the value of lower playing time higher rate players and understate the value of higher playing time lower rate players. But if you insist....he's 10th again. OWAA is total offensive wins above league average. W2AA is OWAA adjusted for standard deviation.

Player         Year Pos SFrac BWAA BRWAA OWAA LgAdj W2AA
Rod Carew      1977   3  1.02  7.4   0.7  8.1  .935  7.5
George Brett   1980   5   .75  6.7   0.3  7.0  .967  6.8
Dwight Evans   1981   9  1.12  6.3   0.8  7.1  .938  6.7
Robin Yount    1982   6  1.02  5.8   0.6  6.4  .984  6.3
John Mayberry  1975   3  1.01  6.0   0.0  6.0  .976  5.8
Reggie Jackson 1974   9   .89  5.7   0.1  5.8  .994  5.7
Eddie Murray   1984   3  1.03  5.5   0.1  5.6 1.014  5.7
Robin Yount    1983   6   .97  4.8   0.9  5.7  .990  5.7
Eddie Murray   1983   3  1.00  5.4   0.2  5.6  .990  5.6
Fred Lynn      1979   8   .92  5.8   0.1  5.9  .942  5.6

Comments on this chart:

1. I REPEAT THESE ARE BAD AND MISLEADING NUMBERS THEY ARE COMPARED TO AVERAGE NOT REPLACEMENT USE THEM AT YOUR PERIL.
2. The same caveats about Evans (gazillions of plate appearances, no regression for strike season, exceedingly high non-SB baserunning runs) still apply.
3. Carew gets dinged badly (6.5%) for having his big season in an expansion year. Nonetheless, it was such a great season that he still comes out on top.
4. Lynn would be 6th in raw offensive wins above average, but the high run environment and proximity of the season to expansion lead his season to be regressed more heavily (5.8%) than others.

I can't get these charts to format for the life of me. Maybe using CAPS for the tags?

Position-adjusted offensive wins above replacement, AL 1973-1984

Player       Year Pos SFrac BWAA BRWAA Rep WARP1  LgAdj WARP2
Robin Yount  1982   6  1.02  5.8   0.6 3.7  10.1  .984    9.9
Robin Yount  1983   6   .97  4.8   0.9 3.5   9.2  .990    9.2
Dwight Evans 1981   9  1.12  6.3   0.8 1.8   9.0  .938    8.5
Rod Carew    1977   3  1.02  7.4   0.7 0.9   9.0  .935    8.4
George Brett 1980   5   .75  6.7   0.3 1.5   8.6  .967    8.3
Cal Ripken   1984   6  1.05  4.1   0.1 3.8   8.0 1.014    8.1
Toby Harrah  1975   6   .92  4.4   0.0 3.3   7.8  .976    7.7
Rod Carew    1975   4   .90  5.0   0.6 2.0   7.7  .976    7.5
Cal Ripken   1983   6  1.06  3.8  -0.1 3.8   7.6  .990    7.5
Fred Lynn    1979   8   .92  5.8   0.1 1.8   7.9  .942    7.4

Non-Position Adjusted Offensive Wins Above League Average, AL 1973-1984

Player         Year Pos SFrac BWAA BRWAA OWAA LgAdj W2AA
Rod Carew      1977   3  1.02  7.4   0.7  8.1  .935  7.5
George Brett   1980   5   .75  6.7   0.3  7.0  .967  6.8
Dwight Evans   1981   9  1.12  6.3   0.8  7.1  .938  6.7
Robin Yount    1982   6  1.02  5.8   0.6  6.4  .984  6.3
John Mayberry  1975   3  1.01  6.0   0.0  6.0  .976  5.8
Reggie Jackson 1974   9   .89  5.7   0.1  5.8  .994  5.7
Eddie Murray   1984   3  1.03  5.5   0.1  5.6 1.014  5.7
Robin Yount    1983   6   .97  4.8   0.9  5.7  .990  5.7
Eddie Murray   1983   3  1.00  5.4   0.2  5.6  .990  5.6
Fred Lynn      1979   8   .92  5.8   0.1  5.9  .942  5.6

Hope this works...

I give up, sorry. I've posted zillions of charts on BTF but it's just not liking these ones. Has the format for inputting preformatted text changed? Would anyone care to enlighten me?

Dan, don't use caps, do use these: []

bracket pre bracket

bracket /pre bracket

Great, let's try that. Thanks, Eric. I'll repost my comments too.

Position-adjusted offensive wins above replacement, AL 1973-1984




Non-Position Adjusted Offensive Wins Above League Average, AL 1973-1984



Reposted comments:

1. I REPEAT THE NUMBERS IN THE SECOND CHART ARE BAD AND MISLEADING THEY ARE COMPARED TO AVERAGE NOT REPLACEMENT AND NOT ADJUSTED FOR POSITION USE THEM AT YOUR PERIL.
2. The same caveats about Evans (gazillions of plate appearances, no regression for strike season, exceedingly high non-SB baserunning runs) still apply.
3. Carew gets dinged badly (6.5%) for having his big season in an expansion year. Nonetheless, it was such a great year that he still comes out on top.
4. Lynn would be 6th in raw offensive wins above average, but the high run environment and proximity of the 1979 season to expansion lead his year to be regressed more heavily (5.8%) than others.

> I know, I really need to come up with a new name for my system so as to avoid confusion
> with BP.

sWARP - little s is common notation for standard deviation

Rick Reuschel was a good pitcher for many years, but he will not ever appear on my ballot.

But I *definitely* don't agree with using those baseball-reference park-adjusted league average numbers as the basis for comparison if they don't include pitchers. If you subtract pitchers from the NL, the other hitters in the lineup FAR outproduce the American League average, by a full 7%--I can show you the numbers. If that's what baseball-reference's OPS+ numbers are based on, then they will *substantially* overrate post-1973 AL hitters in relation to NL and pre-1973 AL ones by 7%, which is a damn big difference. I didn't know that (I don't use OPS+, since it's simply much less accurate than using run estimators). But it's something that everyone who does use OPS+ should take into account. (Conversely, BP's WARP1 do NOT account for the DH, penalizing post-1973 AL players, and people who use those should correct for that).

Okay, I know this is becoming tedious, but I still think there's something fishy going on here: We agree that Lynn, using that Extrapolated Runs formula, rates about 60 runs above average in 1979. The XR I had for Wynn in 1965 was 104.3. Now, the league average line--including pitchers--for the 1965 NL was .249/.311/.374. Adjusted for Houston, that would work out to .241/.301/.362 by my calculations. A hitter with those numbers using 409 outs would have this estimated line:

93 1B, 21 2B, 4 3B, 12 HR, 46 BB+HBP, 7 SB, 4 CS

Plug that into that XR formula and you get 58.1 runs, or 46.2 runs below Wynn. (I have 46.2 RCAA as being worth about 5.06 wins in th 1965 AL.) So even including pitchers in the average--which would significantly overrate Wynn, as he wouldn't be taking those outs away from pitchers--he still comes out as significantly worse than Lynn. What's going on there?

An interesting pattern of candidates, just assuming for the moment that Keith Hernandez is the only viable newbie in 1996.

Backlog (top 20, ? 200 points) + Hernandez

Cluster #1--Hernandez, Wynn (#2 backlog), Perez (9)--post-expansion hitters

Cluster #2--Keller (1), Bob Johnson (10), )Oms (15)--lively ball hitters

Cluster #3--Roush (4), Browning (5), Duffy (13), GVH (17)--19C/deadball CF glut

Cluster #4--Beckley (6), C. Jones (8), Cravath (14)--19C/deadball hitters

Cluster #5--Bresnahan (16), Leach (18)--deadball hybrids (glove + CF)

Cluster #6--A. Redding (12), Grimes (20); B. Walters (11), Dean (19)--deadball/lively ball SP

"Cluster" #7--Fox (3)--pure glove

"Cluster" #8--Fingers (7)--post-expansion/relief pitcher

The 2 obvious patterns are:

1) the top 20 + 1 include 9 whose value is almost purely hitting plus 6 from the CF (including hybrid CF) glut (71 percent of pool)

2) 4 of the 21 are post-expansion (19 percent)

I don't think this shows that we are over-rating hitters BTW or under-rating post-expansion players. What it shows is that we are generally in agreement on the post-expansion players and generally in wild disagreement about the hitters and CF glut, such that we can elect the post-expansion players, leaving a small backlog, but cannot quite find agreement on pre-expansion hitters and CF, leaving them in the backlog year after year.

As to the gloves, I am not sure--I'm not sure they're not in the backlog because we're not electing them, I think it's more a case of not being in consideration sets.

As to the pitchers, I don't know. Maybe 5 out of 21 is what you would expect.

DCW3, I'm very glad that you're insistent on getting this exactly right and understanding why.

1. Your math in #126 is accurate.
2. Wynn's double play avoidance was worth four runs. You'd be wrong to discount that, it's real and it matters. Adding that on brings him up to 50 runs above average.
3. Since you are applying the park adjustment to the league-average player rather than to Wynn (which is a strange way to do things if you ask me, but it gets you to the same place), you're seeing how many runs above league average he would have been *in the Astrodome*. Since the Astrodome is a lower run environment, one run above average buys more wins than it does in an average stadium (which is why we have park factors in the first place). An average 1965 NL team in the Astrodome would score 594 runs. An average 1965 NL team in the Astrodome with Wynn would score 594 + 50 = 644 runs. 644 RS and 594 RA in 162 games is 86.8 wins, or 5.8 above average. And Lynn was 60 park-adjusted runs/5.9 wins above average in the 1979 AL.
4. Then, of course, you have to account for the DH (benefiting Lynn), and also for Wynn's extra PA, which benefit him relative to replacement. The DH adds 0.5 wins to Lynn and the extra PA add 0.1 wins to Wynn, so Lynn winds up 0.5 wins better than Wynn--a small, but meaningful, difference.

Wynn's double play avoidance was worth four runs. You'd be wrong to discount that, it's real and it matters. Adding that on brings him up to 50 runs above average.

Dan, correct me if I'm wrong, but are you somehow giving more credit to Wynn for double play avoidance than Lynn? Wynn grounded into more double plays than Lynn did, not less. Wynn grounded into 12 double playes and lined into two others in 1965 - as far as I know, these numbers aren't particularly impressive. Why is he getting credit for this?

Always on my case, Dizzypaco! (I repeat, that's a good thing, and I appreciate it).

Check Baseball Prospectus's "double play rate for batters" numbers (on the sortable stats pages). In 1965, Jimmy Wynn hit into 15 double plays in 167 opportunities, a 9.0% rate, when the league averaged 13.0%, so he hit into 6.79 fewer double plays than a league-average player. In 1979, Fred Lynn hit into 16 double plays in 134 opportunities, an 11.9% rate, when the league averaged 14.0%, so he hit into 2.81 fewer double plays than a league-average player.

The 5.9 offensive wins above average for Lynn already includes his (lesser) double play avoidance.

I find it funny that I seem to be Jimmy Wynn's best friend on this thread when I don't even vote for him. All I'm trying to do is explain to DCW3 why his 1965 offensive performance was only 0.5 wins worse than Lynn's 1979 offensive performance.

Check Baseball Prospectus's "double play rate for batters" numbers (on the sortable stats pages). In 1965, Jimmy Wynn hit into 15 double plays in 167 opportunities, a 9.0% rate, when the league averaged 13.0%, so he hit into 6.79 fewer double plays than a league-average player. In 1979, Fred Lynn hit into 16 double plays in 134 opportunities, an 11.9% rate, when the league averaged 14.0%, so he hit into 2.81 fewer double plays than a league-average player.

I didn't even know BP had that information. That's great stuff.

Unfortunately BP's NetDP data only go back to 1959, but it's better than nothing.

The second version of my WARP (which should be done by the "1996" election and will include every position player in both leagues with more than 50 PA since 1893) will include sac flies, net double plays, and non-SB baserunning runs, and include a discount for IBB's (valued, according to XR, at 3/4 of a non-intentional BB).

I wonder if yest would have voted Joe Morgan #1 if I had told him that he had the #1, #3, and #7 (as well as #40) seasons for DP avoidance since 1959. A lot of otherwise great seasons are also among the top DP avoidance years (Morgan 75 and 76, Mantle 61, Bonds 01, Griffey 96, Giambi 2000, Sosa 01, ARod 05, Reggie 69, Beltrán 06, etc.,) which I guess makes sense--if you're not making outs generally, you're less likely to make them in DP opportunities.

DCW3, let me know if I've answered you effectively about Wynn vs. Lynn, or if it still isn't adding up for you.

Okay, I know this is becoming tedious, but I still think there's something fishy going on here: We agree that Lynn, using that Extrapolated Runs formula, rates about 60 runs above average in 1979. The XR I had for Wynn in 1965 was 104.3. Now, the league average line--including pitchers--for the 1965 NL was .249/.311/.374. Adjusted for Houston, that would work out to .241/.301/.362 by my calculations. A hitter with those numbers using 409 outs would have this estimated line:

93 1B, 21 2B, 4 3B, 12 HR, 46 BB+HBP, 7 SB, 4 CS

Plug that into that XR formula and you get 58.1 runs, or 46.2 runs below Wynn. (I have 46.2 RCAA as being worth about 5.06 wins in th 1965 AL.) So even including pitchers in the average--which would significantly overrate Wynn, as he wouldn't be taking those outs away from pitchers--he still comes out as significantly worse than Lynn. What's going on there?

The big mystery is why a guy this smart is doing real estate telemarketing for a living.

On DP avoidance - agree it's a great tool. But, it has also been shown to be true that players hit a bit better with men on first base (the "hole"). So, if you are going to crrect for someone hitting into DP's by how many men were on 1B, shouldn't you also correcct for their AVG/OBA/SLG also? That wol dbe more consistnet.
Jim Rice was famous for GIDP. Part (only pat) of his problem was there were many GIDP opps. But it also meant he had lots of people to drive in, and maybe his AVG benefitted as well. The hits he got were real, and so were the DPs. Count em both. Count em both.

TomH, how would we do that, mathematically? And do we have the data?

Check Baseball Prospectus's "double play rate for batters" numbers (on the sortable stats pages).

I'm pretty sure the numbers are wrong. Flat wrong. Fred Lynn did not hit into 16 double plays in 1979.

I went through every box score where some kind of double play was tied to his name - yes, there were 16. In two cases, a player was thrown out stealing, and Lynn later made an out in that at bat. In five cases, Lynn lined out, and the player on base was doubled up - not always from first, so that should not be counted in the same way (unless you want to add all times that a player was on second or third with less than two out as an opportunity). In total, Fred Lynn grounded into 9 double plays in 1979. Lynn grounded into less double plays per opportunity in 1979 than Wynn did in 1965.

You also have to consider that Lynn typically batted with slow runners on first base, while Wynn typically batted with fast runners on base. All in all, there is no reason that I see for giving Wynn more credit for GIDP avoidance than Lynn.

On this last point...

Wynn had a lot of GIDP opportunities, especially for 1965. The reason for this was very simple - Joe Morgan. Wouldn't you expect the speed (and baserunning ability) of the runner on first to have some impact on the number of GIDP?

Wouldn't you expect the speed (and baserunning ability) of the runner on first to have some impact on the number of GIDP?

I could be wrong but I think I remember reading somewhere that the runner on first had almost no impact on DPs.

Outside of a player stealing a base (isn't that already corrected for?), I don't see how a player's baserunning ability would effect a GIDP. You have got to be either really, really fast or have there be a bobble (and therefore a chance to take out the 2B/SS) for baserunning to be a factor. I would have to agree with AJM here, I don't see how that matters.

Is there a similar AL pitching vs. NL pitching bias in WARP post 1973?

Count em both. Count em both.

Get the papers. Get the papers.

What about on line drive/fly ball double plays? Would you agree that a better base runner is less likely to be doubled off first or second than a bad baserunner? This is, after all, the heart of the difference between Lynn and Wynn on the double play issue.

I've rippled the effects of the AL batting runs adjustments through my spreadsheet. Here are the big changes:

1-7 on ballot are unchanged
8) Rusty Staub (up from 13)
12) Graig Nettles (48)
15) Buddy Bell (60)
16) Thurman Munson (44)
17) Ken Singleton (43)
20) Darrell Porter (51)
22) Bobby Bonds (33)
35) Gene Tenace (55)
43) Boog Powell (52)
64) Jim Rice (108) - tied with Colavito now
86) Fred Lynn (116)
97) Chet Lemon (140)

Wynn's double play avoidance was worth four runs. You'd be wrong to discount that, it's real and it matters. Adding that on brings him up to 50 runs above average.

You've basically answered my questions, but in terms of double plays: I imagine that a big reason that Wynn hit into fewer DP than Lynn (if indeed he did) was that Wynn struck out so much--sixth in the league that year, and 47 more times than Lynn. If you're going to take double plays into account, shouldn't you correct for the negative effects of strikeouts (failing to move runners over, etc.) as well?

The big mystery is why a guy this smart is doing real estate telemarketing for a living.

Geez, now you sound like my parents. Although it was *not* smart to stay up an hour later than I ought to have working on this stuff.

(By the way, Daryn, I never did thank you for complimenting me in the Lounge a while back--though I'm not sure how well it reflects on me that I feel the need to admit mistakes so often that my name has become associated with it.)

Daryn, I never did thank you for complimenting me in the Lounge a while back--though I'm not sure how well it reflects on me that I feel the need to admit mistakes so often that my name has become associated with it.

Now, whenever my girls tell me they have made a windy, I think of you.

I've got Hernandez #2 on my prelim ballot. Quiz #29, Lemon #34, Reuschel #42, Lynn #63.

I personally never thought much of Chet Lemon as a near-HoM quality player, do I have an error in my calcs or is anyone else drawing the same conclusion from the stats?

DL from MN--no, I don't think there would be. A 100 ERA+ in the AL is the same as a 100 ERA+ in the NL, but a 100 OPS+ in the AL (compared to the DH) is not the same as a 100 OPS+ in the NL (compared to pitchers hitting).

DCW3--yes, you're right, I should adjust for K's and I don't. I'll make that tweak now. We'll see how much it changes things.

OK, I've now corrected for K's, and the difference is pretty minimal--Wynn comes out with 116.7 park-adjusted XR instead of 117.0, since he had 31 more K's than the league average rate for his AB-H, and each K is 0.01 run worse than a regular out. Still 5.9 wins above average. Lynn had 15 K's more than the league average rate for his AB-H, so he loses 0.2 runs, from 125.4 to 125.2. Nonetheless, every little bit of accuracy helps--thanks for suggesting it.

Has anybody changed their mind about Jim Wynn yet? And if so, from what to what? (e.g. from an obvious HoMer to HoVG? from off ballot to on? etc.)

Failing that, can somebody summarize what we've learned about him here?

I personally never thought much of Chet Lemon as a near-HoM quality player, do I have an error in my calcs or is anyone else drawing the same conclusion from the stats?

That's the conclusion I get from WARP. Win shares is a lot less high on him. We've talked at times about Win Shares overrating centerfield defense, but WARP values it even more highly, I think. Since my system gives equal weight to the two systems (with other tweaks as well), I won't have him near my ballot. He's in my top 60, however, and his overall value in my system is almost identical to Fred Lynn's. It certainly seems like he was better than he is generally remembered as having been. When we get a discussion thread for him next week, I hope there will be a more thorough examination of his career.

1. We have learned that Lynn's 1979 was slightly better offensively than Wynn's 1965. Overall this doesn't tell us much. Lynn's best two seasons (34,33 in WS) are roughly as good as Wynn's (36,32) but it is seasons 3-7 where Wynn begins to put Lynn in the dust.

2. While we have talked about WS overrating CF defense, I dont' think that it does so in the modern era. As we have discussed before there seems to have been a shift in CF defense ar some point in the 50's or 60's. This is part of the reason why CF stops have offensive monsters like DiMaggio, Cobb, and Mantle after the 1960's, CF defense became more important. Long story short, I think that WS overrates CF defense in the pre-integration and immediate post integration eras because it doesn't take this shift into account. This is especially true in the 19th and early 20th centuries (before HR's and therefore flyballs became more common), one reason why Ryan, GVH, and Duffy are so loved by WS. So I am not sure that WS is really overrating the defense of Wynn, Lynn, Lemon, Cedeno, et al. as a group. Maybe individually there are discrepanices, but as a group I think it is pretty accurate.

I could be wrong but I think I remember reading somewhere that the runner on first had almost no impact on DPs.

Apparently they do. I believe Tango wrote a few years ago that fast baserunners on first do have a definite impact on keeping out of DPs. I was surprised, but I pretty much believe anything Tango says.

I just want to keep my name alive, by the way, for these young bucks who don't remember me.

Sunny on Pesky in 1995 Results
28-x-x-x-34-25-20-23. So how many do you give in place of x-x-x? 31 (average of 28 and 34)? Too much? 22 (average of 0-25-28-34)? No, that's clearly too little. 29 (average of 25-28-34)?

Note Pesky's decline, very young, already in year war plus one, to a Win Shares rating below his only prewar season (AAA-28-x-x-x-34-25). That mitigates the problem with your method of using four seasons unless only one is available before or after the war. Consider instead a more likely Win Shares trajectory for someone so young: AAA-25-x-x-x-34-28. The estimate x-x-x = 29-29-29 is the same because the average of the three observed changes is the same, but the evidence for Pesky's ability to put up 29-29-29 is weaker.

. . . in reply to Eric Chalek
</i>I see what you mean. I was thinking a '41 MLE to add into his career totals, but I see that was not your point. To me, using 3 years, though, can't be more than a WS or 2 different from where you're going to end up, can it?<i>

Suppose the quality of that AAA season is 17. Now 17-28...34-25 ==> sum 104, average 26. And career 26-26-26 is nine WS below less than 29-29-29. If the AAA seasons is 13, . . . 25-25-25 and twelve WS below the "sunny" estimate.

--
On Charlie Keller, I agree with Phil C----- that those 1.2 seasons of MVP-level play after the War make a crucial difference, undermining the point by Dan R #16 and Mike Webber #19, who argue by analogy to PECOTA and other projections. Keller in wartime is a form of interpolation where projections are a form of extrapolation.

DaEvans' part-time work in 1969-70 lifts those years to 25 NL HOMers - most, along with 1964, since the one-league NL 1893 total of 27 (minimum 10 G charts).

AL had 10 HOMers apiece (pathetic, yes) in 1969-70, so 35 total HOMers.
1964 also had the 25-10 split for 35.

1963 was 23-13 for 36.

Those are the highest numbers since integration - so far.

Last time over 36 HOMers was 1942, with 38 - and 42 in 1941 and 43 in 1940.

Long story short, I think that WS overrates CF defense in the pre-integration and immediate post integration eras because it doesn't take this shift into account. This is especially true in the 19th and early 20th centuries (before HR's and therefore flyballs became more common), one reason why Ryan, GVH, and Duffy are so loved by WS.

Mark - I don't think I can agree with the totality of your assessment here. I did some research on Duffy, noting that he played less than 50% of his outfield games in CF. Wht was especially fascinating was that in several of the seasons when he wasn't in centre, his range factor actually was the highest of his team's outfielders (and we should know they had some great ones). Also, his range factors were the highest on the team when he was listed as their CF.

So what does this tell us? For starters, the defensive spectrum is not stable and varies from team to team. In some cases, a LF will have more defensive value than a CF. Sometimes, we might be able to say the same for a RF.
We are also seeing this now as we review Keith Hernandez's defensive ability. I think we can surmise from our findings and recollections that a firstbasemen may have an easier time of being average defensively and we may in fact be correct in assigning little defensive value to an average 1B. However (and this is a huge however) a firstbaseman who is above average adds real value to his team He makes the rest of the infield better, as well as his pitchers. Furthermore, while we may have a real cap on average 1B play (similar to what we see in WS) I don't think we can really overestimate the value added by a good to great defensive 1B.

BTW - I think a great way to test 1B ability would be to compare the assist rates and error rates of infielders between year to year when the pitching staff remained mostly stable but there was a change of starting 1B. If anyone can provide examples of teams who meet that criteria, I'd be happy to crunch the numbers.

BTW2 - I posted my findings about Duffy's defensive spectrum manipulations on the Duffy/Ryan?GVH thread, if anyone's interested.
Maybe my findings might help explain how they always seem to outplay their pythags?

Errgh, might take a while to find the studies about GIDP. I can't rememeber anynmoe what I have on hard copy and what's all over the net :) I THINK one of SABR's By The Numbers issues has two studies, I'll have to go look.

But there was one study I recall that showed a small impact on speed of runner on 1st for GIDP; I believe it was more related to over hit rate. IOW, one middle IFer creeps closer to 2nd to cover on the steal, more grounders go through the hole. Less GIDPs. Probably not less GIDPs per groundout tho.

Something I posted to SABR-L in 2001:

------------------------------------

Using Retrosheet play-by-play data from 1978 to 1982 (the years
covered by Retrosheet while Garvey was a Dodger), I determined the
number of ground balls fielded by each 1B with a runner on first and
fewer than two outs and the number of times a 3-6-3 or 3-6-1 DP
resulted. Garvey did, indeed, rank consistently near the bottom of
the league in terms of the number of such DPs turned per
opportunity. I don't want to take up the list bandwidth by posting a
complete rundown (if I have time in the next day or two I'll format
it and post a link to it on the Web someplace), but for a quick
comparison, here are the numbers in each season for Garvey and Keith
Hernandez, who was the best at the 3-6-x DP in the last four of
those five seasons:

1978: Garvey 4/44, Hernandez 4/30
1979: Garvey 1/28, Hernandez 11/53
1980: Garvey 1/25, Hernandez 9/48
1981: Garvey 1/16, Hernandez 7/28
1982: Garvey 0/26, Hernandez 10/44

------------------------------------

I no longer have the full list for those seasons, unfortunately, but it wouldn't be that difficult to regenerate.

-- MWE

1978: Garvey 4/44, Hernandez 4/30
1979: Garvey 1/28, Hernandez 11/53
1980: Garvey 1/25, Hernandez 9/48
1981: Garvey 1/16, Hernandez 7/28
1982: Garvey 0/26, Hernandez 10/44

So taking them as the worst and best at turning 3-6-X dps, the range is

1978 9%-13%
1979 4%-21%
1980 4%-19%
1981 6%-25%
1982 0%-23%
------------
tot 5%-20%

So perhaps from a small sample we can interpolate the mean as around 12%-13%. But I'd hold out for complete numbers myself, because there's some chance that the average 1B is closer to Garvey than Hernandez. After all, this is the extreme end of the spectrum where guys who can't hack it elsewhere go to hit, and it's played in an extremely risk-averse fashion in general. Hernandez may well be performing at an extraordinary rate in terms of his throwing due to those factors.

Regarding GIDP,
Tom Ruane, in SABR-L posts on March 29 and 30, 1998, showed that players who tend to ground into a lot of double plays are, on average, more prodcutive than hitters who strike out and fly out a lot. This is because they reach base on error more and their outs advance runners more (despite the loss of more runners on the DP.)

Mark Pankin, in a paper available on the Retrosheet WWW site, documented that the speed of the runner on first is a significant factor in the rate that a batter grounds into double plays.

ryan,

Don't worry I do have your guy (Duffy, that is) as my #2 backlog guy. However, if I went only be WS and I completely believed what WS was saying he mught be even higher as his WS peak (when adjusted for schedule) is monstrous. I don't believe that (and there are other reasons for it that I am willing to give credit for).

AL had 10 HOMers apiece (pathetic, yes) in 1969-70
Electing Bobby Murcer, Roy White, Thurman Munson, and Graig Nettles would certain help correct that imbalance.

Most HOMers in a given year by position (in this one, you need 1 IP per G or 35 G as a P; or be a 'regular' - about half a team's games in the field/300 AB):

P (12)
1926 - WJohnson Williams Alexander Rixey Faber Covaleski Rogan Vance Lyons Grove Ruffing BFoster
1929 - add Paige, Hubbell, Ferrell; subtract WJohnson, Alexander, Covaleski)
(with 11 in 1925-27-28-30-32; and 1962-65-67-69-70)
NOTES: Grimes would be No. 13 in 1926 or 1929; and No. 12 in 1925-27-28-30. Dean or Bridges would be No. 12 in 1932. Tiant would be No. 12 in 1965-67-69. Blyleven would be No. 12 in 1970.

.........................................

OF (13)
1924 - Cobb, Speaker, Wheat, Carey, Torriente, Heilmann, Ruth, Charleston, Goslin, Stearnes, Simmons, Suttles, Bell
1926 - add PWaner; subtract Suttles
1957 - Slaughter, TWilliams, Doby, Ashburn, Snider, Minoso, Mantle, Mays, Aaron, Kaline, Clemente, Boyer OF-3B, FRobinson
1963 (13) - Musial, Snider, Minoso, Mays, Aaron, Kaline, Clemente, FRobinson, BiWilliams, McCovey, Killebrew, Yastrzemski, Stargell OF(1B)
(with 12 in 1925, 1927, 1951-54-55-56)
NOTES: Roush would be No. 14 in 1924 or 1926; and No. 13 in 1925 or 1927. EHoward would be No. 14 in 1957 or No. 13 in 1955-56. Brock would be No. 14 in 1963.

.....................

C (6)
1939 - Mackey, Hartnett, Dickey, Gibson, Trouppe C-3O, Campanella
(with 5 in 1931-32-33-34-35-38)
NOTES: nothing cooking here.

..........................

1B (8)
1934 - JWilson, Terry, Gehrig, Suttles, Foxx, Charleston, Greenberg, Leonard
1935 - same
1936 - add Mize; subtract Greenberg
(with 7 in 1929-30; 1937; 1974)
NOTES: nothing more from 1929-37; TPerez would be No. 8 in 1974 (which had Yaz, Freehan, BiWilliams, Rose, Torre, and Allen joining lone 'real' 1B McCovey).

.............................................

2B (5)
1892 - Richardson 2B-OF, McPhee, Grant, Childs, Ward
1924 - E Collins, Hornsby, Frisch, Lloyd, Dihigo UT
1926 - add Gehringer; subtract Dihigo
(with 4 in 1888-93-94-98-99; 1925-27-28; 1938-39-40-41-46)
NOTES: nothing cooking here.

.........................................


3B (6)
1966 - Mathews, Boyer, BRobinson, Killebrew 3B-1B, Santo, Allen 3B-OF
(with 5 in 1964-65-71-75)
NOTES: TPerez or Nettles would be No. 6 in 1971. Nettles would be No. 6 in 1975.

.....................................

SS (6)
1940-41 - Wells, Cronin, Appling, Vaughan, Boudreau, Reese
(with 5 in 1901-04-05-06-07-08; 1936-37-42
NOTES: Rizzuto would be No. 6 in 1942.

........................................

DH (4)
1974 - FRobinson, Kaline, Killebrew DH-1B, Santo DH-2B-3B
(with 3 in 1975, 1984)
NOTES: Singleton would be No. 4 in 1984.

Mark - I don't think I can agree with the totality of your assessment here.

This phrasing reminded me of Fargo.

Prelim time!

What I've been doing: Looking at Keller. I've given him peak-level credit for 1944, but still a bit lower than what would come out by averaging the surronding seasons. I'm also using a linear adjustment for 1945. I could be convinced to change either (putting 1944 at the level of the surronding seasons probably puts him in the ballot).

Also, I have changed all the negative BRARs for pitchers into zeroes. As a result, they move up as a group, with one of them (Vic Willis) making it into the ballot.

1) Gavvy Cravath
2) Bus Clarkson
3) Luis Tiant. As a result of the aforementioned BRAR adjustment. In a weak, bunched ballot, it pushes him that far.
4) Ale Oms
5) Tony Lazzeri
6) David Concepción
7) Jimmy Wynn
8) Jake Beckley
9) Vic Willis
10) Tony Perez
11) Keith Hernandez. Another "raw" standing. I wonder if he could be given more defense credit than what the uberstats say.
12) Bob Johnson
13) Charley Jones
14) Toby Harrah
15) Rollie Fingers

As I use the Retrosheet data files more and more (and as Dave Smith and Ted Turocy fix shortcomings with the data analysis tools), I constantly realize how little I knew when I first started using them back in the dark ages around the turn of the century :)

Anyway, I refined the query for DPs started by the 1B (primarily because in my original analysis I missed reverse force DPs). I defined a DP opportunity for the 1B as including every ball fielded by the 1B:

-- with a runner on first and less than two outs;
-- that was not a bunt; and
-- that was not explicitly identified in the Retrosheet data as a pop up, fly ball, or line drive.

I then counted the number of DPs turned in those opportunities. This includes some 3-2-3 and 3-5-3 DPs as well as 3-6-3/3-6-1 plays as well.

Anyway, for the period from 1979-1983, these were the results for 1Bs with at least 50 DP opportunities in that time frame:



"Jackson, R" is Ron, not Reggie :)

-- MWE

"Also, I have changed all the negative BRARs for pitchers into zeroes."

Errr...Why?

I should let Juan V. speak for himself, but I imagine the principle is that bad performance in one season shouldn't be able to cancel out good performance in another season.

A hypothetical reductio ad absurdum example: let's say that instead of retiring, Babe Ruth played until the day he died, batting .100/.150/.200 in 154 games every season. Add on 10-15 years of -4 WARP per season, and his career value might not even be top 10 all time. (I'd have to do the math, but this is just for the sake of argument). Would that really mean that he wasn't the greatest player ever? No. He'd still be the greatest player ever, he just should have retired sooner, or his managers should have had the guts to stand up to his reputation and ego and bench him when he was done. If you don't zero out sub-replacement seasons, you'd be led to the wrong conclusion.

Actually, it's not quite that. It's mostly as a consequence of the fixed replacement level for offense in WARP (BPro's, of course), which most pitchers throughout history fail to meet. The replacement level for pitcher offense is well lower than a .230 EqA (in fact, one could argue that there is no such thing), and this is what I've (somewhat crudely) tried to account for.

Mike,

How much do you think that the 2B, 3B, or C involved had to do with those numbers? Hernandez comes off well but not as well as I would have thought. However, I wonder if it is possible that maybe Chambliss had other defenders better at turning the DP. What are the numbers for 3-2, 3-5, and 3-6's v. 3-2-3, 3-5-3, or 3-6-3/3-6-1. Just wondering.

Oh yeah, I didn't see that you had said BRAR, not total runs. BP's accounting for pitcher hitting is disgraceful. The best-hitting pitchers (Dontrelle Willis, Carlos Zambrano, Jason Marquis) can add a full win to a team with their bats; BP will still have them subtracting value offensively.

Prelim:

1) Bresnahan
2) Keller
3) C. Jones
4) Walters
5) Welch
6) Browning
7) Willis
8) Hernandez - Wasn't sure he was going to make my ballot, but he did and in a strong position.
9) Wynn
10) Cravath
11) Oms
12) Elliott
13) Duffy
14) Traynor
15) Grimes

Lynn and Reuschel were real good, but will never reach my ballot.

Time for my prelim.
1)Hugh Duffy (PHOM)
2)Ben Taylor (PHOM)
3)Edd Roush(PHOM)
4)Lefty Gomez (PHOM)
5)Nellie Fox (PHOM)
6)Tommy Bridges (PHOM)
7)Charley Jones (PHOM)
8)Vern Stephens (PHOM)
9)Gavvy Cravath (PHOM)
10)Keith Hernandez (PHOM)
11)Bob Johnson (PHOM)
((11a)Biz Mackey)) (PHOM)
12)Bobby Veach
((12a)Willie Stargell))
13)Orlando Cepeda
((13a)Ken Boyer))
14)Al Oliver
15)Tony Oliva
16)Jim Rice
17)Wally Berger
18)Dizzy Dean
19)Bus Clarkson
((19a)Darrell Evans))
20)Dan Quisenberry
21)Bruce Sutter
22)Ernie Lombardi
23)Jimmy Wynn
24)Reggie Smith
25)Ron Guidry
26)Al Rosen
27)Mickey Welch
((27a)Jim Bunning))
((27b)Billy Pierce))
28)Sparky Lyle
29)Dick Redding (PHOM)
30)Ron Cey

Some explanations
Keller - serious concerns with in-season durability and career length, plus some mild concerns that his OPS+ numbers do not properly take park factors into effect. (early 50's)
Browning - more serious concerns about in-season durability, coupled with horrid defense and league strength concerns. (early 50's)
Wynn - 23rd. Minor in-season durability concerns. A solid, but not stellar CF. I just don't see what seperates him from the pack. (23)
Beckley - too much career, not enough dominance. Don't buy into his defensive greatness. (early 40's).
Fingers - A want peak for releivers.. Is behind Sutter, Quisenberry, Lyle and Gossage for me. (mid 30's).

rawagman, your explanation for your low ballot placemeet makes no sense. You cite 3 reasons for your concern: all of these reasons are inconsistent with the way you've treated other players on your ballot. Allow me to demonstrate:

1) In-season durability: Did you realize that in 2 of his 5 peak seasons, Keller was in the top 10 in the league in games played? In the other 2, Keller played 140 and 141 games of a 154 game seasons; that projects to 147 and 148 games played of a 162 game season. He's no Cal Ripken, but I don't think that in-season durability can be called a "serious" concern. Besides, WARP and Win Shares numbers already account for playing time, and both rank Keller's extant peak seasons as high-All Star, verging on MVP.

2) Career Length: You have Dizzy Dean ad Al Rose in your top 30. You are not shy about awarding credit, since your ballot includes Cravath and Jones, both of whom hold more tenuous credit arguments than Keller. Note that with appropriate war credit, Keller's prime is 8 seasons long. That is not unusually short. If you credit Keller for his last minor league season, he has 9 prime seasons. (Which is admittedly debatable, but remember, he was Minor League Player of the Year in his 2nd-to-last minor league season, ad was returned to AAA the next year!)

3) OPS+ doesn't accout for Park Effect: You're voting for Cravath, the King of Park Effect. Keller was a dead-pull hitter in Yankee Stadium, so of course he was benefitted by park, but that benefit had value. You either vote on value, in which case you minimize park effect concerns, or you vote on quality, in which case you fully explore them. You can choose either path, but not both, and a vote for Cravath but not Keller is highly internally inconsistent.

rawagman, I suggest that you're assembling a ballot, then providing a rationale which does ot reflect a consistent, quantitative approach to voting. There's nothing wrong with heuristic analysis, but it should be applied equally to all candidates, not as an ex post facto rationalization for not voting for a guy you dob't like.

'zop
A few points - all of these guys have one of those concerns - none have all.
- I do give war credit based on career rates. I'm comfortable with that. I won't be giving Keller any minor league credit. Being held back doesn't hold water with me.
- Keller had a nice prime. But there was nothing outside of it. That sort of thing works better with pitchers and 3B than it does for corner outfielders.
- OPS+ is supposedly adjusted for park effects. That is great when a park is symmetrical. Not so much when it is not.

I may yet move Keller up a bit, but I'm satisfied in my consistency, thanks.

I really have trouble buying the asymmetrical park effect for this project. Keller gave real value to his teams by taking advantage of his park. If our job was to figure out whether or not to sign Keller as a FA in 1945 for a team without the short porch, then such analysis makes a lot of sense. But that it not our job. And why would Keller have taken advantage of his park more than Cravath (who I vote for, not saying this is a bad vote) in the Baker Bowl?

Also, just like 'zop said, Keller's in season durability didn't stop him from having a ton of value per season. He sin't Chance or McGraw, player with great, HOM level peak rates who had trouble staying in the lineup and thus dont' have a lot of value in each season. He still had 8-9 high all-star/MVP level seasons.

However, I can see you not liking what is outside of his prime. That is legit, I just personally don't care too much.

Prelim -- why not, since I've done it.

Keith Hernandez not nearly good enough 2182@129 at bottom of consideration set. Rice minus a chunk. Yet again, thoroughly overrated by WS. Lemon is Hernandez, but for a shorter time. Lynn’s WAY better than Hernandez and better than Rice, WS seriously out even in the same era. Reuschel very close to Tiant, but Tiant moves down to meet him, not vice versa. White Buckner and Boone WAY off the bottom; WS needs Loss Shares to rank them properly.

1. (N/A-9-9-10-7-7-5-5-4-3-5-6-5-3-4-3-4-4-3-2-2-1-1-2-6-4-4-2-1-1-1-3-
1-2-2-1-1-2-2-1-1-2-2-3-1-1-1-1-2-1-3-1-1-2-1-1-2-1-1-2-1-2-1-1-3-1-1-1
-2-2-2-2-3-1-1-1-1-1-1-2-2-2-2-2-1-2) Jake Beckley. Better than Sisler (1 point OPS+, 118 hits, more dangerous/difficult fielding position) and we’ve elected Sisler. Paul Waner is a very close comp (it was 37 years till we found one) and it thus makes no sense to have Waner far above Beckley. Significantly longer career than Clemente or Brock when you adjust the schedule, much longer relative to his contemporaries (he was #2 in AB when he retired, and #5 20 years after he retired.) Adjust his 2930 hits to full seasons and he's up there with Nap, above Babe, over 3200 hits, and OPS+ of 125 better than Van Haltren and slightly short of Wheat’s 129. Isolated power .127 vs “slugger” Wheat .135, in a less power-centered era. TB+BB/PA .455, TB+BB/Outs .707. Played for un-famous teams. Better than Keeler, almost as good as Crawford. More than a borderline HOMer, somewhere in the reaches well above the border but below the immortals. Should have been elected 80 “years” ago.

2. (N/A-7-7-6-8-6-6-7-7-6-7-7-7-9-8-7-7-4-5-3-3-3-5-4-4-4-6-4-4-4-5-2-2
-4-4-3-3-5-4-3) Addie Joss. I’m now even more convinced I missed him earlier, and that adjusting innings down for dead ball pitchers is illegitimate. 2327 IP at an ERA+ of 142. 160-97 by age 30. If you assume the rest of his career would have been 1800 IP, 120-90 with an ERA+ of 110 (somewhat conservative, assuming you boost his last sick season, though pitchers didn’t last as long as they did later) then 50% credit would put him at 3227IP, 220-142, with ERA+ of 130. 25% credit puts him at 2777 IP, 190-120, with ERA+ of 136. Substantially better than Koufax. OPS+20.

3. (N/A-10-8-7-6-4-3-3-5-9-7-8-6-4-4-4-6-4-5-6-5-4-6-7-6-5-5-6-7-5-5-4-
4-5-4-6-4-4-5-4-4-5-4-4-6-5-5-5-6-7-5-5-6-7-6-5-5-7-5-5-5-6-3-4-7-6-4-4-6-5-4) Eddie Cicotte. Only 208-149 and an ERA+ of 123, but 3223 IP, more than Waddell and should get about 25% of the bonus for the 300-win career he should have had (he was, after all, a knuckleballer, who tend to peak late.) Much better than the 20s glut – only loses to Welch on longevity – Newhouser a close comp, but Cicotte had a longer career. Successfully cursed Red Sox AND White Sox for over 8 decades!

4. (N/A-10-9-8-10-11-10-13-12-14-N/A-15-14-13-12-11-10-10-11-9-9
-10-11-10-10-10-11-11-10-9-10-7-8-11-10-8-8-10-9-5) Ernie Lombardi. Up a bit more; we’re forgetting him. berra closely comparable Berra. 2137 hits, normalized to a 130 game season, and an OPS+ of 125 makes him a little better than Schang, but some of it was during the war years and he fielded badly. TB+BB/PA .492, TB+BB/Outs .719., the ratio between the two very low because of strikeouts, I assume. Plus a great nickname!

5. (N/A-14-15-14-13-14-15-14-15-14-15-15-13-12-13-10-11-13-12-10-
11-12-11-6) Vern Stephens. Short career – only 1859 hits, but comparing him to Reese he was much better, and not far short of Doerr. TB+BB/PA .508, TB+BB/Outs .756. OPS+ 119 Best season 1944, however. Sliding up ballot.

6. (N/A-12-10-12-10-11-10-7-7-8-9-7-9-13-11-10-11-12-12-11-11-11-
12-12-11-11-12-13-12-15-14-12-14-11-10-11-11-10-12-11-10-9-9-
10-8-8-9-10-9-8-8-10-10-9-8-9-6-7-10-9-7-7-9-8-7) Sam Leever. Pity he wasn’t able to start at the normal time; 2 more years would have made him a NB. Only 2660 innings, but was blocked till 27 by the one-league 1890s and having a steady job as a schoolteacher. Believe he needs to be looked at seriously by others, and included in pitcher analysis. Mild plus for high level of moral probity.

7. (N/A-9-8-8-9-10-8-10-9-8-7-8-11-11-10-10-10-11-11-10-9-11-12-
11-14-13-11-13-13-13-13-12-11-14-13-12-11-11-12-10-10-11-12-11
-11-11-13-13-11-10-11-8-9-12-11-9-10-11-12-8) Wally Schang. When you normalize his career to 130 game seasons for the first 18 years, as I do for catchers, he gets to 1941 hits, more than Groh at an OPS+ of 117, very similar. Furthermore, TB+BB/PA=.455, TB+BB/Outs=.728, also significantly better than Groh, over very close to the same period. And he was a catcher, more difficult than 3B.

8. (N/A-10-9-8-11-N/A-15-15-14-10-10-11-12-10-12-11-9-8-9-10-10-9-
8-9-10-9-8-7-9-9-8-10-9-8-10-9-8-9-8-8-10-9-8-8-8-9-7-7-8-9-8-7-7-9-9-8-7-8-5-6-9-8-6-6-8-7-9) Charley Jones. Short career – only 1,780 normalized hits, even when adjusted to nominal 130-game-played season (but that’s more than Pike, with much less of an adjustment, and Jones too missed two prime seasons.) But OPS+ 149, TB+BB/PA .473, TB+BB/Outs .722, so above Pike and non-CF 90s OF. Down, but just above Browning

9. (N/A-15-N/A-5-4-4-6-10-8-9-7-5-5-5-7-5-6-7-6-6-7-8-7-6-6-7-8-6-6-5-
5-6-5-7-5-5-6-6-5-6-5-5-7-6-6-6-7-8-6-6-7-8-7-6-6-8-6-6-6-7-4-5-8-7-5-5-7-6-10) Pete Browning. Recalculating, to adjust ’82 as well as ’83-’92, he had 2,177 “normalized” hits, with no AA discount. However, TB+BB/PA .511, TB+BB/Outs .855. the same as Tiernan, not quite as good as Thompson, but he got no significant boost from the 1893-94 run explosion. Career OPS+162 vs. 146 Thompson and 138 Tiernan, but you have to discount a bit for AA. Also discount for not playing full seasons; the normalized hits should thus be about 1900, so drop him to just above Hondo.

10. (N/A-15-N/A-15-N/A-14-13-14-11-12-14-13-11-13-14-13-11) Frank Howard Very slightly better than Kiner – significantly longer career. Underrated by history. OPS+ 142 for 1774 hits. TB+BB/PA .546, TB+BB/Outs .805 in a pitchers’ park and era.

11. (N/A-12) Tommy John 288-231, 4710IP@111. Infinitesimally below Sutton, better than Kaat.

12. (N/A) Fred Lynn. Underrated, considerably better than Rice or Hernandez. 1960 hits at 130, but bonus for playing CF. TB+BB/PA .531, TB+BB/Outs .791. Lovely player to watch, and absolutely top-drawer at his best.

13. (N/A-13-12-13-13-12-14-15-12-13-11-11-N/A-11-9-12-12-N/A-15-15-N/A-14-N/A-15) George van Haltren. Had slipped too far at #44; we need more 90s stars. TB+BB/PA .469, TB+BB/Outs .765. Back on ballot again.

14. (N/A-15-N/A) Rusty Staub. 2716 hits at OPS+124. TB+BB/PA .484, TB+BB/Outs .724. Not quite as good as Beckley, for not quite as long. Will be on soon.

15. (N/A-7-13-11-13-14-14-14-N/A-15-15-15-N/A-14-15-15-15-N/A-
14-N/A-15-15-N/A-15-N/A-14-N/A) Hack Wilson TB+BB/PA = .588, TB+BB/Outs = .954, OPS+ 144. (he does appear to have known about BB, unlike some others.) Very short career, but quality too good to ignore. OPS+ slightly below Jones, so here he goes.

Hack Wilson having just slipped back on my ballot, I copmpared him to Keller, who's comparable as a very short career pure hitter. However even with 100% war credit (which I don't give) Keller's career is shorter than Wilson's, his 1943 has to be discounted for wartime league quality and his OPS+ is only a little better. IOW, Wilson significantly beats him overall. In any case, Keller has nothing like Wilson's 1930, a season for the ages.

rawagman, 'zop is saying that if you are dinging Keller for benefiting from an asymmetrical park, you should ding Cravath, who benefited even more from an asymmetrical park, as well.

A short note from Planet Lurkerguy: Just want to pop in to tell you guys how amazing these threads are, and that we SABR-challenged folks appreciate it (even if we only understand about 1/4 of the discussion). Btw, as a Strat player myself, are you guys like the greatest Strat players in SOM history? Or are you so numbers-oriented that you overthink things and end up losing more often than not because of your 14 hour analysis of gbas among 1b in the pre-integration era? Or, and this is most likely, do you consider Strat so bush league, so statistically infantile that it hasn't been worth your time since you reached puberty, and/or pre-Calculus?

Again, thanks for the great discussion. And thanks for that George Sisler getting in thing some time back. And, oh yeah, if Sisler makes it you gotta put King Kong in. Okay, I'll shut up and go play with my '20 Ruth Strat card now.

How much do you think that the 2B, 3B, or C involved had to do with those numbers?

Not much. I didn't run the numbers for later seasons, but I do remember that Hernandez's numbers were virtually the same once he moved from the Cardinals to the Mets, and Garvey's numbers were also about the same after he left LA for SD.

What are the numbers for 3-2, 3-5, and 3-6's v. 3-2-3, 3-5-3, or 3-6-3/3-6-1

Same table as before, but this time with the plays at other bases that did not result in DPs added in (as "FO", although not all of them were force plays, of course). The Pct column is the percentage of the time a play was made on a lead runner:



Chambliss drops a lot here, suggesting that he took the sure out at first much of the time when he didn't think he had a chance for the DP, rather than trying for the force. Note, also, that Garvey drops well below everyone else on this chart.

-- MWE

Wow, I completely didn't expect to see Kent Hrbek there. It may be because I am too young to remember anything but the end of his career.

Hernandez also seem to have a high DP/FO ratio. A godo bit of this is probalby his ability to get rid fo the ball quickly and make a good throw, but I still wonder how much of it may also be him having good MIers. Who were his MIers? He didn't play much with Ozzie did he?

Hernandez ranks the most highly on the combination of the two lists, befitting his reputation. In Ron jackson is right behind on both, Tony Perez is very close to Hernandez overall. Obviously it would be great to know what these figures looked like over the course of his entire career, but I don't want to push my luck. ; )

Also, these numbers blow my theory out of the water that it would be Garvey and not Hernandez that would be closer to the middle of the pack.

Over that same five-year period, Hernandez handled 70 bunts with runners on base, and got a runner other than the batter out six times. That may not seem like many, but that was both a higher total and a better rate than anyone other than Eddie Murray (6/42) during that same period. Garvey was 2/95.

-- MWE

Can anyone try to put together a quantitative estimate of how many runs Hernández saved in the field, between scooping and throwing, beyond the runs he saved through range?

What the heck, I'll post a prelim too:

1. Pete Browning
2. Hugh Duffy
3. Charlie Keller
4. Thurman Munson
5. Bucky Walters
6. Buddy Bell
7. Graig Nettles
8. Rusty Staub
9. Alejandro Oms
10. Frank Howard
11. Norm Cash
12. Chuck Klein
13. Tommy John
14. Elston Howard
15. Phil Rizzuto

Hernandez slots in at 16, and is hurt by my re-eval of middle infielders, which vaults Scooter from the 20's to #15. He should be afforded as much credit as Keller. I have Hernandez roughly equal in value to Perez, Ben Taylor and Beckley. In fact, those four are all between 16 and 20 on my ballot.

I'm very much sold that Munson was Freehan's equal when looking at the whole body of work, which explains his high placement.

'zop/mark/dan - to be honest, my current ranking of Keller does not reflect my comments on the potential of overestimating the OPS+ of a batter playing in an assymetrical park. It is something I question, but I have not consciously dinged him for it.
My recent ding of Cravath (he was close to elect me spots and has dropped to mid ballot) is more due to my own updating my defensive rankings.
The other concern I have over Keller, which is also reflected in my lack of love for Jimmy Wynn, is the utter lack of ink. I don't really rate players based on ink, but I do have a range of ink that I look for. It tells me that a hitter was better than his peers. I look for it a bit more, obviously, in COF/1B than I do with MIF/C. Keller sticks out as having one of the lowest ink rates among all COF in my consideration set. That's obviously partially a product of his in-season durability issues. 'zop - it's all well and good that Keller was in the top 10 in the league in games played in 2 of his 5 seasons as a full time regular (I'll even call it 4 of 7 to include a very generous war time credit). That still leaves us with a player a player who played 4 full "healthy" seasons, and had three more injury prone seasons.
I will reexamine him, but somehow, I doubt he'll rise too much.

When Thurman's plane crashed was really "the day the music died".

rawagman, consider that your ink system may be penalizing Keller for being an all-around hitter. Keller wasn't extraordinary in any one category (other than BB's, perhaps); but in any given season he could finish in the top 10 in BA, BB, HR, 2B, 3B, R, RBI, OBP, SLG, HBP, K.

Keller is often compared to Kiner, who had similar offense value. Kiner has tons of HR black-ink, but if you took Keller's 2B's and 3B's and turned them into homers, he'd be leading the league in HR, too. Don't discount Keller's total contributions because they were spread out over several offensive categories.

One other neat Keller point. Keller hit a ton of triples: top 3 in the league 3 times, top 10 5 times. That's a whole bunch of triples for a corner OF.

So I figured I'd check to see what the Yankee Stadium component PF was for lefty triples using Retrosheet data for 1957-1966. I couldn't do the calculation perfectly, since I couldn't separate out 3B's by handedness AND home/away at the same time. But we can get a rough idea by assuming 50% of total team PA's were home, 50% away...

Team triple rate, AWAY: 0.72%
Team triple rate, HOME: 0.52%
Lefty triple rate, total: 0.51%
Righty triple rate, total: 0.71%

This strongly suggests that Yankee Stadium played neutral for lefties with respect to triples, and thus Keller's triple rates are real.

Aaaargh! I flipped the AWAY and HOME numbers above. The correct data is:

Team triple rate, AWAY: 0.52%
Team triple rate, HOME: 0.72%
Lefty triple rate, total: 0.51%
Righty triple rate, total: 0.71%

For what it's worth, I don't think Keller was really Kiner's equal on offensive value. Keller's peak years were all in the mid-to-upper 50's in BRAA, with one 62; Kiner's got a 77, 74, and 66. That said, they may have had similar overall peak value after taking into account Kiner's poor fielding.

For what it's worth, I don't think Keller was really Kiner's equal on offensive value. Keller's peak years were all in the mid-to-upper 50's in BRAA, with one 62; Kiner's got a 77, 74, and 66. That said, they may have had similar overall peak value after taking into account Kiner's poor fielding.

Sure, Dan, but the key point is that the difference in black ink b/w Kiner and Keller stems more from Kiner's concentrated contribution in a few categories contrasted with Keller's spread-out contributions. Kiner has a 52 black-ink score; Keller has a 4. Kiner wasn't 13 times as good a hitter as Keller; he wasn't twice as good, he wasn't even 50% better. He's maybe 10% better offensively, a gap that Keller makes up in every other facet of the game. Black ink makes the offensive difference appear much larger than it actually is.

Yeah, According to WARP (I don't have anything else in front of me right now), Keller was a plus player with teh glove until he returned from the war, that is a HUGE imporvement on Kiner.

Also, I do wonder how much of Keller's 'lack of in-season durability' (I mean he did play 138+ plus games every year in a 154 game season) was due to the Yankees having Selkirk and Henrich around. I am pretty sure that his 111 game saeson in 1939 was due to the fact that DiMaggio, Henrich, and Selkirk were the starting outfield in 1938 and played pretty well. In fact 1938 is the only season in which he wasn't close to 30 WS among his full seasons until 1947.

I guess what I am trying to say is that I really doubt that Keller was injury prone so much as he was on a team that afforded him (God Forbid!) 10-12 games of rest a year beacuse of George Selkirk. Not saying that he deserves credit for those games, but to knock him because he was injury prone seems pretty weak.

It feels like I am being accused of using every possible way to single out the flaws in one man's career by a group who seem to be going out of their way to give that same man every possible benefit of every possible doubt.
In basic terms, we are deep in the dredges of our backlog. All the pretty girls are taken and we only have the flawed ones left. That is, they look fine on the surface, but we are seeing them in the worst kind of flourescent lighting which shows them all to least advantage.
1) I have good reason to beleive that Keller was injury prone. He suffered from congenital back problems. If anything, I give him credit for doing what he did in spite of those problems.
2) I have reason to suspect that his OPS+ is overplayed.
3) His career was short. Period.
4) For a corner outfielder, his defensive ability was of limited value (within most norms) and I have no reason to beleive that he outplayed the position (ala Keith Hernandez). I agree that he was a solid corner outfielder, but my system gives that minimal rewardover someone who was a poor corner outfielder.
5) Ink. I can only give you my word that I do no use ink as a be all and end all. Someone with similar ink, but without the other concerns (say Bob Johnson) can make my ballot and PHOM. Someone with tons of ink, but otherwise similar concerns (say Chuck Klein or Pete Browning) will also fall somewhat short of my ballot. The ink is but a tree in a forest.

In summary, with so many different questions concerning the Charlie Keller picture, I cannot in good conscience place him very high. Again, I will do a reval of all high backlog candidates, as I urged all voters to do. Keller may move up, he may move down. That is the charm of a backlog election.

Keller never led the league in OBP or SLG. His career totals are .410 and .518. Kiner led in OBP 1X and SLG 3X with career totals of .398 and .548. His league leading SLG totals were .639, .658 and .627. Keller never did slug .600.

And Keller led the league in OPS+ 1X at 167 but also had 164-163-158 years of 140 games or more. Kiner led the league 2X at 183 and 172 and also had a 182-154, all in at least 150 games. Of course, Keller's 167 came in a weak 1943 context.

Both Kiner and Keller are in my PHoM though I agree that Kiner has the stronger case--a clearly if narrowly stronger case.

Mark Shirk, you're right: Keller was part of a 4-man outfield before the war:



In 1939 and 1940, Selkirk was a hellofa player (OPS+ of 149 and 135, respectively). This explains alot.