User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
Page rendered in 1.0343 seconds
41 querie(s) executed
|
You are here > Home > Hall of Merit > Discussion
| ||||||||
Hall of Merit — A Look at Baseball's All-Time Best Thursday, November 25, 2004Pennants Added 2.0 (updated for the 1949 ballot)I’ve updated the Win Shares version of Pennants Added. For an explanation of the methods, try this thread. I’m going to list players by position. If someone is missing that you’d like to see added, just let me know. One other fix for 2.0 - I’ve updated the team games to team decisions. Since we are using Win Shares, it doesn’t matter if a team played 158 games, if they went 82-70, the player’s total should be based on 152 prorated to 162. There will be a separate thread for pitchers. Click into the discussion for the results. * - estimated Negro League numbers C PenAdd WSaR WS WARP3 Deacon White (3B) 1.208 331 500 84.4 Cal McVey (1B/3B) .891 228 320 34.0 Gabby Hartnett .845 244 342 110.1 Buck Ewing (1B/3B) .806 229 331 83.4 Mickey Cochrane .741 209 292 88.6 Biz Mackey (3B/2B)* .593 178 292 Roger Bresnahan (CF) .585 170 249 57.2 Wally Schang .572 174 262 70.6 Charlie Bennett .531 154 239 68.4 John Clapp .509 147 230 38.6 Deacon McGuire .399 124 232 47.1 Duke Farrell (3B) .395 121 214 57.0 Ray Schalk .394 120 206 65.8 Jack Clements .361 110 185 42.7 Johnny Kling .346 103 170 47.8 Chief Zimmer .339 106 181 56.3 Wilbert Robinson .189 60 143 23.0 1B PenAdd WSaR WS WARP3 Cap Anson 1.895 516 752 127.7 Lou Gehrig 1.508 388 518 142.0 Dan Brouthers 1.327 355 487 102.5 Roger Connor 1.304 350 505 119.7 Mule Suttles (LF)* .821 232 371 Jake Beckley .719 215 369 82.5 Joe Start* .715 207 342 36.0 Bill Terry .704 199 294 78.1 George Sisler (P) .667 190 317 80.1 George Sisler (non-P) .645 184 309 Frank Chance (C) .656 185 257 54.8 Ed Konetchy .632 185 308 70.6 Jack Fournier .573 163 245 55.2 Jim Bottomley .553 162 274 63.3 Harry Davis .543 159 269 52.2 Joe Judge .523 160 286 68.2 Dave Orr .492 135 191 34.6 George Kelly .389 116 205 45.5 Dan McGann .381 116 200 48.0 Jack Doyle (C) .356 109 207 38.9 Tommy Tucker .352 103 207 27.2 2B PenAdd WSaR WS WARP3 Eddie Collins 1.695 447 612 172.2 Rogers Hornsby 1.580 408 539 151.5 Nap Lajoie 1.490 395 545 157.6 Ross Barnes 1.122 266 342 55.4 Charlie Gehringer .952 266 404 126.6 Frankie Frisch (3B) .874 250 386 111.8 Hardy Richardson (3B) .805 226 342 72.6 Bid McPhee .788 233 384 91.9 Larry Doyle .715 205 310 59.8 Cupid Childs .655 183 284 70.5 Johnny Evers .633 184 286 66.2 Fred Dunlap .629 170 254 61.8 Tony Lazzeri .575 168 267 76.7 Buddy Myer .551 164 273 69.4 Tom Daly (C) .528 157 255 53.7 Del Pratt .520 155 262 71.6 Miller Huggins .468 140 236 65.9 Fred Pfeffer .464 139 260 46.9 Max Bishop .389 117 195 58.7 Bobby Lowe .377 114 221 48.9 Kid Gleason (P) .589 167 346 53.7 Kid Gleason (1895+) .161 51 154 23.9 3B PenAdd WSaR WS WARP3 Ezra Sutton 1.090 298 444 47.0 Jud Wilson (1B/2B)* .926 263 398 Home Run Baker .854 231 327 73.0 Tommy Leach (CF) .784 226 355 79.9 John Beckwith (SS)* .774 221 332 Jimmy Collins .710 200 309 75.1 Heinie Groh .701 196 295 79.2 Ed williamson (SS) .630 178 283 48.7 John McGraw .619 173 246 58.7 Pie Traynor .605 177 287 72.3 Billy Nash .603 176 287 58.6 Lave Cross (C) .597 179 324 80.6 Larry Gardner .586 173 280 63.8 Arlie Latham .552 160 281 36.8 Denny Lyons .542 153 229 44.9 Heinie Zimmerman .519 150 232 44.2 Jimmy Dykes .445 138 263 59.3 Bill Joyce .430 124 185 40.3 Freddie Lindstrom .408 119 201 50.9 SS PenAdd WSaR WS WARP3 Honus Wagner 2.170 544 712 179.9 George Davis (3B) 1.091 307 457 113.3 Bill Dahlen 1.052 299 452 118.5 George Wright* .973 246 334 44.3 Joe Cronin .839 235 353 102.6 Jack Glasscock .813 231 367 89.8 Bobby Wallace (P) .807 236 387 107.4 Bobby Wallace (no-P) .727 214 349 102.2 Hughie Jennings (1B) .658 177 257 65.7 Herman Long .650 188 313 63.6 Joe Sewell (3B) .631 184 294 87.5 Joe Tinker .605 177 278 71.0 Dave Bancroft .582 173 289 76.1 Rabbit Maranville** .559 168 322 87.7 Ed McKean .530 155 270 44.4 John Ward (SS/P) 1.772 425 672 79.2 John Ward (1884+) .516 149 249 40.6 Dick Bartell .513 154 269 88.0 Travis Jackson .453 135 224 57.6 Tommy Corcoran .353 109 247 49.7 LF PenAdd WSaR WS WARP3 Jim O'Rourke 1.484 410 604 99.8 Jesse Burkett 1.135 310 454 93.5 Fred Clarke 1.087 305 446 100.6 Ed Delahanty 1.086 291 417 104.5 Al Simmons .971 269 398 106.6 Zack Wheat .934 267 408 88.6 Sherry Magee .925 258 379 78.1 Harry Stovey (1B) .891 246 369 67.4 Joe Jackson .885 234 314 72.8 Jimmy Sheckard .869 245 375 90.5 Goose Goslin .856 244 376 91.2 Joe Kelley .844 236 357 88.5 Charley Jones (CF)** .723 197 287 47.7 Tom York .710 204 321 44.5 George J. Burns (RF) .695 196 314 71.2 Heinie Manush (CF) .653 188 300 63.0 Bobby Veach .641 183 288 74.4 Tip O'Neill (P) .665 182 264 34.9 Tip O'Neill .617 169 242 Topsy Hartsel .568 162 249 50.9 Elmer Smith (P) .855 226 336 22.2 Elmer Smith (no P) .501 143 221 John Anderson (1B/OF) .436 133 234 42.2 Sam Mertes (OF/2B) .436 127 200 40.5 Chick Hafey .433 127 196 49.7 CF PenAdd WSaR WS WARP3 Ty Cobb 2.312 592 775 191.4 Tris Speaker 1.970 509 676 182.2 Paul Hines 1.232 339 507 84.0 Billy Hamilton (LF) 1.058 286 399 96.1 George Gore .965 261 376 83.5 Cool Papa Bell (LF)* .854 254 441 Hugh Duffy (LF/RF) .830 231 348 74.1 Lip Pike (RF/2B)* .819 206 277 36.0 Pete Browning (LF) .813 220 310 55.0 Edd Roush .805 228 340 77.2 Jimmy Ryan (RF/LF/P) .818 235 378 79.9 Jimmy Ryan (non-P) .792 229 368 George Van Haltren (P) .908 259 412 78.7 G. Van Haltren (non-P) .783 225 361 Max Carey (LF) .775 226 375 93.2 Fielder Jones .739 211 327 80.0 Earl Averill .716 200 299 77.2 Roy Thomas .687 193 289 69.4 Mike Griffin .669 192 295 74.9 Wally Berger (LF) .634 177 253 67.4 Dummy Hoy .612 178 302 51.8 Hack Wilson (LF/RF) .579 162 236 61.0 Chick Stahl (RF) .529 152 235 53.3 Earle Combs .518 151 239 55.9 Benny Kauff .511 139 191 41.0 Lloyd Waner .492 146 252 62.3 Jimmy Slagle (LF) .384 113 197 40.2 Steve Brodie .361 110 203 50.8 RF PenAdd WSaR WS WARP3 Babe Ruth (LF/P) 2.498 618 792 225.0 Babe Ruth (non-P) 2.153 539 680 Sam Crawford (CF) 1.185 332 484 107.1 King Kelly (C) 1.083 294 420 87.1 Paul Waner 1.080 301 446 118.4 Harry Heilmann (1B) .919 258 380 92.8 Willie Keeler .879 248 388 97.3 Elmer Flick .878 235 321 83.9 Mike Tiernan .722 203 306 62.1 Kiki Cuyler (CF) .698 200 308 82.5 Harry Hooper .698 206 350 86.5 Sam Rice (CF) .689 206 346 79.2 Sam Thompson .676 192 294 79.5 Babe Herman .555 161 245 63.6 Chuck Klein .540 155 252 72.3 Gavy Cravath .539 152 220 47.1 Ross Youngs .538 152 227 55.8 Tommy McCarthy (LF) .408 118 205 27.2 Patsy Donovan .370 113 234 38.1 JoeD has the Imperial March Stuck in His Head
Posted: November 25, 2004 at 10:46 AM | 180 comment(s)
Login to Bookmark
Related News: |
BookmarksYou must be logged in to view your Bookmarks. Hot Topics2024 Hall of Merit Ballot Discussion
(169 - 1:15pm, Nov 26) Last: kcgard2 Most Meritorious Player: 2023 Ballot (10 - 1:16pm, Nov 25) Last: lieiam Mock Hall of Fame 2024 Contemporary Baseball Ballot - Managers, Executives and Umpires (11 - 6:01pm, Nov 24) Last: Ron J Most Meritorious Player: 2023 Discussion (14 - 5:22pm, Nov 16) Last: Bleed the Freak Reranking First Basemen: Results (55 - 11:31pm, Nov 07) Last: Chris Cobb Mock Hall of Fame Discussion Thread: Contemporary Baseball - Managers, Executives and Umpires 2023 (15 - 8:23pm, Oct 30) Last: Srul Itza Reranking Pitchers 1893-1923: Results (7 - 9:28am, Oct 17) Last: Chris Cobb Ranking the Hall of Merit Pitchers (1893-1923) - Discussion (68 - 1:25pm, Oct 14) Last: DL from MN Reranking Pitchers 1893-1923: Ballot (13 - 2:22pm, Oct 12) Last: DL from MN Reranking Pitchers 1893-1923: Discussion (39 - 10:42am, Oct 12) Last: Guapo Reranking Shortstops: Results (7 - 8:15am, Sep 30) Last: kcgard2 Reranking First Basemen: Ballot (18 - 10:13am, Sep 11) Last: DL from MN Reranking First Basemen: Discussion Thread (111 - 5:08pm, Sep 01) Last: Chris Cobb Hall of Merit Book Club (15 - 6:04pm, Aug 10) Last: progrockfan Battle of the Uber-Stat Systems (Win Shares vs. WARP)! (381 - 1:13pm, Jul 14) Last: Chris Cobb Syndicate |
|||||||
|
About Baseball Think Factory | Write for Us | Copyright © 1996-2021 Baseball Think Factory
User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
|
| | ||||||
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
jschmeagol - that's just the way the numbers fall. For players that truly had a significantly great peak - guys like Ross Barnes, Hughie Jennings, Home Run Baker, etc. - you can see the difference.
Barnes for example has the same number of WS as Hardy Richardson, but 33% more Pennants Added. Look at Baker and Lave Cross as another example. Compare Sam Rice to Harry Heilmann. Jennings has 65 fewer WS than Maranville, but more Pennants Added.
Also, this is only based on a context through 1939. There was more spread between the teams the further back you go in history - so a great peak didn't have nearly the pennant impact that you'd think. The tighter the teams, the more impact a big year has. When the spread is huge one of two things usually happen - 1) Your team would have won anyway; 2) Your big year wouldn't be enough to close the gap anyway.
So as we go through history and more team-seasons get added into the calculation - more team-seasons that are tighter with lower standards for winning a pennant - the players with more big years will move up faster than those without.
OTOH I have some questions. Thanks Joe for the info, BTW, I had been agitating for it from time to time, this is conceptually a very elegant measure, and I plan to take it very seriously, despite the fact that it doesn't help some of my faves like Hughie and Charley Jones. (OTOH it does help other faves like Pike (not even a complete career there), and it very much justifies players I've supported in the past who are now HoMers, from Barnes and G. Wright and McVey and Start.
But like I said I have a couple of questions.
1. Joe, it sounds like each player/season was matched only with other teams for the same season, is that correct? You said you've now gone only through 1939. There would be no logic to matching all previous seasons (only) but not future seasons. IOW if you match 1895 with 1929 or 1939, why not 1949? So I assume Jennings' 1895 was only matched up to the rest of the NL 1895. Is that correct?
2. And we still have timeline issues, right? (Not that I timeline. Maybe right now, however, we need that reverse timeline I was once accused of using.) IOW Player A and Player B have the same season, one in 1895 the other in 1935. Player A (1895) is at a disadvantage because of the happenstance (beyond the player's control, IOW) that all but 2 teams finished 20 GB or more, so there is a structural bias against Player A creating any PAs. Whereas in 1935 (let's just say, e.g.) all but 2 teams are within 20 GB, so there is a structural bias in favor of PAs. Is that what you are saying? How would one construct a timeline that would fairly compare eras?
But we see that players in the earliest days have some huge PA scores. That I assume is because of the very low replacement levels from a primitive player pool that makes it easy to dominate. So the structural disadvantage you mentioned, therefore, appears only to affect, say (as of today) about 1890 through...when? Could one infer that Childs was probably better than Doyle--i.e. close enough despite the structural disadvantage? Is Browning really better than Roush and Carey? Maybe Sam Thompson really better than Sam Rice after you adjust for this structural bias?
3. Also is there any merit to developing a rate for this? i.e. Jennings earned .737 in probably just 5 seasons for .14 per season. Is that more valuable than Herman Long's .770 in perhaps 10-12 above average seasons or .07 per season?
4. This goes to the next question or point of clarification. Presumably each player's total is the sum of a set of seasonal totals, right? So are 5 .14s really more valuable than 10 .07s, especially when you equalize for the unusually large spreads that character Jennings' era. (This may be a bad example because Long was more or less contemporary. But say Jennings vs. Sewell, Jennings' .14s would have had a bigger impact in 1930 while Sewell's .07 would have less impact in 1895. Is this a comparison worth thinking about--that is, again, the rates?
Great stuff but hoping for some clarification and discussion. Not sure yet if this will reflect in my 1940 ballot, but it will soon.
Also, if you put Jennings' 1895 on the other 11 rosters in 1895 and play 'em out, what does .14 pennants mean? I would think it means that out of 100 seasons equally distributed among those 11 teams (9+ each), 1.4 of them would win the pennant with Jennings at SS instead of an average (?) SS while the Orioles meanwhile got an averatge SS in Jennings' place. Or it is 14 out of 1000 or whatever.
That is several questions in one.
Finally, what is the ceiling on single season PAs? Looking at Barnes it must be at least .25, or looking at Collins and Wagner it must be no more than about .15 in the 20th century. Is that right? Can we get a list of the best single seasons?
Thanks gain, great stuff.
Trouble here is that Hughie didn't play 5 seasons, he played about 12. Plus, I think pennants-Added was designed to not be a rate stat.
We've already had access to the rate numbers and the career numbers. The more peak-oriented of us also bring in Top-3 or Top-5 Consecutive (or some variation).
Pennants-Added is one way to give peak-bonuses to players' career values and also lessen the contribution of below-average seasons. It eliminates issues of the definition of peak (how long, consecutive seasons or not)... these issues were often problematic as certain players scored better than other depending on how you defined peak. This method is nice in that its published.
This is just one number for us to add into each of our mixes, but I'm grateful for all the work that went into this. Thanks, Joe! The WSaR numbers are pretty cool as well.
*based on the ones on my spreadsheet, which just has active candidates.
Oh, and adding Roy Thomas to the list would be nice.
George Van Haltren......1.009
Hugh Duffy...............947
Jimmy Ryan...............927
Edd Roush................914
Pete Browning............905
Lip Pike.................895
Jake Beckley.............863
Fielder Jones............854
Harry Hooper.............835
Mike Tiernan.............826
Marc - players are evaluated using all seasons 1876-1939 in terms of pennant impact. I use the cutoff of the year we are voting in (since we are voting in January 1940, I use 1939 as the last season). But all seasons are weighed equally.
Things that aren't accounted for - timeline, league quality (one reason I included the WARP3 numbers). Also unaccounted for - defensive spectrum shifts - aside from WS accounting for the 2B/3B shift.
I would certainly give 1B a boost pre-1920, probably pre-1925 (to account for the few years it took to change attitudes, etc.).
Also catchers - like relief pitchers will be in the future they just don't get enough on their own, either because of manager tendencies or wear-and-tear - so unless you don't want many in, I think you need to give them a boost. I boost them 33% (if all they did was catch). I boost pre-1920 1B about 10%.
WS loves CF, they tend to be a touch overrated because they get too much credit for overall OF defense - so I drop CF about 10% and LF/RF about 5%.
Devin - Van Haltren is higher because his pitching WS are included. I didn't use Chris Cobb's numbers, I figured them on my own - and I've revised my sheet to use team decisions, not games, because WS are only based on team wins, not games. So if a team played 159 games with 6 rainouts, for WS it's not fair to adjust using 159, you need to use 153. For WARP, I believe 159 is the correct number, but I didn't do anything with WARP, those numbers are right off the site.
One other caveat - pitching replacement level is signficantly higher - I've got it set to where 220 IP = 1 full-time regular season in terms of what I subtract from replacement level.
I wouldn't use pennants added/seasons to come up with a rate stat - I'd use pennants added/win shares or WSaR. That would be a good way to see which players had higher peaks. But as David said it isn't designed to be a rate stat.
It's higher in the raw sense, because pitchers threw a ton of innings - not because pitching is more easily replacable or anything.
30 WS for a pitcher in 375 IP isn't nearly as impressive as 30 WS for a SS playing 130 games, that's all I meant.
Elmer Flick
Larry Gardner
Jack Fournier
Topsy Hartsel
I don't think it necessarily hurts Charley Jones.
Jones missed his age 31-32 seasons due to being blacklisted. Here's his rundown:
Age - WS - WSaR
25 - 3 - 2
26 - 22 - 15
27 - 25 - 18
28 - 33 - 26
29 - 40 - 33
30 - 24 - 18 (missed the equivalent of 35 games)
31 - xx - xx
32 - xx - xx
33 - 30 - 24
34 - 40 - 33
35 - 35 - 28
36 - 21 - 15
37 - 13 - 9
38 - 0 - 0
I think it's entirely reasonable to give him two more seasons of 30 WS, 24 WSaR (his 1883), which is about .17 more Pennants Added and puts him at 347 WS in basically a 12-year career. That's pretty conservative. Total is .985 PA.
If you give him 35 WS, 28 WSaR (his 1885), you are looking at 357 WS, and another .21 PA. That puts him over 1.000 PA (1.025). At that point it's reasonable to compare him to a guy like Harry Heilmann. That's where I have Jones, a little bit short of Heilmann.
I don't remember, but is there any reason to give him credit for anything pre-1875?
Replacement level is considered the same throughout histroy. The players from the earliest days had some huge PA scores because they had some monster years - a short schedule contributed somewhat, but a pennant is a pennant.
"Is Browning really better than Roush and Carey?"
No way - for one, WS can't dock him enough for his defense, which was truly abysmal. Also it doesn't account for the poor quality of the early AA, which Browning dominated. I guess you only meant that because of a structural bias (which doesn't exist, as I said in post 13).
"Also is there any merit to developing a rate for this? i.e. Jennings earned .737 in probably just 5 seasons for .14 per season. Is that more valuable than Herman Long's .770 in perhaps 10-12 above average seasons or .07 per season? "
Just to elaborate on what I mentioned earlier - NO! This is that stat. Hughie Jennings' 257 WS were worth .737 PA, Herman Long's 313 WS were worth .770 PA. So if you assume WS is a perfect metric, Long's longevity (no pun intended) was worth more than Jennings' peak.
But, Jennings did have 78% of the WS and 96% of the PA, so the gap was signficantly closed. Closed enough that if one could make a compelling arguement that Jennings is underrated or Long overrated for some reason (besides Jennings' peak, we've already accounted for that!) you could flip-flop them.
"Can we get a list of the best single seasons?"
I don't have that handy right now - but basically just find the top single-seasons for WS/162 games and that will be close (since PA are based on WSaR). A few examples:
Ross Barnes 1873 - estimated at 63 WS, works out to .258 PA. His 1876, 49 WS, .174 PA.
Honus Wagner 1908 - 62 WS, .255 PA.
Tris Speaker 1912 - 55 WS, .212 PA.
Bob Caruthers, 1885 - 73 WS, .238 PA - lower than the others because his replacement level was higher, due to picking up 65 of his WS as a pitcher in 482 IP.
Hoss Radbourn 1884 - 129 WS (100 WSaR!), .542 PA.
Hughie Jennings 1896 - 45 WS, 38 WSaR, .158 PA - George Davis 186 was 26 WS, 19 WSaR, .067 PA.
So two 19's are .134, one 38 is worth .158 - Jennings gets an 18% bonus for having the big year.
That's from two things:
1) changing the season normalizing factor from games to decisions.
2) adding the 1939 season into the body of seasons used to determine the typical pennant winning percentage and team record standard deviations.
So two 19's are .134, one 38 is worth .158 - Jennings gets an 18% bonus for having the big year."
If you look at raw WS, the boost is bigger. Jennings had just 45 WS as opposed to 52. So despite having just 87% of the raw WS, he gets 118% of the PA. So a 45 WS season gets an 18% boost over two 26 WS seasons.
The gain would be even bigger over two 22.5 WS seasons. Sherry Magee had a 22.5 WS season in 1918, that was worth .058 PA. Jimmy Sheckard's 1899 was 22.5 WS and .055 PA and Speaker's 1927 was .057. So a 45 WS season gets a 40% boost over two 22.5 WS seasons. I don't see how anyone could argue that this metric doesn't give enough credit to a big peak.
C - Roger Bresnahan .657, Wally Schang .652 (Charlie Bennett .609).
1B - Jake Beckley .863 (Joe Start .846, not including pre-1871 play).
2B - none (Hardy Richardson .922)
3B - Tommy Leach .909 (Heinie Groh .799)
SS - none (Bobby Wallace .925)
LF - none (Joe Kelley .963) - Charley Jones would be at roughly 1.000 with credit for blacklist seasons.
CF - George Van Haltren 1.009, Hugh Duffy .947, Jimmy Ryan .927 (Max Carey .918) - Edd Roush at .914; Van Haltren would have 361 WS, 261 WSaR, .916 PA w/out pitching credit.
RF - Harry Hooper .835, Mike Tiernan .826, Sam Rice .822 (Sam Thompson .777)
Pre-1880 pitchers - none.
1880-1892 pitchers - Jim McCormick 1.532, Mickey Welch 1.308 (Bob Caruthers 1.308; John Clarkson 1.405 only pitching).
1893-1910 pitchers - Clark Griffith .713 (Joe McGinnity .689) Vic Willis .679
1910-1929 pitchers - Carl Mays .631, Eppa Rixey .630, Wilbur Cooper .615, Burleigh Grimes .609, Jack Quinn .605 (Stan Coveleski .598).
Pitchers, using Fibonacci-RSI
pre-1880 - none.
1880-1892 - Mickey Welch 302-215 (263), Tony Mullane 287-214 (237) (Pud Galvin 359-315, 235).
1893-1910 - Clark Griffith 231-152 (218), Vic Willis 251-203 (187) (Ed Walsh 193-128, 181)
1911-1929 - Eppa Rixey 280-237 (195), Herb Pennock 230-173 (188), Jack Quinn 256-209 (188), (Red Faber 256-213, 183) Urban Shocker 186-118 (182)
I'd say that the standard at C is low for obvious reasons. I'd say a player is a serious candidate at .600 PA if he caught full time - Bresnahan is tricky because of not catching full time - his bonus isn't as high.
3B is a little lower than the rest, .800 making you a serious candidate instead of .900 - not sure why, since WS gives extra defensive credit for playing 3B. Maybe we just overrated Groh and Collins? WARP3 sees them (and Lave Cross) equal to Baker and Leach not sure who is right there . . .
I'm also trying to think of why I was missing on Leach. I guess his skills were a little more subtle. He needs a bit of a docking, due to playing CF for 1/2 of his career - but at worst that could lower him to the Collins-Groh level.
Either Thompson is underrated by WS, or we overrated him (I lean towards the former), but either way, I don't believe .777 should be the standard for RF, I believe .900 (as with RF/CF) is obviously what makes sense.
Overall though, our elections have been amazingly consistent with the Pennants Added metric. Pretty cool.
Beckley is the only tweener candidate (I guess Sisler too), but it's been tough, because each career is so one-sided - in terms of the peak-career value spectrum. If Beckley had the same PA, but 320 WS instead of 369, he've been in a long time ago (because he would have won over at least a few of the heavy peak guys).
If Sisler had 350 WS instead of 317, he'd have a few more of the career voters, etc. . . . We haven't had a balanced tweener candidate at 1B, so the standard really hasn't been set yet . . .
The pennants added metric measures a players' total contributions over the course of his career. It is, therefore, by definition a career value metric as "career value" has traditionally been defined. What the metric does is produce a total value for all of a player's contributions over the course of his career that takes into account the fact that units of value (WS, WARP, etc.) make different contributions to the ultimate end of winning a pennant depending on the time and place that they are earned.
There are two debates we need to have about the pennants added metric (and one uncontestable point we need to acknowledge):
(1) Is pennants added a better way to measure career value (total value under the line) than aggregating season-by-season value above some replacement level? This debate turns on whether one buys the assumptions built into the pennants added metric.
(2) If we can accurately measure the total additional value a player's high peak brings to his teams, should we stop there? Or should a high peak or high prime or long prime count further in our evaluation process? Joe and many others (even some who claim to be "peak" or "prime" voters) ASSUME that the answer is obviously the former. Perhaps they are right. But, as Bill James noted in the first historical abstract, many, many people in evaluating the greatest baseball players of all-time either assume or argue that we should be answering a very different question, not who brought the greatest value to his teams over the course of his career but who established the highest level of play (with very different definitions of how long one has to play at that level to have "established" it). I started out essentially trying to balance the two questions equally in constructing my ballot. I have moved (or been pushed by the group) to valuing the "total value under the line" question more (measuring in a sophisticated, career-shape considering way not unlike the PA calculcation), but still count the "highest established level of play" assessment as something like 1/3 of my system. The choice is, of course, up to each of you.
(3) (The uncontestable point): Pennants added are only as good as the underlying statistic used to calculate them. Therefore, treat Joe's numbers with however much respect you treat Win Shares, no more, no less.
The reason probably is that third basemen break down faster than many of the other positions which is not going to be highlighted by WS. Also, does WS give extra credit for the pre-Mathews third basemen?
Also catchers - like relief pitchers will be in the future they just don't get enough on their own, either because of manager tendencies or wear-and-tear - so unless you don't want many in, I think you need to give them a boost. I boost them 33% (if all they did was catch). I boost pre-1920 1B about 10%.
WS loves CF, they tend to be a touch overrated because they get too much credit for overall OF defense - so I drop CF about 10% and LF/RF about 5%.
It's not clear whether the Pennants Added numbers reflect these boosts, or they are just your personal preferences.
Win Shares doesn't do a very good job adjusting the pitching/fielding split for changing conditions. I think even the most ardent Win Shares supporters agree with that. Exhibit A for this assertion is the pitcher Win Shares during the 1870's/1880's. Pitcher Win Shares are also out-of-whack, but by smaller amounts, during the 1890's and probably the deadball era also.
The corollary to this is that the fielding Win Shares for this period are also out of whack, because whatever was mistakenly given to the pitchers should have gone to the fielders. This means that the Pennants Added numbers are also depreciated for those who played key defensive positions (C,SS,3B,2B), and the farther back we go, the more the PA numbers have been depreciated.
This affects Bennett more than Bresnahan and Bresnahan more than Schang.
This affects Long and Jennings more than Sewell.
This affects Dunlap more than Childs, Childs more than Evers, and Evers more than Doyle.
This affects Williamson more than J.Collins, Collins more than Leach, and Leach more than Groh.
IOW, when there are close calls at a key defensive position, this flaw in Win Shares acts as a reverse timeline, because the farther we go back, the more important the fielding actually was.
OTOH, if you really believe in the 1880's Pitching Win Shares numbers, then there's another dozen 1870's/80's pitchers available whose 3-yr/5-yr peaks will put to shame any of these players from the 20's/30's.
Joe is right. It would appear that, with only a couple of exceptions, we've done a very good job sorting out the best position players. About .900 seems to be the PA between in and out.
Bennett's peak is far enough back that it is probably seriously underrated by the WS fielding/pitching split flaw mentioned above. If that were somehow corrected for, I'm confident he would move ahead of Bresnahan, though probably not quite to .900.
For the same reason, I think that Dunlap would move up the 2B list. Whether he'd move over the .900 mark is open to debate, but is a serious possibility.
Leach appears to be an oversight; Williamson too (1880's fielding, again); Collins and Groh are the exceptions to the .900 rule.
Thompson certainly doesn't look good. With a real 1890's CF glut we've rejected them all (so far) in favor of Carey (who may be questionable).
The base premise behind these calculations is where the WS replacement level lies. If I'm understanding correctly, Joe, you're using either 6.5 or 7 WS for a full season. This may be too low as 10 Win Shares is the average regular on the 1962 Mets. A higher replacement level should drop all PA numbers, but hurts long careers with low peaks (e.g. Van Haltren, Beckley) more than short careers with high peaks (Flick, Kelley).
Is there a spread-sheet available for those interested to play with?
Thanks, Joe.
The metric does NOT reflect those boosts. The only adjustments reflected in the metric above are for replacement level (6.5 WS per 162 G or 220 IP) and for season length.
I will post the spreadsheet to the Yahoo group as soon as I get a chance (it's on a different computer).
WS has a .200 replacement level built in already - I can't see 10 WS per season as being replacement level.
Bret Boone this year had 9 WS in 148 G (basically 10 per 162 G). He had an OPS+ of 98 while playing 2B. I don't see how anyone could convince me that is a typical replacement level season.
Eric Hinske had 8, while playing 155 G as a 3B with an OPS+ of 74 - that could be considered replacement level.
David Eckstein had 9, while playing 142 G as a SS with a 77 OPS+ - I don't think that's replacement level.
Alex Cintron had 8 WS, while playing 154 G as a SS with a 69 OPS+ - I could see that as replacement level.
Maybe I could see bumping the replacement level to 8 or 8.5 WS per 162 G, but I can't see going any higher than that.
A .250 team (like the 1962 NYMets) would go 40.5-121.5, and accumulate 121.5 Win Shares. Each of the 12.33 positions would then get 9.9 Win Shares (assuming they were all equally bad, and that Win Shares portions them out evenly - I don't believe it actually does, though I would have to work through the math).
If you believe that those Mets represent replacement level, then this provides a bench-mark for full-time replacement level. The replacement level amount for a .500 team may be somewhat lower due to Pythagorean distortion, but I can't see it being much below 9 (though I could be wrong). If you believe that those Mets are above replacement level, then the Win Shares replacement number is correspondingly lower.
All of the examples you cite are from "glove" positions. Win Shares is biased against them, with a higher replacement level for fielding/pitching than for hitting. Also note that the DH messes with the calculations because it steals OWS from the other positions (2+ per position). Find some NL 10's at 1b/corner-of and see what your impression of their quality is.
Bret Boone this year had 9 WS in 148 G (basically 10 per 162 G). He had an OPS+ of 98 while playing 2B.
Something's funny there. An average offense on an average team gets 116 Win Shares to spread around, which is 13 average at 9 hitting positions (including DH). A DH with a 98 OPS+ playing 90% of the time should get 10-11 Win Shares. (Seattle missing their Pyth-proj by 6 games didn't help either.)
Should be (1.5 per position)
If I added up every player in history's "Pennants Added", will it add up to the total number of Pennants (or close to it)?
To hazard a guess, for the past 100 years (200 pennants), I suppose we would see about 250 "pennants added" in total. Which would mean if all players in MLB history were put in a dollar auction, immortals like Ruth (2.5 pennants)would be worth 1% of the total budget.
Tike Redman: 155 G mostly CF and a 79 OPS+, 11 WS (HardballTimes)
Danny Bautista: 141 G mostly RF and an 87 OPS+, 11 WS (HardBallTimes)
Pedro Feliz: 144 G half at 1B and a 98 OPS+, 10 WS (HardBallTimes)
These guys were 20-25% better than Hinske/Eckstein/BBoone?
Feliz 3.1
Boone 3.0
Eckstein 2.8
Bautista 2.7
Redman 2.1
Hinske 2.0
Cintron 1.0
Philly, I think it will add to "pennants above replacement level" :)
Hypothesis: When Pennants Added does add up to the total number of pennants, the proper replacement level will have been found.
Corollary: The total sum of Pennants Added for a league for a season would be approximately 1.
OTOH, this might be a different statistic, Pennant Shares. ;-)
Interesting question Philly Booster - honestly, I have absolutely no clue.
"Corollary: The total sum of Pennants Added for a league for a season would be approximately 1"
I would think not, but I don't know for sure.
For one thing teams in a league - if a player has a 25 WS season in the 1963 NL (10 team league) or 2004 NL (16 team league) they would be weighed the same, get the same pennants added, etc.. But in a 16-team league there are more teams, seasons, etc..
Interesting thing to mull over at least . . .
Okay for replacement level, I've been convinced, I'll bump it to 8 WS/162 games. But Jim - are you saying I should be giving pitchers a replacement level of only 8 per 330 IP? That 330 IP = a full season as a hitter? I can't fathom that.
For one, that's about 1300 BF or ~double the number of PA in a full-season.
Let me try the math . . .
.675*.52*81*3 = 85.293 WS for an average staff.
An average staff pitches 1458 innings. That works out to 12.87 WS for an average pitcher in 220 innings.
We know that WS has a built in replacement level of about .200. That's what 0 WS means, that you are a .200 player.
So if you want to set replacement level in the middle (.350 W%) you'd get about 6.435 WS per 220 IP, right?
But here's my problem with all of this . . .
I think a replacement team would be worse than the 1962 Mets. First, the 1962 Mets were 10 games under their Pythagorean record. They were a 50-110 quality team that was really unlucky. The next season, they were 51-111 with a 50-112 Pythagorean record.
So I think a .250 team with a .313 pythagorean is above replacement level, but I think a But I think a replacement player is roughly a .350 player. How do these two thoughts peacefully co-exist?
Take an offense of replacement level players, the team will play .350 baseball - with an average pitching staff. If you put replacement pitchers into the mix, now you've got a .350 offense and a .350 pitching staff and the team plays significantly worse than .350 ball.
A team that scores 700 runs (league average) and gives up 962 plays .350 baseball. A team that scores 494 runs and gives up 700 (league average) plays .350 baseball.
But a team that scores 494 runs and gives up 962 - they play .223 baseball.
So maybe we should look at the WS of a .223 team to figure out where replacement level should be?
A .223 team isn't that much worse than the .250 team that Jim used, but let's run the numbers anyway real quick.
Dividing the batting and fielding WS up between 8 position players it works out to 8.8 WS/162 games and for pitchers it works out to 5.7 WS/220 IP. Or 8.6 for the 330 IP pitcher Jim used (I still have trouble with that though).
Okay, are these the figures I should use to recalculate the numbers? Are we theoretically correct here?
A team that scores 308 Runs and allow 700 plays .200 baseball.
A team that scores 700 and allows 1367 plays .200.
a team that scores 308 and allows 1367 - they play .052 baseball (8-154) - you can't get much closer to zero but it just helps me to confirm that we are on the right track here theoretically, that's all.
81-81 243 Win Shares
116.6 Offense 41.1 Fielding 85.3 Pitching
Pitching staff
4 Average inning-eater pitchers, 336 IP apiece, and one bullpen guy, also average, 112 IP, total 1456 IP
Each "ace" gets 19.7 WS, the bullpen guy gets 6.5, total 85.3
8 Average everyday players: each gets 19.7 WS, total 157.6
Under Win Shares, a 336 IP average pitcher is equivalent to a play-every-inning average regular. I'm not quite sure how this translates to replacement level in real life, but theoretically the equivalency has some validity.
Pitcher's hitting screws this analysis up to some extent, though it's not clear to me how much. In general, 20th century pitchers haven't hit enough to earn positive Win Shares, though specific individuals do. Pinch hitters also can earn positive OWS ("stealing" them from the position players). These two factors would lower the 336 equivalency number somewhat (don't know how much, but adding a full-time DH lowers it to 300 IP, so it can't be less than that).
I think it's not out of line to say the WS underrates AL hitters by as much as 11% on average - possibly more.
Before we get to considering players whose careers took place after 1972 I (or someone) needs to figure out exactly what an average DH hits. Retrosheet's breakdowns will be a big help here.
My theory is that there is only one way to handle it. You have to add offensive WS to all AL teams (or more specifically offensive players). I realize that this goes against the concept of everything 'adding up' but there is no other way that I can see to handle the problem. The AL artificially adds offense to their game - in essence it doesn't add up there either. A player with the same batting events (assuming equivalent parks, etc.) gets fewer WS in the AL for no other reason than that the offensive credit is divided 9 ways instead of 8.
I've talked to Studes about this and he disagrees - but I really think that it's because I haven't had the time to explain it well. I don't see any other way around it, but maybe I've got tunnel vision on the issue.
There is no subsequent adjustment needed for AL pitchers because they are still being compared to league averages (they are already adjusted for the tougher 'degree of difficulty'). But the AL adds offense, so I think the only way to account for this is to add WS to the mix, even if it violates the everything must add up rule.
I've talked to Studes about this and he disagrees - but I really think that it's because I haven't had the time to explain it well. I don't see any other way around it, but maybe I've got tunnel vision on the issue.
Joe, I believe your thinking is correct, but your conclusion may be wrong?!
It will take more offense in the AL to produce a "WIN" than in the NL, so the same batting events in the AL SHOULD result in fewer win shares, as they are LESS VALUABLE than if they were produced in the NL.
What you seem to be trying to adjust for is not value, but ABILITY. If you're trying to get to ABILITY, then you'd need some type of adjustment.
On top of this, the "problem" is made worse I think by the silly zeroing out of "negative" batting win shares, which are primarily created by NL pitchers....
"Dividing the batting and fielding WS up between 8 position players it works out to 8.8 WS/162 games and for pitchers it works out to 5.7 WS/220 IP. Or 8.6 for the 330 IP pitcher"
I'm going to use these numbers unless someone speaks up and tells me I shouldn't.
I admit I'm sorta on the fence on this one. Basically, the AL is playing with 10 players vs. the NL with 9, so Win Shares is not going to be the ideal measuring tool in this case since both NL and AL teams have the same number of wins to divide up, unless you also had "loss shares" which hasn't really been invented correctly yet..
C - no movement.
1B - Frank Chance (was. 726) moves ahead of Ed Konetchy (was .748).
2B - Hardy Richarson (was .922) moves ahead of Bid McPhee (was .933).
3B - John McGraw (was .691) moved ahead of Billy Nash and Lave Cross. Nash (was .710) moved ahead of Cross as well (was .729). The gap is close enough that I would nudge Cross up to about equal with Williamson, due to his catching.
SS - Hughie Jennings (was .737) moves ahead of Herman Long (was .770) and Joe Sewell (was .738).
LF - Charley Jones (was .815) moved ahead of Tom York (was .825).
CF - Edd Roush (was .914) moves ahead of Jimmy Ryan (was .927) and Max Carey (was .918). Pete Browning (was .905) moves ahead of all 3. Lip Pike (was .895) moves ahead of all 4.
RF - Mike Tiernan (was .826) moves ahead of Harry Hooper (was .835).
Looking at these numbers and adjusting for WS's flaws, for league quality, for peak and prime (which I independently value, not just as a multiplier when calculating career value), and for position, I think these numbers very much reinforce the conlusions I have reached on all the eligible position players save one: Tommy Leach. I'm going to reconsider him thoroughly before the next election.
This approach strikes me as conceptual similar to hypothetically adding more pitching Win Shares to all teams after the introduction of the 5-man rotation to allow for direct comparison to pitchers from 4-man rotations. I'm not sure what I think about that.
I think, in general, Joe, you are attempting to use a favored statistic for some measure that it was not intended for. There are statistics that measure ability rather than value. If it is important to measure ability, then use one of those instead of Win Shares -- rather than warping WS for one purpose, but not for all the others in which WS boosts shortchanges a player due to non-ability-related factors.
There are going to be all sorts of issues when the DH comes into play
Complete agreement.
I think KJOK has it right when he says that offense is less valuable in the modern AL. Simple supply and demand; surplus vs scarcity. I think this is best handled with a league quality adjustment, though it only affects BWS. (I have no idea how WARP deals with this issue, not having looked that closely at the modern players.)
KJOK:
On top of this, the "problem" is made worse I think by the silly zeroing out of "negative" batting win shares, which are primarily created by NL pitchers....
Disagreement. Adding negative batting WS to the NL makes the problem worse. When compared with James' WS, this in effect takes pitching WS and converts them to batting WS; the smaller group of NL hitters divvy up a larger pool of NL batting WS, accentuating the difference wrt the AL.
jimd
About .900 seems to be the PA between in and out.
This appears to be about .750 after the revisions (with a .700 exception for 3B).
Again, Joe, thank you very much. I know it was a lot of work, but the results are very interesting.
Agreement. You're correct, the zeoring out of negative batting win shares LESSENS the problem, but as Joe has discovered, it's still a problem.
I think, in general, Joe, you are attempting to use a favored statistic for some measure that it was not intended for. There are statistics that measure ability rather than value. If it is important to measure ability, then use one of those instead of Win Shares -- rather than warping WS for one purpose, but not for all the others in which WS boosts shortchanges a player due to non-ability-related factors.
Agree here also. Win Shares is simply not a good tool for doing "HOM" evaluation - low replacement level, incorrect position adjustments over time, incorrect pitching/fielding splits over time, etc., etc.
.856 Van Haltren
.812 Ryan
.811 Leach
.810 Hooper
.800 Duffy
.799 Roush
.797 F. Jones
.767 Beckley
.767 Rice
.756 J. Sewell
Here's what I have, please let me know if this is accurate before I move them up top . . .
Rube Foster - 241-176 266 WS + batting (guesstimate 25 since Walter Johnson had 30 in a longer career)
Jose Mendez 195-133 257 WS
Dick Redding 230-177 267 WS
Joe Rogan - no estimate
Smokey Joe Williams - 377-300 (no WS estimate)
John Beckwith ~297 WS
Frank Grant - no estimate
Pete Hill - no estimate (would get ~350 WS)
Home Run Johnson - no estimate
John Henry Lloyd - ~490 WS
Ollie Marcelle ~220-235 WS
Bill Monroe ~340 WS
Dobie Moore ~300 WS (with a high peak)
Bruce Petway ~181 WS
Spotswood Poles ~332 WS
Ben Taylor ~326 WS
Jules Thomas - no estimate (I'd say a little lower than Poles)
Cristobal Torriente - ~374 WS
Louis Santop - no estimate
Please let me know if I missed anything in my detective work. As best I can tell all of these numbers are set to 154 game seasons, so they Win Shares numbers will need a 5% boost to be on par with those for their white counterparts.
"Dividing the batting and fielding WS up between 8 position players it works out to 8.8 WS/162 games and for pitchers it works out to 5.7 WS/220 IP. Or 8.6 for the 330 IP pitcher"
I'm going to use these numbers unless someone speaks up and tells me I shouldn't.
I'm not sure whether this confirms your decision or not (my first thought is that it does) to use a significantly lower replacement level for pitchers than for hitters. If anything, perhaps replacement level for pitchers may need to be even lower.
This article from today's Hardball Times:
http://www.hardballtimes.com/main/article/a-good-starter-is-hard-to-find/
suggests and provides some evidence (for the 21st century at least) what I have long suspected -- for pitchers much more than other positions, showing up (and not sucking) is much of the battle.
The conclusion:
[The replacement level for e]very position comes in between 60% and 70% except starting pitching. The replacement level for starting pitchers is around 40% [38%] -- just slightly more than half that of all other positions. This finding reinforces the commonly heard complaint that Win Shares undervalues starting pitchers."
In other words, if my starting leftfielder goes down, I can probably replace about 70% of his value with a fairly readily available backup, but if my starting pitcher goes down, then I might not be able to replace much more than a third.
I don't believe Chris ever did anything with them, Joe. I don't think it's possible with Grant, anyway.
I will try to get to this tonight OCF - I'm heading off on vacation this morning, and it kind of slipped my mind. Worst case, I'll print the relevant info before I leave work tonight, and do it on the plane - I assume the hotel have a way for me to hop on line with my laptop, even if it's just dialing up through a local #.
What would you suggest I use as my replacement level. I think it's already pretty low - 8.8 WS per 337 IP.
There's one important variable missing here, which is TIME.
For players just injured and missing today's game, there is one (very low) replacement level.
For players injured and about to miss several games (giving the team time to call up someone, or trade for someone, etc.) then there will be a higher average replacement level.
For players injured in early spring training and about to miss the whole season, they replacement level will be even higher.
For players retiring in October, the replacement level will be even higher, etc.
What all of these models fail to account for is that the overall level of the league declines when a player is injured.
Also, teams need to expend resources to find replacement player, so the longer the injury (or retirement) the more resources that can't be used elsewhere.
I don't think long-term replacement level approaches anything near average, as some have suggested.
Ruth is the new all-time leader in Pennants Added, joining Ty Cobb as the only players over 2.000.
Hornsby is a clear #2 at 2B to Eddie Collins, by WARP, and WSPA. He's actually behind Lajoie on WS, he only passes him with his incredible peak.
This is true, but the question is what value we are trying to evaluate -- a players' value to his team, or a players' value to the league as a whole. I think most of us think of value as the former. Babe Ruth is great because of what he did to help the Yankees wins pennants, not because he improved the overall quality of the American League. If that is the case, then the relevant replacement level is the quality player the Yankees are likely to get, not the inevitable scrub who would trickle up into a backup role for the Indians or Browns.
What would you suggest I use as my replacement level. I think it's already pretty low - 8.8 WS per 337 IP.
Of course, we'd need to see more information on how the 1930s replacement level compared to the 1930s, but assuming that they are similar, then using jimd's numbers for post 45, if an average pitcher on an average team (77-77 record) pitching 330 innings earns 8.1% of 231 win shares (77*3), then he would earn 18.7 Win Shares per year. If we assume that replacement level for pitchers is 38% of the average level, the we get 18.7*0.38, or 7.1 Win Shares per 330 innings.
Honus Wagner says "Ahem!" :-)
Thanks, Joe.
How about Pie Traynor?
The Lake Wobegon replacement-level. If replacement-level was average over any time period, then practically all teams would be above average.
This concept confuses expectation with reality. 20 years from now, your expectation at each position is average, but that's not the replacement level; you have to expend resources to get to average. If you don't expend the resources and just drift along, you become the 1918-1948 Phillies, or the Browns 1930-1953, etc.
If that is the case, then the relevant replacement level is the quality player the Yankees are likely to get, not the inevitable scrub who would trickle up into a backup role for the Indians or Browns.
A concept worth exploring and related to Brent's (I think it was his, but I could be misremembering) championship level.
But please don't call it "replacement level". It's different from what is usually meant by that term (which is a league-wide concept). And we have enough trouble communicating our thoughts sometimes without deliberately using the same name for different concepts.
I'm not sure I follow this. The replacement level of starting players is actually slightly ABOVE league average long-term...
20 years from now, your expectation at each position is average, but that's not the replacement level; you have to expend resources to get to average.
Resource expenditure is moving more towards an economic evaluation. Plus you have to spend resources to even be the old Phillies or Browns, including the resource of a roster spot, etc.
But please don't call it "replacement level". It's different from what is usually meant by that term (which is a league-wide concept). And we have enough trouble communicating our thoughts sometimes without deliberately using the same name for different concepts.
Well, this is exactly what some of us are arguing against. Measuring "above replacement level" by using the absolute worst player in a league to measure HOM worthiness. A better measure would be the level that the "average" player doing the replacing would be performing at.
However, those conditions are fictional. Not all teams are smart. Not all teams have the resources to be on the market for talent at all times. There are rules on when and how you can trade. Teams stick with below replacement level guys out of loyalty or inertia. Etc.
As a result, when the smart, resource-heavy teams need to replace a player (and they are the teams we care about if we are focusing on winning pennants) the market is structured such that they don't pay full price (in $ and talent) for their acquisition. As a consequence, they can usually obtain a player who is somewhat better than the best non-major leaguer at little or no cost or a player who is fairly good at a discounted cost. I have absolutely no idea the magnitude of this effect, and am fairly certain it fluctuates over time (due to different distributions of resources and savvy, different rules, etc.) but am certain that it exists at all places and times.
This can only be true if you DEFINE replacement level as an average. The definitions I've seen define replacement level as the minimum available. If you change the definition, then you're talking about a different concept, which is fine, but it's not "replacement level".
IOW, if you take the average value of all replacements 20 years from a given date, then that value will be slightly above league average. But that's not replacement level, that's average replacement value, and they're not the same thing.
I'm not sure that it would be "better", but it would be a different measure. Count as positive all value above average, count as zero all value below average. This just means that you are applying a higher standard. Zeroing the negative is important, because again you shouldn't penalize the guy for playing - making a living because he has value to some team even if not to the best teams - just because he doesn't meet the higher standard. It's kind of like time in the minors, usually just ignored, not actively penalized.
jimd :I'm not sure that it would be "better", but it would be a different measure. Count as positive all value above average, count as zero all value below average.
EEP! My head is spinning...
I believe KJOK meant "average replacement" here not "average". The point here is that due to fluctuations in farm system strength, who is on waivers, poor management, etc, the "replacement level" for a team at a given position can vary. Thus, there is almost guaranteed to be several "below replacement" players in the league at any given time. So, the hypothetical "replacement level" should not necessarily be defined by the worst players in the league, but a little bit higher than that.
Comparing to the absolute average instead of the replacement is another debate.
This is just another reason why I take value to be a zero-based phenomenon. Every player who takes the field has zero value until the first pitch is thrown and then their value goes up (not down) from there. A fielder makes an error, a pitcher gets rocked, a batter takes a donut, that's an absence of positive value.
The calculation of value against any other baseline just looks too hard, too conceptual, too much guesswork to me.
Van Haltren is higher because his pitching WS are included.
Would you be able to show Van Haltren's WS and Pen. Add. without the pitching, much like you did with Ward? It would give a bit more perspective for a player with a different career than is usual.
I believe KJOK meant "average replacement" here not "average".
Yes. But he also said "The replacement level of starting players is actually slightly ABOVE league average long-term", so to him they are close to interchangeable.
So, the hypothetical "replacement level" should not necessarily be defined by the worst players in the league, but a little bit higher than that.
No quarrel with that, but that's not what KJOK is saying.
sunnyday2:
whether it is really germane to the HoM I'm not sure.
Methinks any discussion about the way we evaluate players is germane.
Clearly, replacement value for the New York Yankees circa 1997-2004-2005 is a little different than it is for other clubs
This may be true, but it does not lead to a workable definition of value for a project such as the HOM. A team-by-team definition leads to evaluations where good players on bad teams are more valuable than good players on good teams simply because they play for bad teams, while using a league-wide replacement level in the same case would have preferred the good players on the good teams.
KJOK:
I'm not attacking your player rankings, please understand that. I think that peak value is more important (and valuable) than many here do; I just don't have a good handle on quantifying it. So incorporating "value over average" as part of a player evaluation system seems reasonable to me (just no penalty for "value below average"). I hope to get my old BP's out of the attic and look at Pennants Added more closely some weekend soon, to understand it better.
I also think that using WS or WARP to evaluate careers has problems, because much of that total value represents value under replacement, value awarded simply for playing time, not value above replacement (using the worst regular definition).
I also think that using WS or WARP totals to evaluate careers has problems
Yes. But he also said "The replacement level of starting players is actually slightly ABOVE league average long-term", so to him they are close to interchangeable.
So, the hypothetical "replacement level" should not necessarily be defined by the worst players in the league, but a little bit higher than that.
No quarrel with that, but that's not what KJOK is saying.
I apologize for being ambiguous, especially on a subject where you need to be very precise.
What I meant to say was the avearge replacement level for STARTING players IS slightly above ".500" OVER THE LONG TERM. In the short term, if you've got to replace your starting SS 10 minutes before gametime, the "average" replacement in this situation is obviously much less than a .500 player.
I think the basic difference in what I'm proposing "should" be the method is assigning 'value' is that "WARP" basically measures from the 25th man on the roster of the Arizona Diamondbacks while I would propose measuring from the average "10th man" (11th in AL I guess) with average calculated across the Red Sox/Cardinals/Yankees all the way down to the Brewers/Diamondbacks, etc.
I think I'd much rather use Major League average than AA average for any measurement....
Again, this doesn't make sense to me. Everybody who is playing a "below .500" player is ignoring an above .500 player who is out of work?
Or I'm not understanding at all how this is being calculated.
Again, this doesn't make sense to me. Everybody who is playing a "below .500" player is ignoring an above .500 player who is out of work?
It just means that the "average" starting player is around a .514 player. So, if a player retires, the most likely performance level of his replacement on the team the following year would be a player who is slightly above the major league average.
Do the players who are below the "average replacement value" get zero value or negative value?
So it is an "average" based replacement-value system, as opposed to a "minimum" based replacement-value system
Yes, because unless you're the Diamondbacks or the old Browns or Phillies, that is the more realistic replacement IMO.
Do the players who are below the "average replacement value" get zero value or negative value?
I know people get hung up on this 'negative value', but each game begins with both teams theorically having a 50% change of winning. So if a player over the course of the year was less than a 50% of league player, he would receive negative value for that year as he moved his team towards losses more than towards wins.
I confer with my stat guys, to decide whether to accept his terms. They tell me my only immediate replacement in a "minimum replacement" guy, but there are some free agents who might be available and who might be average (but don't come with crystal balls, and are asking for 3 year, $21 million contracts), a couple of guys in AA and AAA who certainly look like they could add to the team in future. Of course, they could all crash and burn too.
They run the numbers and tell me, "If you sign this guy, you'll be one game below average at that position for the next 15 years. If you don't sign him, you'll probably give that starting job to betwen three to six different guys, some of whom are only 12 years old today. If you add up their combined performance, they should give you about one game above average over the next 15 years."
Oh, I decide. Sign this guy for one game below average, or go into the unknown for an expected one game above average. I tell Player X to take his ball and go elsewhere, I'm not interested.
If you disagree with the above -- to show this idea in starker terms -- consider that Player X brings 24 of his friends with him. They all have their own crystal balls, and they ALL show they will perform exactly one game below average for the next 15 years. You own an expansion team with no one currently under contract. They all want 15 year guaranteed contracts with no-trade clauses. Do you sign them all? They all have "value," don't they?
The problem is, theoretically, if some idiot owner had allowed Lefty Grove to pitch for his team until the day Grove died, Lefty's value would have wound up deep in the negatives. It doesn't make any sense to me that the player should be penalized for this stupidity. If anybody should be given a negative, it should be the GM and owner.
Only a moron would sign them all, but that still doesn't mean that a below-average player doesn't have some value in the real world. Maybe a not a lot, but not negative either.
History is filled with teams that would have loved to have one of those "-1" guys to fill a hole and push them over the top.
I'm afraid I'm torn on this whole debate. I acknowledge that a low replacement level is good in showing that the "-1" guys are better than an above average september call-up or even John Paciorek. The lower baseline is also great for single-season debates where in-season durability is an issue (Ripken vs. Larkin, Mantle-62-MVP). In a single season, its logical to assume your replacement will be fairly low.
Still the higher baseline is nicer for measure "greatness" over a long career. This gives higher relative weight to peak seasons. Also, I agree with KJOK in that as time passes on your "next-year replacement" will regress fairly quickly back to the mean. So when comparing the 10-year player to the 15-year player, there are many advantages to using a higher baseline.
Now, if I could only figure out how to combine those two. :-) Somehow rewarding the peaks of a Barry Larkin or a Willie Stargell, yet acknowledging that each spent a good chunk of nearly every season on the DL. I'll just look at both numbers for now.
The guy who is "1-game below .500" has positive real value; he represents an upgrade for at least 1/3rd of the league, probably more. Nobody is likely to want him for his whole career, but he will play for some organization until they acquire a better player, and he will then move on to another organization. He will not be out of work until he retires, and he will never be near being the worst player at his position in the league. There is no question that he has positive real value.
The HOM dilemma is when such a player has an unusual career shape; no peak combined with unusual durability. The "1-game below .500" full-time player is worth about 16+ Win Shares or 5+ WARP a year, and 20 years of that is over 330 career Win Shares or 1000 career WARP, benchmarks that usually indicate HOM worthiness, at least to the career voter. If such a player manages to play 20-25 years while very rarely being above average, is he a HOMer?
After all, he had more hits and triples than Babe Ruth!
Sorry, had to be done.
Shouldn't Charlie Grimm's picture be next to that definition?
But what if he were still pitching today? And had started at age 12?
What if, in 2014, he retires at age 66, so instead of 216-216, he was 516-516? He's now the major league leader in wins (and losses), he's kept up that 95-99 ERA+ he had in his 40s, so when he retires his career number is exactly 100.
But, except for a few scattered years in the 1980s, he was never actually that good.
True, but his durability quotient is going through the roof. The career part of me would be taking notice of Mr. Hough when he became eligible.
You must be Registered and Logged In to post comments.
<< Back to main