|
Post by bleedthefreak on Mar 15, 2024 12:09:15 GMT -8
Bleedthefreak, it looks like there are two different sets of WAR numbers available at the BaseballProjection.com site. When I access the site via the url you provided above -- www.baseballprojection.com/war2/glossary.htm -- I get to the glossary page. If I navigate to players from the index on that page, I get to the stats you quote above. When I access the site via its own homepage -- www.baseballprojection.com -- and access what is advertised as "Wins Above Replacement, 1871-2009" link in the upper left hand corner of the page, I end up in a different index that leads to a different set of player stats. When I discover that the big banner "WAR and Victory Shares" is in fact a link and click on it, I get to the glossary page and the index that leads to the stats you have presented. So there are two different sets of WAR stats on the site. It happens that for Buddy Bell, the career WAR total is exactly the same in both systems -- 60.6 -- but that may be a rare accident. Do you (or anyone else) know the status of that other set of WAR values that are included on the site? Maybe those are WAR 1.0, as it was when BBRef picked it up in 2009 and began advancing it year by year and making minor tweaks? The baseball projection, 1871-2009 is Sean's original WAR calculation.
I think his plan is to take those old pages down and have them available as a csv download.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 15, 2024 12:36:34 GMT -8
Bleedthefreak, it looks like there are two different sets of WAR numbers available at the BaseballProjection.com site. When I access the site via the url you provided above -- www.baseballprojection.com/war2/glossary.htm -- I get to the glossary page. If I navigate to players from the index on that page, I get to the stats you quote above. When I access the site via its own homepage -- www.baseballprojection.com -- and access what is advertised as "Wins Above Replacement, 1871-2009" link in the upper left hand corner of the page, I end up in a different index that leads to a different set of player stats. When I discover that the big banner "WAR and Victory Shares" is in fact a link and click on it, I get to the glossary page and the index that leads to the stats you have presented. So there are two different sets of WAR stats on the site. It happens that for Buddy Bell, the career WAR total is exactly the same in both systems -- 60.6 -- but that may be a rare accident. Do you (or anyone else) know the status of that other set of WAR values that are included on the site? Maybe those are WAR 1.0, as it was when BBRef picked it up in 2009 and began advancing it year by year and making minor tweaks? "Wins Above Replacement, 1971-2009" is WAR 1.0, the other you described is WAR 2.0.
|
|
nate
Hall of Merit Voter
Posts: 27
|
Post by nate on Mar 16, 2024 8:30:46 GMT -8
I wouldn't want to claim authority to make any official decision, but I will happily weigh in. I don't think I like relying too hard on the 'excluded' players category - it's very easy to make the case that Japanese players were excluded unfairly before the establishment of the posting system, and if there is one class of players that is indubitably excluded it is those who played exclusively in the NPB. I'm honestly 50/50 whether Linares is more similar to an NPB player, or someone like Perucho Cepeda or Carlos Moran who had few direct contributions to US baseball (although in different ways) but played in the great North American baseball system. If I had to pick now it would be that it's more like the NPB situation - while the political situation is distinct, both are leagues that were distinguished on national lines, not discriminatory ones. That being said, I think it's also easy to fall back on the idea of 'North American' baseball, which is a clean way to continue to not consider pure NPB players while avoiding any strange edge cases for Caribbean ones.
Linares is a tough one. I wonder if he was ever discussed on the old BBTF site? The idea that the HOM is meant to enshrine the best North American players is something that was said often enough on the old site, but it never made its way into the constitution, apparently.
Not that this is the be all, end all, but... Moran would clearly be eligible for the HOF, since, while he mostly played in Cuba, he did play in the US as part of the pre-NgLs Cuban Stars team. I suspect Cepeda wouldn't be, but I'm not sure... he never played in the US, apparently because he didn't want to deal with the racism here at the time. (Of course, Puerto Rico is part of the US, but it's usually been treated separately, baseball-wise.) Linares clearly wouldn't be.
|
|
|
Post by bleedthefreak on Mar 16, 2024 10:05:54 GMT -8
I've done a sweep of about 1000 position players, anyone with 20 + career baseball-reference war and others that we reviewed for the BBF ranking game and some active guys.
Guys gaining > 5 wins or losing > 5.5.
2.0 1.0 Diff Name 79.5 55.6 23.9 Max Carey 56.1 38.7 17.4 Pie Traynor 39.1 26.1 13.0 Mike Scioscia 27.2 14.7 12.5 Johnny Edwards 91.0 79.1 11.9 Joe DiMaggio 49.7 38.0 11.7 Steve Garvey 60.4 49.5 10.9 Nellie Fox 103.6 92.8 10.8 Al Kaline 69.4 58.7 10.7 Darrell Evans 79.8 69.4 10.4 Tim Raines 58.4 48.0 10.4 Kiki Cuyler 89.5 79.2 10.3 Pete Rose 65.8 55.8 10.0 Freddie Freeman 34.4 24.7 9.7 Michael Young 37.2 27.5 9.7 Randy Winn 53.6 44.1 9.5 Rabbit maranville 81.3 71.8 9.5 Frankie Frisch 165.3 156.1 9.2 Willie Mays 54.0 44.8 9.2 Buster Posey 137.5 128.6 8.9 Stan Musial 69.4 60.5 8.9 Gary Sheffield 48.9 40.1 8.8 Dave Parker 48.6 39.9 8.7 George J Burns 37.9 29.2 8.7 Jayson Werth 81.7 73.1 8.6 Jim Thome 37.0 28.4 8.6 Terry Pendleton 38.7 30.3 8.4 Bobby Bonilla 62.6 54.2 8.4 Sam Rice 63.3 55.2 8.1 Joe Mauer 27.6 19.5 8.1 Joe Carter 119.2 111.2 8.1 Rickey Henderson 51.0 43.0 8.0 Tony Oliva 115.1 107.3 7.9 Frank Robinson 76.0 68.3 7.7 Al Simmons 29.1 21.4 7.7 Matt Kemp 22.7 15.2 7.5 Vic Saier 32.3 24.8 7.5 Mike Lowell 77.5 70.1 7.4 Gary Carter 106.9 99.7 7.2 Joe Morgan 67.5 60.4 7.1 Jim Edmonds 57.0 49.9 7.1 Dave Bancroft 23.5 16.5 7.0 Brad Ausmus 40.9 34.0 6.9 Elvis Andrus 149.8 143.0 6.8 Hank Aaron 51.0 44.2 6.8 Steve Finley 70.9 64.2 6.7 Dave Winfield 64.2 57.6 6.6 Enos Slaughter 47.1 40.6 6.5 Tim Salmon 77.0 70.5 6.5 Barry Larkin 60.2 53.8 6.5 Ron Cey 22.8 16.4 6.4 Bubbles Hargrave 23.4 17.0 6.4 Jimmy Johnston 26.7 20.4 6.3 Daniel Murphy 17.8 11.5 6.3 Shawon Dunston 48.5 42.3 6.3 Nelson Cruz 35.9 29.7 6.2 Javy Lopez 24.1 18.0 6.1 Angel Pagan 36.2 30.1 6.1 Joe Kuhel 39.5 33.4 6.1 Ken Caminiti 20.6 14.7 5.9 Ryan Howard 27.5 21.6 5.9 Yoenis Cespedes 20.3 14.4 5.9 Tony Armas 37.9 32.0 5.9 Corey Seager 63.4 57.6 5.8 Willie Stargell 33.6 27.8 5.8 Garry Templeton 36.6 30.9 5.8 Hunter Pence 83.1 77.4 5.7 robin Yount 24.5 18.8 5.7 Earl Battey 34.7 29.0 5.7 Red Rolfe 48.0 42.4 5.6 Lenny Dykstra 23.6 18.0 5.6 Cleon Jones 22.3 16.8 5.5 Melvin Upton 35.1 29.6 5.5 Marquis Grissom 38.6 33.2 5.4 Anthony Rendon 73.8 68.4 5.4 Carlton Fisk 28.2 22.8 5.4 Bret Boone 44.2 38.9 5.3 Paul O'Neill 31.5 26.2 5.3 Johnny Hopp 29.1 23.8 5.3 Prince Fielder 34.1 28.9 5.2 Matt Carpenter 25.4 20.2 5.2 Casey Stengel 51.5 46.4 5.1 Mark Grace 14.0 8.9 5.1 Don Kessinger 25.4 20.3 5.1 Bud Harrelson 30.5 25.4 5.1 Joe Rudi 58.1 53.1 5.0 Jack Clark 37.4 32.4 5.0 Edgar Renteria 20.1 15.1 5.0 Matt Lawton 42.1 37.1 5.0 Jay Bell
2.0 1.0 Diff Name 108.3 127.1 (18.8) Rogers Hornsby 76.9 93.5 (16.6) Adrian Beltre 33.0 48.0 (15.0) Wally Schang 29.5 44.3 (14.8) Brett Gardner 55.1 68.4 (13.3) Kenny Lofton 8.5 21.6 (13.1) Terry Kennedy 29.6 42.7 (13.1) Jorge Posada 58.9 71.9 (13.0) Rafael Palmeiro 104.6 117.5 (12.9) Alex Rodriguez 24.7 37.3 (12.6) Andrelton Simmons 72.0 83.8 (11.8) Ken Griffey Jr. 56.8 68.4 (11.6) Edgar Martinez 40.5 51.9 (11.4) Dustin Pedroia 61.9 72.2 (10.3) Harry Heilmann 26.3 36.6 (10.3) Bill Mazeroski 18.2 28.3 (10.1) Carlos Lee 12.6 22.6 (10.0) Charles Johnson 34.6 44.6 (10.0) Chuck Knoblauch 18.6 28.3 (9.7) Bob Meusel 20.5 30.1 (9.6) Trevor Story 29.5 39.1 (9.6) Rico Petrocelli 10.7 20.0 (9.3) Earl Sheely 26.3 35.5 (9.2) Kevin Kiermaier 24.8 33.8 (9.0) Tony Cuccinello 32.0 41.0 (8.9) Mark Belanger 29.0 37.9 (8.9) Starling Marte 16.1 24.8 (8.7) Ival Goodman 36.7 45.3 (8.6) Dixie Walker 29.9 38.5 (8.6) Lonnie Smith 24.4 33.0 (8.6) Salvador Perez 38.7 47.3 (8.6) Devon White 64.2 72.7 (8.5) Larry Walker 30.3 38.7 (8.4) Hack Wilson 83.0 91.4 (8.4) Wade Boggs 17.2 25.5 (8.3) Carlos Pena 21.1 29.3 (8.2) Mickey Tettleton 31.8 40.0 (8.2) Clyde Milan 26.8 35.0 (8.2) George H Burns 40.1 48.0 (7.9) Ralph Kiner 58.1 65.9 (7.8) Duke Snider 39.4 47.2 (7.7) Curtis Granderson 57.1 64.8 (7.7) Andre Dawson 38.1 45.8 (7.7) Edd Roush 27.3 35.0 (7.7) Brady Anderson 28.9 36.5 (7.6) George Scott 67.6 75.1 (7.5) Johnny Bench 19.9 27.4 (7.5) Buck Herzog 15.4 22.9 (7.5) Darren Daulton 38.1 45.6 (7.5) Omar Vizquel 53.0 60.4 (7.4) Zack Wheat 22.2 29.5 (7.3) Brian Roberts 40.8 48.1 (7.3) Heinie Manush 34.9 42.2 (7.3) Darryl Strawberry 48.4 55.6 (7.2) Chet Lemon 61.5 68.7 (7.2) Ivan Rodriguez 54.7 61.8 (7.1) Todd Helton 16.5 23.5 (7.0) Jack Wilson 27.4 34.4 (7.0) Travis Fryman 12.6 19.6 (7.0) Claudell Washington 17.2 24.2 (7.0) Gene Alley 41.1 48.0 (6.9) Bobby Veach 11.9 18.8 (6.9) Dick Schofield 31.8 38.7 (6.9) Juan Gonzalez 9.0 15.8 (6.8) Jody Davis 20.6 27.4 (6.8) Benito Santiago 44.7 51.4 (6.7) Bobby Doerr 50.2 56.8 (6.6) Larry Doby 14.4 20.9 (6.5) Mike Stanley 35.5 42.0 (6.5) Harlond Clift 20.6 27.1 (6.5) Lefty O'Doul 14.4 20.9 (6.5) Muddy Ruel 46.2 52.6 (6.4) Fred McGriff 18.0 24.4 (6.4) Carlos Gonzalez 29.2 35.6 (6.4) Tommy Holmes 15.1 21.3 (6.2) Mark Kotsay 13.6 19.8 (6.2) Willard Marshall 39.4 45.6 (6.2) Del Pratt 20.2 26.4 (6.2) Steve O'Neill 20.3 26.5 (6.2) John Mayberry 23.8 30.0 (6.2) JD Martinez 17.0 23.1 (6.1) Bobby Higginson 14.1 20.2 (6.1) Colby Rasmus 31.5 37.6 (6.1) Dick McAuliffe 14.6 20.6 (6.0) Melky Cabrera 45.6 51.6 (6.0) Earl Averill 19.7 25.7 (6.0) Steve Sax 62.0 67.9 (5.9) Graig Nettles 90.0 95.9 (5.9) Cal Ripken Jr. 22.2 28.1 (5.9) JJ Hardy 23.5 29.4 (5.9) Jim Piersall 11.9 17.8 (5.9) Jeff Burroughs 50.5 56.3 (5.8) Johnny Damon 17.0 22.8 (5.8) Michael Bourn 11.3 17.1 (5.8) Eddie Bressoud 14.1 19.9 (5.8) Jeromy Burnitz 16.3 22.1 (5.8) Gary Pettis 16.7 22.4 (5.7) Ken Oberkfell 60.6 66.3 (5.7) Buddy Bell 13.9 19.6 (5.7) Dick Hoblitzel 31.7 37.4 (5.7) Cy Williams 18.9 24.6 (5.7) Corey Koskie 35.5 41.2 (5.7) Jack Fournier 16.8 22.4 (5.6) Jim Gantner 16.5 22.1 (5.6) Austin Jackson 22.4 28.0 (5.6) Terry Steinbach 15.7 21.2 (5.5) Trot Nixon 33.0 38.5 (5.5) Lorenzo Cain 16.9 22.4 (5.5) Spud Davis 19.2 24.7 (5.5) Josh Reddick 41.2 46.7 (5.5) Chuck Klein 13.1 18.6 (5.5) Grady Hatton 32.8 38.3 (5.5) Sid Gordon 44.7 50.2 (5.5) Fred Lynn
|
|
|
Post by bleedthefreak on Mar 16, 2024 10:12:15 GMT -8
Guys with votes in the 2024 election: 10.4 Kiki Cuyler 8.4 Sam Rice 7.1 Dave Bancroft 6.5 Ron Cey 5.0 Jack Clark 3.9 Phil Rizzuto 3.3 Jose Cruz 2.6 Tony Perez 1.9 Willie Davis 1.8 Gene Tenace 1.7 Cesar Cedeno 1.3 Kirby Puckett 0.4 David Wright (0.2) Gavvy Cravath (0.3) Norm Cash (0.6) Dave Concepcion (0.6) Bob Johnson (0.9) John Olerud (1.2) Jason Giambi (1.3) Albert Belle (1.4) Bernie Williams (1.7) Thurman Munson (2.8) Toby Harrah (3.2) David Ortiz (3.4) Jason Kendall (3.4) Jim Fregosi (3.6) Luis Aparicio (3.7) Brian Downing (4.3) Jim Sundberg (4.4) Brian Giles (4.4) Tony Phillips (4.4) Harry Hooper (4.7) Bert Campaneris (4.7) Sal Bando (4.7) Lance Parrish (5.1) Bob Elliott (5.3) Robin Ventura (5.7) Buddy Bell (6.4) Fred McGriff (7.2) Chet Lemon (13.1) Jorge Posada (15.0) Wally Schang
|
|
nate
Hall of Merit Voter
Posts: 27
|
Post by nate on Mar 16, 2024 10:59:51 GMT -8
Pie Traynor is an interesting name. He was a guy with a good defensive rep in his time, but TZ has always been down on him. His batting value looks pretty similar at BRef and in WAR 2.0, but the latter is much higher on his defense (-32 runs for TZ vs 121 for WAR 2.0). I don't have the numbers at hand, but I remember DRA also being a lot higher on Traynor than TZ, so I guess the question is... How good was he actually defensively and why do various defense metrics have such different opinions about him? If you just look at WAR 2.0, he's no slam dunk, but he's the kind of borderline guy who could get some votes, whereas he was on no ballots last election.
|
|
|
Post by bleedthefreak on Mar 16, 2024 11:45:56 GMT -8
Pie Traynor is an interesting name. He was a guy with a good defensive rep in his time, but TZ has always been down on him. His batting value looks pretty similar at BRef and in WAR 2.0, but the latter is much higher on his defense (-32 runs for TZ vs 121 for WAR 2.0). I don't have the numbers at hand, but I remember DRA also being a lot higher on Traynor than TZ, so I guess the question is... How good was he actually defensively and why do various defense metrics have such different opinions about him? If you just look at WAR 2.0, he's no slam dunk, but he's the kind of borderline guy who could get some votes, whereas he was on no ballots last election. He's at 47.2 gWar, +9.4 defensive wins.
|
|
|
Post by chriscobb on Mar 16, 2024 15:14:17 GMT -8
So, what are folks thinking of doing with WAR 2.0?
Dr. Chaleeko has suggested that "'2.0' should not supplant current WAR but should complement it."
I am considering the possibility of making it a third component in my system, equally weighted with BBRef and Fangraphs WAR for post-1910 position players. I am getting a sense for what it is doing with position players, but I'm less clear about what is happening with pitchers, aside from the ways in which the assessment of their fielding support has changed. I'd like to know why pitchers' batting WAR seems to be consistently different, whereas batting WAR for position players is the element of WAR 2.0 that is least changed from WAR 1.0/BBREf WAR. Given that it is an RA/9-based assessment, WAR 2.0 theoretically compatible with BBRef WAR for pitchers in ways that Fangraphs FIP approach is not (and I am not including Fangraphs pitching WAR in my system), and I'd rather have a system for pitchers that isn't based entirely on a single metric, but I don't think I know enough to take that step yet.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 16, 2024 15:45:33 GMT -8
So now Ron Cey and Sam Rice look like guys we should have elected, according to WAR 2.0.
|
|
|
Post by chriscobb on Mar 16, 2024 16:56:58 GMT -8
"So now Ron Cey and Sam Rice look like guys we should have elected, according to WAR 2.0."
Probably yes, according to WAR 2.0, and Dave Bancroft, probably more so than Sam Rice, even, who is such a career compiler that even with 62.4 WAR 2.0, his peak is not going to be that impressive. With respect to Cey, we haven't elected Sal Bando on the strength of 61.5 BBRef WAR in 2019 games, so I am not sure it is the case that we should have elected Ron Cey on the strength of 60.2 WAR 2.0 in 2073 games, unless we should already have elected Bando, too, only to reconsider now. Kiki Cuyler is probably on the "should have elected, according to WAR 2.0" list, too.
|
|
nate
Hall of Merit Voter
Posts: 27
|
Post by nate on Mar 16, 2024 20:57:44 GMT -8
Pie Traynor is an interesting name. He was a guy with a good defensive rep in his time, but TZ has always been down on him. His batting value looks pretty similar at BRef and in WAR 2.0, but the latter is much higher on his defense (-32 runs for TZ vs 121 for WAR 2.0). I don't have the numbers at hand, but I remember DRA also being a lot higher on Traynor than TZ, so I guess the question is... How good was he actually defensively and why do various defense metrics have such different opinions about him? If you just look at WAR 2.0, he's no slam dunk, but he's the kind of borderline guy who could get some votes, whereas he was on no ballots last election. He's at 47.2 gWar, +9.4 defensive wins.
Ah, so certainly higher than bWAR, but not as high as 2.0. Hmm. I think he's a guy who could appear on the back half of some ballots, but even if a voter is only considering 2.0 I'm not sure he's quite strong enough to get into an elect me position. His peak (32 WAR for 5 year peak, and 1927 as a MVPish type year) and career are both pretty good in 2.0, but I'm not sure they're quite good enough.
|
|
nate
Hall of Merit Voter
Posts: 27
|
Post by nate on Mar 16, 2024 21:14:01 GMT -8
So, what are folks thinking of doing with WAR 2.0? Dr. Chaleeko has suggested that "'2.0' should not supplant current WAR but should complement it." I am considering the possibility of making it a third component in my system, equally weighted with BBRef and Fangraphs WAR for post-1910 position players. I am getting a sense for what it is doing with position players, but I'm less clear about what is happening with pitchers, aside from the ways in which the assessment of their fielding support has changed. I'd like to know why pitchers' batting WAR seems to be consistently different, whereas batting WAR for position players is the element of WAR 2.0 that is least changed from WAR 1.0/BBREf WAR. Given that it is an RA/9-based assessment, WAR 2.0 theoretically compatible with BBRef WAR for pitchers in ways that Fangraphs FIP approach is not (and I am not including Fangraphs pitching WAR in my system), and I'd rather have a system for pitchers that isn't based entirely on a single metric, but I don't think I know enough to take that step yet.
I'm considering adding it to my system. Last election I only used bWAR and fWAR. I was already considering adding Tom Thress' eWORL and now I'm considering adding this too. Right now I'm still making sure I have everyone's WAR values scaled to 162 games (using the formula ericj posted earlier in the thread), so I don't have to make a decision quite yet. Neither eWORL or WAR 2.0 go all the way back to 1871, so I'd probably only use either one for players whose careers are fully captured. I wouldn't mind seeing some more info or more opinions on 2.0 though.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 17, 2024 5:00:38 GMT -8
Dawson, Palmeiro, Lofton, and Averill are likely mistakes, according to 2.0. And Edgar and Helton are iffy. Those downgrades actually align with my feelings on those players. But Duke Snider? Doesn't seem like a borderline out guy to me, like 2.0 says. Zack Wheat, Larry Doby? Jorge Posada finally gets "credit" for his awful defense and is a laughable candidate according to 2.0.
Some of these new numbers feel...very weird, I will say.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 17, 2024 5:33:36 GMT -8
I'm also considering adding it, but not at the same weight as fWAR or 1.0, at least until I understand it a little better.
Tango says Naive WAR is scaled to team wins. Is this true for the numbers we've been quoting, or is that for WAR_Rec or whatever the column is called (the "reconciled" version)? If 2.0 is going down that path, I won't be incorporating it into my rankings. But it's hard for me to decipher whether that's baked in or not. Negative loss shares also just bugs the conceptual shit out of me, but required to make the math work out the way they've set it up.
I look at Max Carey, and he now has a WAA higher than his WAR from 1.0!! And the fact that he's 74 WAR/56 WAA in over 10,000 PAs - like, is that possible? Almost every win he ever added was above average?? His rookie season, 1912 - 5.8 WAR and 4.5 WAA in *690* PAs. An average player was only 1.1 WAR (per 600) that year, I guess?? I'm having trouble making sense of this stuff right now. Does 2.0 have data on baserunning during Carey's early career yet, or still not?
I think it is safe to say plays+ is not a fully baked stat yet, Sean admits some of the values are batshit crazy in the glossary, it's not adjusted for anything (batter handedness, ballpark, GB v FB, etc.), yet it is a component in the defensive ratings.
Somebody help me out with some of these things. Am I taking crazy pills and 2.0 is a pure advancement? Or does it seem like a bit of an unfinished product at the moment, like this is the beta testing phase?
|
|
|
Post by bleedthefreak on Mar 17, 2024 6:17:57 GMT -8
I'm also considering adding it, but not at the same weight as fWAR or 1.0, at least until I understand it a little better. Tango says Naive WAR is scaled to team wins. Is this true for the numbers we've been quoting, or is that for WAR_Rec or whatever the column is called (the "reconciled" version)? If 2.0 is going down that path, I won't be incorporating it into my rankings. But it's hard for me to decipher whether that's baked in or not. Negative loss shares also just bugs the conceptual shit out of me, but required to make the math work out the way they've set it up. I look at Max Carey, and he now has a WAA higher than his WAR from 1.0!! And the fact that he's 74 WAR/56 WAA in over 10,000 PAs - like, is that possible? Almost every win he ever added was above average?? His rookie season, 1912 - 5.8 WAR and 4.5 WAA in *690* PAs. An average player was only 1.1 WAR (per 600) that year, I guess?? I'm having trouble making sense of this stuff right now. Does 2.0 have data on baserunning during Carey's early career yet, or still not? I think it is safe to say plays+ is not a fully baked stat yet, Sean admits some of the values are batshit crazy in the glossary, it's not adjusted for anything (batter handedness, ballpark, GB v FB, etc.), yet it is a component in the defensive ratings. Somebody help me out with some of these things. Am I taking crazy pills and 2.0 is a pure advancement? Or does it seem like a bit of an unfinished product at the moment, like this is the beta testing phase? I'm under the impression that it's a finished product.
I joined the BJOL site to see the discussion on this topic and otherwise, but haven't commented yet.
It's free to sign-up, put I can ask questions for the group if desired too.
Sean openly speaks about 2.0.
|
|
|
Post by bleedthefreak on Mar 17, 2024 6:27:30 GMT -8
Dawson, Palmeiro, Lofton, and Averill are likely mistakes, according to 2.0. And Edgar and Helton are iffy. Those downgrades actually align with my feelings on those players. But Duke Snider? Doesn't seem like a borderline out guy to me, like 2.0 says. Zack Wheat, Larry Doby? Jorge Posada finally gets "credit" for his awful defense and is a laughable candidate according to 2.0. Some of these new numbers feel...very weird, I will say. Dawson gets to ~63 wins if you take out his late career below replacement years + extend the 1981 season to 162 games. Edgar and anyone that DHs I think is hurt by the position penalty and has upside.
Snider is a surprise, Kiko's leveraging of retrosheet data is a big fan of his D, he's at ~77 wins at his site.
Doby falls by ~5 wins, with NLE credit, might just look more marginal, the 1950s aren't loaded with candidates, I think he should still be ok.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 17, 2024 7:33:47 GMT -8
Wait, Rally isn't Sean? I've been misapprehending that for a long time. Or, is it two different Seans?? Lol. Is Rally Sean Smith?
|
|
ericj
New Member
Posts: 7
|
Post by ericj on Mar 17, 2024 8:00:22 GMT -8
Surprised to see the DH penalty get steeper in WAR 2.0; most of the conversation I'd seen about it tended toward it being too steep. (5 runs below 1B always seemed roughly correct to me, FWIW.)
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 17, 2024 8:03:59 GMT -8
Having read the thread about Naive WAR (I agree we should call it this, as it does not seem intended as an evolution of bbref WAR, and mainly for reasons I'm about to discuss about defense), I would be willing to bet most of the largest discrepancies come down to defense and baserunning. Defense will be easier to explain after reading this: www.baseballprojection.com/defense/home.htmDefense is going to vary pretty highly for a lot of players compared to the systems used by B-R and FG. The reason is that Naive WAR uses a naive defensive engine so that players can be directly comparable for the full PBP era (that is, back to 1912). The entire defensive engine in Naive WAR is based on team DER, apportioned to individual players. There are no adjustments for difficulty of the play. There are *no park adjustments.* Infield popups are counted (equal weight as any other play made on the field). All of the other adjustments built into the more advanced defensive engines are gone. The only considerations are batter handedness and type of hit (single vs double vs triple). I think I would tend to trust the defensive values in Naive WAR the least of any system out there. Unquestionably so for players from 1950 onwards, and double unquestionably for modern times. But that may just be me. The part I still have trouble with is baserunning. Rally says the differences may be attributable, at least partly, to the fact that he takes outs into account (meaning, did the extra base get taken with 0 outs or 1 out or 2 outs in the inning). I guess I don't have a big problem with that approach theoretically. But I do wonder how much baserunning data is still incomplete. To answer my earlier question: it looks like WAR_Rec is reconciled to team wins, WAR is not. So you can have your pick of whether you want to do that.
|
|
|
Post by bleedthefreak on Mar 17, 2024 9:14:48 GMT -8
Wait, Rally isn't Sean? I've been misapprehending that for a long time. Or, is it two different Seans?? Lol. Is Rally Sean Smith? Yes
|
|
|
Post by herrdoktorchaleeko on Mar 17, 2024 15:41:24 GMT -8
Based on what I’m seeing in our discussion and Sean’s explanation, while Naive WAR may be a finished product, I’m unsure how “reliable” it is. I mean that not in a statistical sense but in terms of nitty-gritty can I trust this? I’m dubious so far.
I’ve been with you guys (on and off but mostly on) for the better part of 20 years. I fear I’ve become an old fart right in front of your eyes: “I don’t trust this new-fangled mathy stat! And get off my lawn!” I’m questioning myself because the discussion of player value is coming full circle. Win Shares tied everything to wins IRL. WAR went totally opposite. Then came WPA and feathers ruffled. Thress sort of bridged the WAR/WOA/IRL questions with his work. And now Naive WAR is just about full circle.
I jumped off the IRL tieback 10-15 years ago. I primarily value an understanding of the player without interference from the many contexts that influence his playing record. I’m feeling as though this version of WAR may be allowing some of that context to creep back in. Or it is introducing new uncertainties as has been mentioned upthread.
So I’m feeling very hesitant to jump on the Naive WAR bandwagon.
THEN comes the question of MLE compatibility….
|
|
|
Post by chriscobb on Mar 17, 2024 19:15:13 GMT -8
I am generally with Dr. Chaleeko in being skeptical of the benefits of "allowing some . . . context to creep back in." My distrust of context-based measures is one reason why I haven't adopted player won-lost records to my system, although its treatment of fielding value and of replacement level are equally large factors.
It’s probably worth inventorying the contextual elements of naive WAR/WAR 2.0 and examining them individually. Here are the ones I am aware of:
• RISP – performance with runners in scoring position is included in batting value. This factor seems generally to be small but swings some players a couple of wins positive or negative. This contextual factor is broken out as its own item, so it could be kept or removed at the user’s discretion.
• Baserunning – According to kcgard2 above, “Rally says the differences may be attributable, at least partly, to the fact that he takes outs into account (meaning, did the extra base get taken with 0 outs or 1 out or 2 outs in the inning).” This element is baked in to BSR, it seems.
• Fielding support – pitcher fielding support is based on defensive efficiency on balls in play in the pitcher’s games, not on the team’s average defensive efficiency. Presumably there is some credit to the pitcher for helping the defense, but this isn’t shown.
• Fielding analysis – using defensive efficiency would seem to allow for the uncalculated influence of certain contextual factors that more granular fielding systems adjust for.
• Reconciliation to team W-L record – this inserts a contextual factor, although since one can use the “unreconciled” WAR, there is no obligation to include this contextual element.
• Pitching leverage – this is a contextual element that has long been included in BBRef WAR, so one might not even remark on it being here, but it is a contextual element. However, BBRef WAR rebalances league WAA to zero to account for leverage in its WAAadj. Stat: “For relief pitchers, we multiply WAA by (1+gmLI)/2. This is done in recognition of the added importance of high leverage. WAAadj is the additional value of this leverage adjustment. Also, the manner in which this and the WAA calculations are performed cause the league total WAA to move away from zero, so we also do an operation to recenter the entire league. The recentering forces the league sum to 0 which is as it should be for Wins Above Average. So for the league as a whole, WAA+WAAadj will equal zero and WAR = WAA + WAAadj + Replacement value.” As far as I can tell, WAR 2.0/Naïve WAR does not include the step of recentering the entire league, unless this is included somehow in the “Role Adj.” I am not sure whether WAR 2.0 and BBRef WAR differ here, although it is clear that WAR 2.0’s methods account for the “opener,” where BBRef WAR does not appear to do so.
Of these, I am most dubious about the elements that can be disregarded or removed by the user: RISP and reconciliation with team W-L record. With respect to baserunning, there is an element of opportunity, but judging opportunities is in fact an important element of baserunning skill. When is it optimal to take the extra base? How “smart” is the player at assessing risk vs. reward in the in-game situation? Without having reached a definite conclusion, I can see a clear argument for including situational information in an assessment of baserunning, but I don’t know that I’ve considered all of the implications yet.
With respect to fielding, it’s been pretty well demonstrated that team defensive efficiency can vary considerably from game to game and that a team average efficiency may not accurately represent the defensive support that individual pitchers actually received, so here the additional context seems to add accuracy. On the other hand, there’s still the question of how the pitcher’s contribution to fielder efficiency can be disentangled from the fielder’s support of pitcher effectiveness on bip. I’d like to know more about how WAR 2.0 / Naïve WAR handles that before making a call on this. With respect to using defensive efficiency as such as the basis for fielding assessment, I think that it makes sense to have a fielding measure derived from defensive efficiency as part of a fielding assessment denominated in runs that are compatible with the values generated by other fielding systems. Particularly with fielding values derived from game information that doesn’t support a multi-faceted analysis of each batted ball, I see a defensive-efficiency assessment as a useful complement to a TZ or DRA approach.
With respect to pitching leverage, this is already a part of BBRef WAR and the shift to a more flexible and exact way of establishing leverage from role is needed for the contemporary game. I’d like to know if there is a rebalancing step or not, though.
Are there other contextual measures that I haven’t included? What do others think about the way in which these factors are being handled in the system?
|
|
nate
Hall of Merit Voter
Posts: 27
|
Post by nate on Mar 17, 2024 22:09:16 GMT -8
Former longtime voter here, returning after a long sojourn (I have to admit the BBTF implosion kind of brought my attention back, so...thanks, Jim?)! As those who were here when I last voted in...(checks watch)...the 2012 election will perhaps recall, I was an extreme peak voter, and so I pretty much remain, if slightly less so than when I began. And while I got re-intrigued in time for the 2024 election, I quickly realized that stats have evolved a great deal in the years since my last participation, so it took me some time to convert/develop a new system, but I think I've managed to wrangle one into a respectable enough place to start actually voting again this time around. I was looking at some MLEs, when I was reminded of the case of Charlie Smith. As an extreme peak voter, you might want to look at his numbers, if you haven't already. He only played 7 years, but his peak was quite good.
|
|
alex02
Hall of Merit Voter
Posts: 39
|
Post by alex02 on Mar 18, 2024 13:58:25 GMT -8
This isn't an argument for or against WAR 2.0, but one of my first impressions is that its results seem more in line with common wisdom among non-stat nerds than other existing WARs.
Pie Traynor as one of the best 3Bs of the early 20th century, Joe Dimaggio as an all-time great even before war credit, Steve Garvey closer to the Hall line, a bunch of dubious HOFers (Kiki Cuyler, Sam Rice, Nellie Fox, Rabbit Maranville) looking more credible, Buster Posey as an easy choice, Adrian Beltre as a great but not necessarily an inner-circle great, Kenny Lofton a good player but below the Hall line, Brett Gardner down, Ryan Howard up, etc. I'm not sure if this is about the RISP component or what, but all of this would have the typical non-stathead nodding along.
But of course there are exceptions (Ken Griffey Jr. being the biggest), and it's entirely possible I've imagined this pattern altogether.
|
|
|
Post by herrdoktorchaleeko on Mar 18, 2024 17:16:34 GMT -8
I think another layer of context we aren’t considering yet is external to Naive WAR, itself. Alex02 was headed there. Namely that seasons prior to the Live Ball era have some very sparse pickings. Harry Hooper’s arm demonstrates this well. No one with an assists record anywhere close to his has. Anywhere near as bad an arm rating as nWAR is giving him (-5). Why? Some possible explanations: 1) Hooper was actually a meh thrower 2) nWAR isn’t accounting for base runner holds accurately 3) Most of Hooper’s best throwing seasons occurred during years when PBP in retrosheet is still kinda sparse. 4) A combo of these
Re #3, Hooper’s seasons prior to WWI all have about ⅓ or less the total opportunities as the 1920s years when Hoop was an older player.
As to #1, his arm is strongly reputed, but Carey and Speaker both kicked butt in nWAR throwing runs, and they have comparably outstanding assists records and reputations (tho mostly in CF). Sam Rice, whose assist record is about half as good as Hooper’s has +10 arm runs.
I’m unsure how Rally is measuring arm runs beyond using PBP, so it’s all a little unclear to me. But Hooper really stuck out for me.
|
|
|
Post by chriscobb on Mar 18, 2024 18:22:24 GMT -8
Now that you point it out, the gap between Hooper and Speaker is staggering.
Pitching records as well as fielding records from the 1910s appear to be quite incomplete. Walter Johnson is a good example. My initial impression is that AL records are thinner than AL records from the decade, but I haven't been systematic in my searches.
It looks like the general take-away is that although nWAR/WAR2.0 includes seasons beginning with 1911, its data set isn't complete enough yet for its pre-1920 numbers to be useful for evaluating players.
|
|
|
Post by bleedthefreak on Mar 18, 2024 20:01:03 GMT -8
Now that you point it out, the gap between Hooper and Speaker is staggering. Pitching records as well as fielding records from the 1910s appear to be quite incomplete. Walter Johnson is a good example. My initial impression is that AL records are thinner than AL records from the decade, but I haven't been systematic in my searches. It looks like the general take-away is that although nWAR/WAR2.0 includes seasons beginning with 1911, its data set isn't complete enough yet for its pre-1920 numbers to be useful for evaluating players. I'll ask Rally what games are missing...
|
|
|
Post by bleedthefreak on Mar 19, 2024 7:17:47 GMT -8
Now that you point it out, the gap between Hooper and Speaker is staggering. Pitching records as well as fielding records from the 1910s appear to be quite incomplete. Walter Johnson is a good example. My initial impression is that AL records are thinner than AL records from the decade, but I haven't been systematic in my searches. It looks like the general take-away is that although nWAR/WAR2.0 includes seasons beginning with 1911, its data set isn't complete enough yet for its pre-1920 numbers to be useful for evaluating players. Missing games link here: www.retrosheet.org/wanted/index.htmlI'm going to download the years into a spreadsheet and can post a summary. It's not consistent team by team, year by year.
EDIT: Rally mentions that all of 1912 is available, while the Retrosheet page hasn't been updated and is a gap...checking to see if I can reconcile the gap...
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Mar 19, 2024 12:32:57 GMT -8
OK at least I discovered the discrepancy with WAA. You should not look at the column called WAA, because it does not represent WAA at all (at least not as anyone understands the term in any WAR framework). In Naive WAR Rally has called WAA Wins+. But still keeps a column called WAA because it is not possible for someone to invent a WAR implementation without creating uselessly confusing terms and contradictions to what everyone else does.
So Carey in 1912 has 5.8 WAR and 3.8 Wins+. That at least puts the average player in the ballpark of 2 WAR per 600 PAs. 1.8 WAR per 600 is close enough (unless that pattern is true across all/most/much of baseball history, in which case if you want to use nWAA then you'll have to make mental or real adjustments to the fact that the "average" line is set at a different spot than other WARs, which are more like 2.2 per 600).
|
|
|
Post by herrdoktorchaleeko on Mar 19, 2024 13:02:00 GMT -8
As we await more word on nWAR, I’ll circle back to an earlier topic. Thank you for clarification on Omar Linares’ status. I asked b/c the MLE I’ve run puts him near/at the borderline at 3B. That may be more true for me than others because I use a segregation adjustment that pushes a few pre Jackie guys down a bit in my rankings. I’m in the midst of working up what Castro Cubans I have sufficient data for. I also did El Duque, but he wasn’t coming out high enough to merit my asking about him. This MIGHT become a more interesting question years from now when Céspedes and Jose Abreu find their way onto the eligible list. But for now, it’s not a pants-on-fire thing.
|
|