MVP '99

[patsen]

One more thing, how do these Total Conversion mods handle retired/retiring players? I assume anyone with stats in 1999 makes the set, but BBRef has a list of all players who ended their major league in 1998 available (http://sandbox.baseball-reference.com/leagues/MLB/1998-finalyear.shtml).

Paul Molitor, Dennis Eckersley and Cecil Fielder seem to be the best of the retirees. Do we keep them in the game in FA, or remove them from the listing entirely?

The opening roster/free agent file might be interesting too. What date of the season are we targetting? April 4th seems to be the first day of the season...

You can look at http://sandbox.baseball-reference.com/leag...nsactions.shtml to see the April moves. We'd need to pick a point in time, and try to apply the rosters as of then.

[Borgi]

Had some good times making the AL rosters for the 99. Hope one day they will be used, that would be cool to re-up this

[patsen]

Sorry this took so much time, but I've been busy all month. To create 1999 rosters, I'm using 1996-1998 stats, and a modified version of Marcels to run these projections. These are still preliminary, but they should give an idea of what can come out. We can then cross-reference them with batting splits (which I have for the majors over 96-99), and create the ratings. I'll have to find out how to assign stars, but given the future can be seen, we can cheat a bit.

I made a quick sample of the '99 Yankees. I'm only using MLB-AAA-AA stats from 1996-1998, and I forgot to apply the aging modifier, so the older players will be penalized, and the younger players will improve. I also plan to use these for hitting stats for pitchers, but I'll need to use different regression lines (hitters to hitters, pitchers to pitchers).

'99 Yankees		  Ag   G  PA  AB   H 2B 3B HR  BB HB  SO  SB CS  Avg  OBP  Slg wOBA (Rel)

Bernie Williams	  30 123 551 482 154 29  5 22  68  1  75  14  7 .320 .405 .537 .408 (.86)

Paul O'Neill		 36 138 593 525 162 35  1 19  66  2  86   9  3 .309 .388 .488 .383 (.87)

Tino Martinez		31 133 566 504 144 27  1 27  59  4  74   3  1 .286 .366 .504 .377 (.87)

Chili Davis		  39  87 348 301  84 15  0 14  46  1  56   3  2 .279 .376 .468 .371 (.81)

Chuck Knoblauch	  30 135 620 533 154 25  7 12  72 15  70  36 10 .289 .389 .430 .367 (.88)

Darryl Strawberry	37 109 380 332  84 16  2 20  45  3  86   8  6 .253 .347 .494 .363 (.70)

Derek Jeter		  25 138 620 559 171 25  6 13  54  7 104  20  7 .306 .374 .442 .362 (.88)

Tim Raines		   39 112 423 366 105 19  2  8  54  2  55  10  4 .287 .381 .415 .357 (.78)

Jim Leyritz		  35 124 406 350  95 15  0 12  48  8  79   3  1 .271 .372 .417 .354 (.78)

Jeff Manto		   34 114 389 340  86 17  1 18  46  3  80   3  3 .253 .347 .468 .355 (.76)

Scott Brosius		32 136 548 492 133 27  1 17  49  7  95   9  5 .270 .345 .433 .343 (.85)

Jorge Posada		 27 116 425 372  95 22  1 13  51  2  85   2  2 .255 .348 .425 .342 (.78)

Shane Spencer		27 121 464 418 104 26  1 21  43  3  90   2  3 .249 .323 .467 .341 (.84)

Chad Curtis		  30 140 510 444 115 23  1 13  61  5  77  16  7 .259 .355 .403 .340 (.84)

Ricky Ledee		  25 124 455 411 106 23  2 16  41  3 116   6  3 .258 .330 .440 .337 (.80)

Dale Sveum		   35  93 264 242  62 15  1  9  21  1  55   2  2 .256 .318 .438 .329 (.71)

Tony Tarasco		 28 138 405 363  90 18  1 12  41  2  65   5  4 .248 .328 .402 .324 (.74)

Joe Girardi		  34 103 378 350  96 18  3  4  25  3  50   5  4 .274 .328 .377 .315 (.79)

Homer Bush		   26  91 283 264  69 13  1  4  17  2  61   7  5 .261 .311 .364 .301 (.69)

D'Angelo Jimenez	 21 115 466 422  97 19  3  8  42  3  84   6  6 .230 .305 .346 .293 (.71)

Clay Bellinger	   30 121 450 418  98 24  2  9  28  5  85   5  3 .234 .291 .366 .291 (.82)

Luis Sojo			34 103 310 291  74 11  1  3  18  1  30   4  2 .254 .300 .330 .283 (.70)

I expect to make this a total minors roster, mostly because we can. If this works, then we can make future total conversion mods (or past) total minors as well. So, I'll include A ball, and Short-season A for more stats to base the ratings off of. I assume you have your algorithms to create the rest of the MVP ratings from stats, so I'll likely use those to automate the job. We'll see what happens.

Also, we have Baseball America's top 100 prospects for 1999. All of them, and where they started the season are on this page: http://web.archive.org/web/20000304134612/...wherestart.html

[patsen]

I fixed a few things in the spreadsheet, and ran it on the 1999 Padres. As you can see, I dug a bit deeper, and included most of the AAA Las Vegas starting lineup as well. I probably included late-season acquisitions, but when we run it on the league, we'll do each player once, then assign them to rosters, so it's not an issue.

'99 Padres		   Ag   G  PA  AB   H 2B 3B HR  BB HB  SO  SB CS  Avg  OBP  Slg wOBA (Rel)

Tony Gwynn		   39 124 512 473 154 33  1 12  37  2  30   8  3 .326 .377 .476 .373 (.85)

Reggie Sanders	   31 126 508 450 118 22  4 17  52  5 122  19  9 .262 .344 .442 .345 (.82)

Wally Joyner		 37 129 497 443 128 27  1 11  52  2  58   4  3 .289 .366 .429 .353 (.84)

Jim Leyritz		  35 124 406 351  94 15  0 11  47  8  80   3  1 .268 .367 .405 .348 (.78)

Dave Magadan		 36 108 293 256  71 13  1  5  36  1  39   2  2 .277 .369 .395 .344 (.68)

John Vander Wal	  33 179 296 264  70 17  2  8  31  1  66   3  2 .265 .345 .436 .343 (.64)

Quilvio Veras		28 132 565 489 129 22  2  6  71  5  77  22 10 .264 .363 .354 .329 (.85)

Aaron Guiel		  26  96 346 310  78 20  3  8  29  7  72   6  4 .252 .329 .413 .328 (.79)

Greg Myers		   33 106 317 291  77 17  1  7  25  1  52   2  1 .265 .325 .402 .321 (.70)

George Arias		 27 135 490 455 116 27  2 16  31  4  96   2  2 .255 .308 .429 .320 (.84)

Phil Nevin		   28 106 359 325  79 15  1 13  30  3  85   2  1 .243 .312 .415 .319 (.77)

Chris Gomez		  28 139 511 456 118 24  2  6  50  5  90   4  4 .259 .339 .360 .316 (.84)

Ed Giovanola		 30 121 332 294  74 13  2  3  36  1  49   3  3 .252 .334 .340 .308 (.76)

Carlos Baerga		30 136 520 489 130 26  1 10  26  5  53   2  3 .266 .310 .384 .307 (.84)

Wiki Gonzalez		25  80 251 231  56 13  1  6  18  3  34   2  2 .242 .307 .385 .307 (.46)

Eric Owens		   28 128 401 366  93 14  2  7  33  2  58  15  7 .254 .319 .361 .305 (.80)

Gary Matthews		24  93 347 310  73 14  2  6  35  2  73   7  3 .235 .317 .352 .302 (.62)

Dusty Allen		  26 130 487 440  96 23  2 13  44  2 112   2  3 .218 .292 .368 .292 (.80)

Chris Prieto		 26 112 423 385  98 18  4  3  34  4  64  13  7 .255 .322 .345 .302 (.76)

Damian Jackson	   25 140 538 487 114 25  3  7  46  6 108  16  8 .234 .309 .341 .294 (.84)

Mike Darr			23 115 467 430 105 25  2  5  33  3  79  12  6 .244 .302 .347 .290 (.71)

Rico Rossy		   35 116 392 357  81 20  1  7  33  2  65   3  2 .227 .296 .347 .288 (.80)

Ruben Rivera		 25 130 363 326  69 15  2  8  33  4  97   9  4 .212 .292 .344 .286 (.74)

Carlos Garcia		31  93 337 315  76 14  2  5  19  3  53   9  4 .241 .291 .346 .284 (.76)

David Newhan		 25 116 478 437  99 20  2  8  38  2 111  12  7 .227 .291 .336 .281 (.76)

Ben Davis			22 112 421 391  89 20  1  8  26  4  68   3  2 .228 .283 .345 .280 (.67)

Frank Charles		30 122 440 417  93 25  1  8  19  4 101   2  2 .223 .264 .345 .268 (.79)

For the record, 'league average' is .267/.340/.421 with a wOBA of .336. You can estimate replacement level as somewhere between .280 and .320, depending on position/defence.

As I said last post, I only used data down to AA. I expect to go down to short-season A, and that will probably make the minor leaguers look a bit worse. I'm also applying no park adjustments, mainly because they aren't available for minor league seasons at that time, and because the stats we use for ratings should be park neutral. The only difference will be major league stats will be adjusted.

In the future, I might also make a rough estimation of minor league park factors using team (RS+RA), but that will probably be heavily regressed. The MLEs I'm using do account for league run environments, however, so players playing in the Pacific Coast League won't have their hitting stats artificially inflated.

EDIT: I'd like to add that as with all projections, we are assuming the player plays in the majors for a full season. So the fact that I'm projecting more than 6000 PA simply means we're covering multiple rosters. I might drop those PA estimates and build durability as a skill measured like any other, we'll see.

[patsen]

Hmm, I was sort of hoping to hear something about this progress, but it's probably a busy time for people, with exams and all. I've added A+ and A ball to the stats database, and that should give enough data for most of the players. Ironically, the biggest issue might be the top prospects, as they're more likely to be just drafted, so I'll probably have to run their MLEs by hand, and re-run the projection.

I'm also preparing the 1996-98 park factors, which are only used to normalize ML stats before plugging them into the projections. So, Rockies aren't being awarded or penalized for their park.

So, here's another team for your viewing pleasure, and I dug deeper, and basically filled a AA roster with hitters too, to show how it should spread out.

Chipper Jones		27 142 616 541 160 30  4 24  74  1  84  14  4 .296 .381 .499 .383 (.87)

Ryan Klesko		  28 132 495 440 119 23  3 21  51  3  94   5  3 .270 .349 .480 .359 (.84)

Javy Lopez		   28 133 508 468 132 23  1 25  35  5  84   4  3 .282 .339 .496 .359 (.84)

Brian Jordan		 32 133 525 483 141 28  3 16  34  8  69  15  5 .292 .349 .462 .354 (.82)

Walt Weiss		   35 113 451 389 105 19  3  5  58  3  60   7  2 .270 .368 .373 .337 (.83)

Jose Hernandez	   29 174 484 445 115 21  5 18  37  2 114   5  5 .258 .318 .449 .333 (.79)

Andruw Jones		 22 155 559 510 127 26  4 23  45  4 123  21  8 .249 .315 .451 .332 (.85)

Eddie Perez		  31 114 303 278  74 15  1  9  22  3  48   2  2 .266 .327 .424 .330 (.64)

Gene Schall		  29 109 391 353  88 16  1 14  32  6  87   2  2 .249 .322 .419 .326 (.77)

Keith Lockhart	   34 138 414 381 100 22  2 10  31  2  45   5  3 .262 .321 .409 .322 (.77)

Greg Myers		   33 108 322 296  77 17  1  7  25  1  53   2  1 .260 .320 .395 .316 (.71)

Gerald Williams	  32 136 402 376 100 22  2  9  21  4  64  13  7 .266 .311 .407 .314 (.80)

Bret Boone		   30 148 568 520 129 29  2 15  44  4 100   6  4 .248 .312 .398 .313 (.85)

Brad Tyler		   30 138 470 421  99 19  4 13  47  2 112   9  4 .235 .315 .392 .313 (.82)

Otis Nixon		   40 114 511 460 124 13  3  3  50  1  65  38  9 .270 .342 .330 .309 (.85)

Brian Hunter		 31 186 718 667 171 35  4 11  47  4 117  37 11 .256 .309 .370 .302 (.91)

Freddy Garcia		26 166 519 484 112 27  2 19  31  4 119   2  3 .231 .283 .413 .302 (.84)

Edwin Whatley		27 139 499 452 105 27  2  9  43  4  88   9  5 .232 .305 .361 .297 (.83)

Jorge Fabregas	   29 115 347 323  81 11  1  6  22  2  46   2  1 .251 .303 .347 .291 (.75)

Steve Sisco		  29 120 424 398  97 18  1 10  25  1  71   4  3 .244 .290 .369 .291 (.81)

Marty Malloy		 27 129 510 469 117 21  2  5  38  3  71  11  7 .249 .310 .335 .291 (.84)

Demond Smith		 26 109 429 385  87 17  4  6  39  5  91  17 10 .226 .305 .338 .291 (.83)

Ozzie Guillen		35 118 404 381  96 18  4  4  22  1  32   4  4 .252 .295 .352 .287 (.81)

George Lombard	   23 130 474 433  95 18  3 11  36  5 146  18  8 .219 .287 .351 .284 (.84)

Toby Rumfield		26 134 465 435 103 24  1  7  26  3  69   4  3 .237 .284 .345 .279 (.81)

Howard Battle		27 117 385 361  83 20  1  7  22  2  67   3  3 .230 .278 .349 .278 (.76)

Randall Simon		24 136 503 480 115 24  1 10  20  3  74   3  4 .240 .274 .356 .277 (.85)

Steve Goodell		24 124 418 377  80 17  1  7  32  9  91   2  3 .212 .289 .318 .277 (.80)

Buck McNabb		  26 133 443 405  94 17  2  3  34  4  72  10  8 .232 .298 .306 .276 (.81)

Adam Johnson		 23 128 461 430  92 21  1 13  28  3  90   7  6 .214 .267 .358 .275 (.79)

Mike Glavine		 26 130 457 409  77 16  0 14  44  3 131   1  2 .188 .271 .330 .270 (.80)

Pablo Martinez	   30 110 345 323  75 12  1  4  21  1  73   7  7 .232 .281 .313 .268 (.75)

Mark DeRosa		  24 128 476 439  95 17  1  6  31  6  82   5  8 .216 .277 .301 .263 (.78)

Steven Norris		27 113 382 357  75 20  1  7  21  4  71   2  4 .210 .262 .331 .263 (.73)

Lonell Roberts	   28  94 310 288  63  9  2  4  21  1  67   8  8 .219 .274 .306 .262 (.69)

Jayson Bass		  25 180 604 559 114 24  3 12  41  4 178  19 12 .204 .263 .322 .261 (.87)

Mike Mahoney		 26 111 378 352  70 17  1  6  21  5  85   2  2 .199 .254 .304 .250 (.77)

Tyrone Pendergrass   22 133 518 484 100 16  3  4  29  4 109  28 14 .207 .257 .277 .242 (.82)

Pascual Matos		24 120 408 392  78 15  1  9  13  3 123   4  3 .199 .230 .311 .238 (.79)

Jose Pimentel		24 115 371 352  70 10  1  4  15  4  83  18 12 .199 .240 .267 .229 (.79)

Steve Lackey		 24 120 447 424  85 15  1  3  20  3  96  11  6 .200 .242 .262 .228 (.81)

Glenn Williams	   21 129 467 440  79 19  1  7  23  4 143   4  4 .180 .227 .275 .225 (.79)

Fernando Lunar	   22 114 379 361  65 13  0  4  11  7  69   2  2 .180 .219 .249 .212 (.79)

Projecting a 40-year old for 38 SB is fun, funner is if he would have played all year, he probably would have made it. Mark DeRosa is a strange case. He was a late bloomer, so his projection isn't really out of whack. Throw on 3 stars or so, and he'll probably be useful by the mid 2000s.

Just going by eye, it seems to make sense, we can see where the AAA guys end and the AA guys start. We'll also see random minor leaguers show up as useful players, odds are, they're AAA 1B and OF. Theoretically, an average hitter at DH is a replacement level player, so as long as people are under .330, it should be fine. There are plenty of Shelley Duncan type players who don't do enough for a major league job, even if he can hit decently.

Also, being a full minors mod, I looked at the teams. AAA and AA are fine (the American Association already disbanded), but not all teams have a hi-A team, and some have two. So, either we get creative in the assigning of teams, or we 'promote' some A-league teams to fill the league. I doubt having the right minor league standings is the biggest issue we'll face.

The big issue at this point is indexing. The more minor league stats I look at, the more likely we'll get naming issues. That's the main reason I haven't just posted one large spreadsheet (that and it's still just running lookups and calculating in Excel).

I assume converting to ratings isn't the biggest issue, hence why I'm delivering them as a raw stat line. I know Dylan Bradbury has written some nice tools to run it, but it's impractical to run on thousands of players, and from what I read, the 'official' policy is to use 'official' formulae anyway, so I'll leave that up in the air for now. If it's a by-hand process, I can probably add it to my script, and automate it.

My goal is to simplify the number crunching involved in making these seasons so people are free to work on the more creative and artistic aspects of the mods.

[Homer]

so are you saying you're coming up with a way to automate a stats generator? if so, i would be highly interested in this.

[patsen]

Basically, yes.

I've been playing around with 1996-98 data to project 1999, as this was the thread (and open project) that caught my attention. I'm currently able to create stat lines for almost any player, and we can arguably use this technique for any season. I haven't finished the pitching section yet, but it'll work the same way. Defence could use some work, but we're not rushed yet, so we'll take it one step at a time.

I know there are methods to convert stats to ratings. I've been quietly following Bradbury's work, but the only formulae I've seen are very simple CAP guidelines, and I assume there are more sophisticated ones, used for the official roster mods. Then people only need to worry about the visual things.

Perhaps we can talk later about more details. I'm sure there's a lot of things that can be automated. You can contact me privately if you wish.

EDIT: Should work for any season

[patsen]

I've been working on the pitching half of the spreadsheet, and it seems to be working well. I'm also working on the indexing, so we shouldn't be misattributing someone's minor league numbers, or merging two players together anymore.

So, I ran the spreadsheet on the 1999 Yankees, and we've got something that looks like this:

Player			  Pos Ag   G GS  W  L Sv  IP	 H  ER HR  BB HB  SO  BFP   ERA WHIP   K/9 wOBA  PC (Rel)

Roger Clemens		SP 36  27 27 14  7  0 203.7 162  64 12  70  2 202  809  2.83 1.14  8.93 .278 116 (.71)

Mariano Rivera	   RP 29  50  0  4  2 28  63.0  54  20  4  20  2  48  254  2.86 1.17  6.86 .290  19 (.60)

David Cone		   SP 36  27 27 14  6  0 183.3 161  70 17  65  2 169  746  3.44 1.23  8.30 .302 106 (.67)

Orlando Hernandez	SP 33  24 24 11  6  0 155.0 146  69 15  63  4 133  650  4.01 1.35  7.72 .321 103 (.61)

Andy Pettitte		SP 27  29 28 14  8  0 191.7 196  82 15  67  2 129  806  3.85 1.37  6.06 .322 105 (.69)

Mike Stanton		 RP 32  66  0  4  2  3  71.3  66  34  9  27  2  56  297  4.29 1.30  7.07 .322  17 (.65)

Jeff Juden		   Sw 28  21 10  5  3  0  73.3  69  34  9  31  1  56  308  4.17 1.36  6.87 .327  74 (.44)

Jeff Nelson		  RP 32  55  0  3  4  2  54.3  51  23  5  24  4  45  232  3.81 1.38  7.45 .328  16 (.53)

Hideki Irabu		 SP 30  27 25 11  8  0 153.3 145  74 23  59  2 111  639  4.34 1.33  6.52 .331  92 (.63)

Ramiro Mendoza	   Sw 27  33 14  8  5  1 119.3 130  54 11  32  4  63  502  4.07 1.36  4.75 .331  76 (.66)

Dan Naulty		   RP 29  38  0  2  2  1  43.0  45  24  6  18  1  30  185  5.02 1.47  6.28 .349  18 (.42)

Allen Watson		 SP 28  26 20  7  8  0 125.3 140  68 18  46  1  85  541  4.88 1.48  6.10 .354  86 (.58)

Don Wengert		  Sw 29  35 13  4  7  1 108.0 129  65 15  39  3  65  476  5.42 1.56  5.42 .365  72 (.65)

Greg McCarthy		RP 30  52  0  3  3  2  51.0  52  29  7  34  3  44  233  5.12 1.69  7.76 .367  17 (.51)

Mike Buddie		  RP 28  42  5  5  4  2  77.0  97  47  9  36  3  46  353  5.49 1.73  5.38 .378  27 (.62)

Tony Fossas		  RP 41  70  0  2  4  0  46.7  56  25  5  24  2  32  213  4.82 1.71  6.17 .373  11 (.47)

Trevor Wilson		Sw 33  30 17  6  6  0 115.7 135  70 17  61  6  67  529  5.45 1.69  5.21 .378  82 (.62)

Dave Pavlas		  RP 36  43  2  3  3  8  60.3  76  37  9  24  2  40  273  5.52 1.66  5.97 .379  22 (.56)

Ryan Bradley		 Sw 23  35 14  7  5  3 109.0 122  73 18  66  5  79  501  6.03 1.72  6.52 .381  77 (.64)

Jay Tessmer		  RP 27  60  0  5  3 18  66.7  86  38  8  31  2  46  307  5.13 1.76  6.21 .382  19 (.62)

Jason Grimsley	   RP 31  39  7  4  5  1  85.0  99  59 12  55  5  53  399  6.25 1.81  5.61 .386  32 (.63)

Todd Erdos		   RP 25  52  0  3  3 14  55.7  72  40  8  32  3  40  264  6.47 1.87  6.47 .397  19 (.55)

Ed Yarnall		   SP 23  26 25  8  7  0 133.3 168  92 22  79  7  91  630  6.21 1.85  6.14 .399  93 (.56)

Ben Ford			 RP 23  56  0  3  4 10  70.0  90  51 10  42  4  46  332  6.56 1.89  5.91 .400  22 (.64)

Todd Erdos		   RP 25  52  0  3  3 14  55.7  73  40  8  32  3  40  264  6.47 1.89  6.47 .401  19 (.55)

Chris Nichting	   Sw 33  36 11  4  7  1  93.7 129  77 17  45  4  65  443  7.40 1.86  6.25 .408  71 (.62)

Luis de los Santos   SP 21  25 24  6  9  0 130.7 210 115 20  60  5  63  643  7.92 2.07  4.34 .429  95 (.56)

Dave Zancanaro	   SP 30  23 16  4  6  1  93.0 134  78 19  54  5  50  454  7.55 2.02  4.84 .430  84 (.48)

Matt Williams		Sw 28  42 17  5  8  0 117.3 176  98 21  72  6  68  586  7.52 2.11  5.22 .431  73 (.67)

The pitch counts shown are in terms of actual (estimated) pitches per game. I named every pitcher SP, RP or Sw depending of how many of their innings were believed to be start innings. I then estimated PC in starts and relief, and applied an appropriate mix to get a single number. I don't expect them to be used as Stamina directly, but it should be a good start. We can compare methods, and come up with a single number.

[patsen]

Alright, things are going well... I added in the vsL/vsR splits. I got myself the 1996-99 numbers, regressed them, and put them in a table to look up. We should be getting realistic splits now.

And, because of the regressions, the formulae will work whether someone has thousands of ABs, or zero.

Here's the 1999 All Star teams for another sample. Oh, and I got park effects working too, so Coors-inflated players are brought back down to earth.

N.L. All Stars	   Ag   G  PA  AB   H 2B 3B HR  BB HB  SO  SB CS  Avg  OBP  Slg wOBA (Rel) Av:vL/VR  Sl:vL/vR

Barry Larkin		 35 131 536 465 138 27  6 17  68  4  58  23  5 .297 .392 .490 .386 (.84) .302/.295 .511/.483

Larry Walker		 32 131 519 453 144 36  3 25  58  8  80  20  5 .318 .405 .576 .420 (.84) .298/.324 .515/.595

Sammy Sosa		   30 143 625 571 158 24  2 41  51  3 146  17  8 .277 .339 .541 .374 (.87) .283/.274 .555/.536

Mark McGwire		 35 142 603 484 137 22  0 52 112  7 136   3  1 .283 .425 .651 .450 (.87) .289/.281 .657/.648

Matt Williams		33 131 544 500 133 25  2 23  40  4  99   7  3 .266 .325 .462 .341 (.85) .274/.263 .490/.452

Jeff Bagwell		 31 135 600 490 146 32  2 32 101  9  88  19  7 .298 .427 .567 .428 (.88) .310/.294 .577/.564

Mike Piazza		  30 137 576 513 173 29  1 32  61  2  76   3  2 .337 .410 .585 .426 (.87) .344/.335 .598/.580

Jeromy Burnitz	   30 159 597 526 140 30  3 30  66  5 127  12  7 .266 .353 .506 .370 (.84) .254/.270 .459/.520

Jay Bell			 33 143 585 514 133 26  3 17  65  5 110   6  5 .259 .347 .420 .340 (.86) .266/.256 .431/.416

Sean Casey		   24 124 457 414 114 27  1 10  38  5  66   2  2 .275 .344 .418 .337 (.80) .254/.281 .375/.430

Jeff Kent			31 136 556 506 140 34  2 23  43  8 100   9  4 .277 .344 .488 .360 (.85) .281/.275 .494/.486

Ed Sprague		   31 127 511 463 110 25  2 19  39  9  94   2  2 .238 .309 .423 .320 (.85) .244/.235 .443/.416

Mike Lieberthal	  27 110 417 381  95 19  2 13  31  5  61   3  2 .249 .314 .412 .319 (.78) .261/.245 .435/.403

Dave Nilsson		 29 118 447 401 114 24  1 15  44  2  63   3  3 .284 .358 .461 .359 (.83) .253/.294 .380/.486

Luis Gonzalez		31 191 728 652 167 37  4 16  68  8  90  13  8 .256 .334 .399 .326 (.88) .243/.260 .351/.413

Gary Sheffield	   30 126 531 427 124 26  2 24  95  9  59  14  6 .290 .429 .529 .418 (.86) .302/.286 .550/.522

Vladimir Guerrero	24 144 576 532 160 32  5 25  37  7  76   9  8 .301 .354 .521 .376 (.85) .305/.299 .529/.517

Tony Gwynn		   39 124 519 479 159 36  1 12  38  2  29   8  3 .332 .383 .486 .380 (.85) .324/.335 .469/.492

Brian Jordan		 32 134 526 484 143 28  3 16  34  8  68  15  5 .295 .352 .465 .356 (.82) .309/.291 .493/.455

Alex Gonzalez		26 144 549 508 122 25  2 13  36  5 109  16  6 .240 .297 .374 .297 (.85) .249/.237 .392/.367


A.L. All Stars	   Ag   G  PA  AB   H 2B 3B HR  BB HB  SO  SB CS  Avg  OBP  Slg wOBA (Rel) Av:vL/VR  Sl:vL/vR

Kenny Lofton		 32 132 595 530 156 25  5 11  63  2  78  41 13 .294 .371 .423 .354 (.87) .282/.299 .380/.437

Nomar Garciaparra	25 132 592 554 163 31  7 28  33  5  71  14  6 .294 .340 .527 .370 (.85) .297/.293 .542/.521

Ken Griffey		  29 141 624 552 162 28  3 45  65  7 101  15  4 .293 .375 .600 .411 (.87) .277/.299 .572/.608

Manny Ramirez		27 138 586 513 152 33  2 31  68  5 104   5  3 .296 .384 .550 .400 (.87) .309/.292 .576/.540

Jim Thome			28 124 524 432 124 26  2 30  88  4 122   2  1 .287 .412 .565 .419 (.86) .267/.293 .485/.590

Cal Ripken		   38 143 599 546 146 29  1 15  49  4  68   2  2 .267 .332 .407 .327 (.87) .272/.266 .417/.403

Rafael Palmeiro	  34 142 626 552 153 31  2 32  70  5  88   8  3 .277 .364 .514 .379 (.88) .264/.282 .491/.523

Ivan Rodriguez	   27 133 568 529 158 33  3 17  34  4  76   7  1 .299 .345 .469 .353 (.87) .305/.296 .482/.464

Roberto Alomar	   31 132 574 513 156 34  2 15  59  2  62  14  5 .304 .378 .466 .371 (.86) .300/.306 .461/.468

Ron Coomer		   32 143 527 500 140 24  2 15  26  1  74   4  2 .280 .317 .426 .324 (.83) .294/.274 .458/.413

Jose Offerman		30 137 596 526 158 28  9  6  67  3  89  25 10 .300 .383 .422 .360 (.86) .303/.300 .421/.423

Tony Fernandez	   37 136 515 472 139 28  2 11  37  7  58  10  7 .294 .355 .432 .348 (.79) .314/.287 .458/.423

Brad Ausmus		  30 135 481 430 112 18  2  7  47  4  70  10  5 .260 .339 .360 .317 (.82) .263/.260 .370/.357

Bernie Williams	  30 124 554 484 154 29  5 23  69  1  75  14  7 .318 .404 .541 .409 (.86) .325/.315 .565/.532

Harold Baines		40 115 416 371 108 21  0 13  44  1  53   2  1 .291 .368 .453 .362 (.82) .273/.296 .408/.465

Jose Canseco		 34 134 571 504 124 24  1 33  61  5 134  16  8 .246 .333 .494 .355 (.84) .253/.243 .511/.488

John Jaha			33  89 357 305  78 13  1 15  47  5  74   3  2 .256 .364 .452 .360 (.77) .268/.251 .479/.442

B.J. Surhoff		 34 144 573 522 146 31  3 18  48  3  73   5  4 .280 .344 .454 .348 (.86) .270/.283 .415/.466

Shawn Green		  26 152 590 542 149 29  4 23  44  5 114  20  6 .275 .336 .470 .349 (.85) .255/.281 .412/.488

Magglio Ordonez	  25 141 545 511 136 27  1 13  28  6  69   8  8 .266 .312 .399 .313 (.85) .272/.264 .411/.395

Derek Jeter		  25 138 622 561 167 25  6 13  54  7 107  19  7 .298 .367 .433 .355 (.88) .302/.296 .445/.429

Omar Vizquel		 32 142 579 523 145 26  4  6  53  3  58  32 10 .277 .347 .377 .326 (.86) .270/.280 .372/.378


Pitchers			 Ag   G  PA  AB   H 2B 3B HR  BB HB  SO  SB CS  Avg  OBP  Slg wOBA (Rel) Av:vL/VR  Sl:vL/vR

Curt Schilling	   32  31  73  71  11  1  0  0   2  0  27   0  0 .155 .178 .169 .160 (.87) .159/.153 .176/.166

Pedro Martinez	   27  15  36  34   4  1  0  0   2  0  15   0  0 .118 .167 .147 .149 (.80) .121/.116 .153/.145

I also added a switch to project pitchers. They need ratings, so pitchers who have hit will have those stats added in, those who haven't get regressed. Once I get my indexing issues sorted out, I'll try to post a full spreadsheet, covering at least the major leaguers, in normal 'not total minors' style. If everything seems right from there, I'll proceed down to hit the prospects and organizational players.

The spreadsheet seems to work fine for established players and established minor leaguers. Who they struggle with are the good prospects, those without a lot of minor league time, and haven't broken into the majors yet. Magglio Ordonez is a great example of one of these breakouts.

I might need to cheat in some 1999 stats to try to get these players closer to the mark. It's better than artificially inflating prospects' ratings up arbitrarily. Or, change the regression factor. Ordonez's minor league numbers are bringing him down.

EDIT: Shrunk table, since they're getting wide.

[andy82]

Probably the wrong thread, but it would be awesome for Bradbury's calculators to take stats straight from Baseball Reference, and then have those results put straight into MVPEdit on the required player. Highly unlikely to happen, but it would be fantastic if it was possible.

[Homer]

that's been on my wish list for awhile too. i just don't think anyone can do it.

[andy82]

that's been on my wish list for awhile too. i just don't think anyone can do it.

I remember a while back that someone produced an editor for PES4 (A soccer game) that extracted stats from FM2005 (Soccer Management Game), converted them into the PES4 format and replaced the selected players attributes.

What is available at the mo is the conversion method and the editor, we just need to integrate them all together and take stats from a stats site (Like baseball-reference.com) or Lahman database.

If this was put together then the creation of season mods would be so much easier and quicker. I just wish I had the knowledge and ability to do that.

[Jim825]

MVPEdit already allows you to pull in teams from the Lahman database. That's always the first step we take when creating single season mods. After that, we run a series of Global Tweaks that stecropper developed which make results even more realistic. Finally, we run multiple sims and make individual player tweaks to get the sim results to be comparable to the results from the seasons you are modding.

I've used this method for all of the single season mods I've created and have gotten pretty realistic results, even to the point of the simmed league leaders and award winners for the season (Cy Young, MVP, Rolaids reliever) being the same as the real life league leaders and award winners for the season that I modded.

[patsen]

That's basically what I'm working on, I've been digging around for the better part of the last week on finding how to run the ratings. If I could find the ratings MVPEdit does, then everything else should fall into place.

I've also found a thread on how SwingingSoriano did it, at http://www.mvpmods.com/index.php?showtopic...&hl=pitcher

I know there was a thread on an algorithm for doing pitcher pitches properly, but I can't seem to find it. I know DylanBradbury incorporated that algorithm in his ratings, but I can't find the original thread. Once my exams are over, I'll look deeper.

But yea, I basically take techniques, and try to automate them. Right now, it's one large Excel spreadsheet so I can change things easier, but ideally, it would run directly off B-R data, with very little intervention.

That's why I've been asking around, trying to see what goes into roster creation, so I can see what gets automated. But, unless I get the MVPEdit code, or something similar, I'm kind of stuck.

[DylanBradbury]

I know DylanBradbury incorporated that algorithm in his ratings, but I can't find the original thread. Once my exams are over, I'll look deeper.

My pitching formulas were very loosely based on the formulas inside mvpedit, not from any thread posted here. They've involved and improved to an almost unrecognizable state [from the original formula] now.

[Jim825]

To come up with the proper pitches for pitchers, I use two methods:

First, I use the Neyer / James Guide to Pitchers: http://www.amazon.com/Neyer-James-Guide-Pi...8767&sr=8-1

It has entries for hundreds of pitchers with the pitches they threw as well as their delivery.

Secondly, if a player isn't listed in the book, I use Baseball Mogul 2007. I select the year I want to play and then look through the rosters for that particular year and look at the player attributes. For pitchers, it lists (pretty accurately) the pitches they threw and shows ratings to help you determine which one is their #1 pitch, their #2 pitch, etc. as well as their control settings. I've looked through the Neyer / James book and compared a pitcher's listed pitches with the ones listed for him in Baseball Mogul and they are usually the exact same.

Stecropper and I had talked about this a few years ago and he found that the ratings for both pitchers and hitters in Baseball Mogul are pretty close to those used in MVP. You can always talk to him to get more information about this.

Stecropper helped develop and test some of the formulas that rglass used in MVPEdit, so he can also provide you with more information on that as well.

[patsen]

I'm talking about the one where you determine a pitcher's pitches and ratings (was 6-steps or so, started by picking the pitcher type and working from there). There was an algorithm on that, and I recall you implementing it. That's the thread I'm trying to find.

I also know there are a lot of loose threads on ratings, mostly based on some charts tying contact to Avg, and so forth, like the ones listed above. I assume MVPEdit does something similar. There's a lot of work to do, which is why I didn't want to duplicate work, but if you want to protect your ratings, I'll respect your wishes.

This is a big project, so it'll likely involve a lot of people. We aren't just creating ratings for a player, we're creating a full league at a time.

(I have an exam in 20 min, so I'll be back later.)

[DylanBradbury]

I'm talking about the one where you determine a pitcher's pitches and ratings (was 6-steps or so, started by picking the pitcher type and working from there). There was an algorithm on that, and I recall you implementing it. That's the thread I'm trying to find.

I also know there are a lot of loose threads on ratings, mostly based on some charts tying contact to Avg, and so forth, like the ones listed above. I assume MVPEdit does something similar. There's a lot of work to do, which is why I didn't want to duplicate work, but if you want to protect your ratings, I'll respect your wishes.

This is a big project, so it'll likely involve a lot of people. We aren't just creating ratings for a player, we're creating a full league at a time.

(I have an exam in 20 min, so I'll be back later.)

Patsen, I know the thread you mean, and I know I never implemented the "6-Step Plan" you're referring to.

[patsen]

Oh, sorry about that. Guess I'll need to find it, since it looks like there's a lot of useful information there. But, I never intended to try to build ratings from scratch. I was hoping to streamline already existing techniques so that the long, number crunching stuff can all be done quickly. However, from what I can hear, people are either using MVPEdit's formulae, with some standard modifications, or they're KG and using more proprietary methods.

I'm trying to save people's time by making the long and boring part one-click. Trying to recreate something that already exists isn't beneficial to me, as I could be spending my time doing something more productive. I never meant to offend you, or steal your work. I'm just trying to see what's on the table, and how it can be improved.

[patsen]

Ok, the algorithm may not be perfect, but I've got the hitting stats (not ratings) of every position player who had a PA in 1998. In we were making MVP 99, we'd start here. Then, we'd add the total minors.

Arguably, if we could import that into MVPEdit, we'd have a nice chunk of the work done. Once we get ratings from stats, what I may do is have the file export to MVPEdit player .mep files, so we can just dump the players on the correct teams. Heh, if I'm handling the entire league at a time, it may be simpler to just modify all the .dat files instead.

I could do the same for pitchers, but I'm debating whether we should solve for pitchers and fielders simultaneously... probably better to keep it simple, then improve it.

99_mvp_projection_mlb.txt

[andy82]

After that, we run a series of Global Tweaks that stecropper developed which make results even more realistic.

Where can these be found? I've looked over the boards but no sign?

[DylanBradbury]

I was hoping to streamline already existing techniques so that the long, number crunching stuff can all be done quickly. However, from what I can hear, people are either using MVPEdit's formulae, with some standard modifications.

As Jim has already said, MVPEdit does this for you...

MVPEdit already allows you to pull in teams from the Lahman database.

[DylanBradbury]

Where can these be found? I've looked over the boards but no sign?

I think this is what you and patsen want.

Hopefully that helps.

[patsen]

The Global edits is found at http://www.mvpmods.com/index.php?showtopic=32568

And thanks for the MVPEdit tutorial, but it doesn't answer the question. MVPEdit draws data from the Lahman database and creates ratings from only that data. And, it needs to be run 30 times. And rates a 1/1 cup of coffee 100/100.

I'm cross-referencing multiple sources of data (MLB stats, MiLB, retro splits for now, but I'll have more later), and trying to convert to ratings. I'll run them through SwingingSoriano's lookup tables later, but I wouldn't mind doing some in-game experiments, first. Like Jim's posted earlier, but more generalized.

I've started some quick, but I'm wondering if anyone else has done something similar before I spend half a week repeating someone else's work. I made a roster where every hitter has ratings of 70 across the board. That's about league average. The experiment is setting plate discipline to 80, and see the effect. 90... 60... and so forth, see the effect of the changes on stats. From there, we should see what stats should be factored in, and the general effect. With all the dynasty options off (and no DH), the league totals (since every hitter should be identical) should be an adequate sample.

I've always thought SO should be part of the contact formula (since contact% is usually defined as [AB-SO] / AB, the rate of actually making contact), so hopefully we'll have more info, but when I started this, I was really hoping to just handle the stat collecting and have someone else convert to ratings, but I guess it isn't always easy.

(EDIT: Heh, guess I owe Dylan a coke, now.)

[DylanBradbury]

And thanks for the MVPEdit tutorial, but it doesn't answer the question. MVPEdit draws data from the Lahman database and creates ratings from only that data. And, it needs to be run 30 times. And rates a 1/1 cup of coffee 100/100.

Patsen, I don't know if you're still doing ratings for this mod, but an idea hit me today: Replace the hitters '99 stats in the Lahman DB with your compiled stats, and then import with MVPEdit. That way, you'll have to only add minor leaguers without Major league experience.

Hopefully that helps!