Offensive Projections
Name P Age AVG OBP SLG G AB R H 2B 3B HR RBI BB K SB CS OPS+
David Wright 3b 27 .307 .396 .507 148 566 104 174 37 2 24 101 84 130 24 7 140
Carlos Beltran# cf 33 .294 .381 .514 113 432 79 127 28 2 21 83 62 78 16 2 137
Jose Reyes# ss 27 .294 .363 .455 109 462 79 136 24 10 10 49 49 58 43 11 117
Carlos Delgado* 1b 38 .271 .351 .479 74 284 44 77 15 1 14 56 32 62 0 0 119
Fernando Tatis 1b 35 .277 .345 .458 107 336 46 93 19 3 12 49 30 64 4 2 113
Angel Pagan# cf 28 .285 .335 .446 92 298 48 85 16 7 6 36 24 53 12 4 107
Chris Carter* 1b 27 .278 .334 .428 128 493 57 137 27 1 15 69 41 79 1 0 102
Gary Sheffield lf 41 .248 .347 .411 97 302 43 75 14 1 11 45 43 60 4 1 102
Luis Castillo# 2b 34 .290 .368 .349 120 441 72 128 14 3 2 38 55 48 18 5 94
Daniel Murphy* 1b 25 .274 .326 .408 150 525 66 144 34 3 10 69 41 79 6 4 95
Jeff Francoeur rf 26 .269 .310 .425 155 602 75 162 34 3 18 86 32 111 4 3 94
Jason DuBois lf 31 .242 .310 .425 116 388 50 94 18 1 17 59 31 113 2 1 94
Josh Thole* c 23 .273 .335 .360 124 447 48 122 26 2 3 51 40 55 5 2 87
Chip Ambres rf 30 .252 .317 .393 125 448 57 113 23 2 12 54 42 100 4 4 89
Emil Brown rf 35 .259 .314 .396 100 351 48 91 19 1 9 55 27 72 5 3 89
Jeremy Reed* lf 29 .274 .320 .394 139 368 47 101 20 3 6 38 24 52 5 5 90
Andy Green 3b 32 .248 .325 .369 93 290 44 72 13 2 6 29 32 58 4 2 86
Mike Lamb* 3b 34 .266 .314 .377 108 342 41 91 19 2 5 44 23 40 0 1 84
Nicholas Evans lf 24 .240 .295 .402 130 438 53 105 25 5 12 50 33 113 1 0 84
Fernando Martinez cf 21 .251 .297 .406 86 335 43 84 21 2 9 41 19 75 4 2 86
Cory Sullivan* cf 30 .262 .319 .364 127 390 51 102 18 5 4 38 33 70 8 4 83
Jesus Feliciano* lf 31 .284 .327 .349 112 398 46 113 18 1 2 34 23 43 8 6 82
Ike Davis* 1b 23 .227 .302 .388 122 472 46 107 24 2 16 58 50 154 0 2 83
Alex Cora* ss 34 .251 .322 .336 85 235 28 59 10 2 2 17 19 26 4 2 77
Robinson Cancel c 34 .256 .299 .358 69 215 21 55 11 1 3 19 13 31 5 2 75
Anderson Hernandez# 2b 27 .255 .304 .338 129 462 58 118 19 5 3 40 32 81 9 7 72
Ruben Tejada ss 20 .247 .299 .324 139 527 56 130 23 3 4 42 34 88 11 4 67
Omir Santos c 29 .247 .289 .342 92 295 28 73 14 1 4 33 15 55 1 1 69
Javier Castillo 3b 26 .246 .288 .343 127 472 43 116 22 3 6 43 27 107 3 3 68
Ramon Martinez ss 37 .243 .300 .313 49 144 14 35 7 0 1 18 12 20 1 1 65
Rene Rivera c 26 .226 .268 .353 75 266 23 60 13 0 7 29 14 72 0 1 65
Shawn Bowman 3b 25 .216 .258 .320 70 269 24 58 12 2 4 24 14 93 1 1 54
Defensive Projections
Name CThr 1b 2b 3b ss lf cf rf
Wright Av/100
Beltran# Vg/97
Reyes# Av/93
Delgado* Fr/103
Tatis Av/116 Pr/116 Fr/120 Fr/120
Pagan# Av/92 Av/105 Av/92
Carter* Pr/146 Pr/108 Pr/137
Sheffield Pr/75 Pr/75 Pr/75
Castillo# Fr/100
Murphy* Vg/144 Fr/173 Fr/144 Av/129 Av/129
Francoeur Av/97
DuBois Fr/125 Pr/129 Pr/129
Thole* Fr Fr/111
Ambres Vg/104 Pr/131 Vg/104
Brown Av/127 Fr/119
Reed* Av/75 Fr/84 Av/75
Green Av/100 Fr/120 Pr/120 Fr/100
Lamb* Av/121 Pr/114
Evans Av/95 Fr/123
Martinez Av/93 Fr/93 Av/93
Sullivan* Av/84 Fr/84
Feliciano* Vg/78 Fr/71 Vg/78
Davis* Av/105
Cora* Av/86 Av/99
Cancel Av Fr/114 Fr/114
Hernandez# Vg/107 Av/108
Tejada Vg/131 Vg/131
Santos Av
Castillo Av/146 Fr/175
Martinez Fr/61 Fr/61 Pr/61
Rivera Av
Bowman Vg/79
* - Bats Left
# - Switch Hitter
ODDIBE (Odds of Important Baseball Events)
Name PO EX VG AV FR PO COMP 1 COMP 2 COMP 3
WrightDavid 3B 85% 14% 1% 0% 0% AlfonzoEdgardo JacobyBrook SantoRon
BeltranCarlos CF 97% 3% 0% 0% 0% LankfordRay AndersonBrady van SlykeAndy
ReyesJose SS 86% 11% 3% 1% 0% FurcalRafael ApplingLuke RenteriaEdgar
DelgadoCarlos 1B 12% 38% 28% 19% 3% AdcockJoe JohnsonCliff FairlyRon
TatisFernando 1B 5% 25% 25% 31% 14% KarrosEric PippWally MormanRuss
PaganAngel CF 24% 27% 33% 15% 2% KeoughMarty WynneMarvell SandersDeion
CarterChris 1B 2% 10% 20% 45% 24% CoxSteve McCrawTommy JohnsonRon
SheffieldGary LF 5% 17% 21% 27% 30% SauerHank SlaughterEnos WanerPaul
CastilloLuis 2B 20% 18% 22% 25% 15% RandolphWillie VinaFernando BlasingameDon
MurphyDaniel 1B 0% 3% 9% 35% 53% KrukJohn DalenaPete MastellerDan
FrancoeurJeff RF 1% 4% 11% 26% 59% JonesJacque ChamberlainWes JamesChris
DuboisJason LF 1% 5% 10% 25% 59% JohnsonDeron HatcherChris ClarkBobby
ReedJeremy LF 0% 3% 5% 14% 79% UnserDel MayDerrick WilliamsDallas
BrownEmil RF 1% 5% 6% 15% 73% SierraRuben WardGary HoagMyril
TholeJoshua C 1% 13% 31% 42% 13% NavarroDioner LevisJesse KendallJason
AmbresChip RF 1% 2% 4% 15% 78% CordovaMarty SpencerShane CimoliGino
MartinezFernando LF 0% 1% 4% 17% 77% JamesChris O’NeillPaul HartsockBrian
GreenAndy 3B 0% 3% 12% 32% 54% TatisFernando RoysterJerry BarnesSkeeter
LambMike 3B 0% 2% 8% 24% 66% ChristmanMark StrippJoe BarnesSkeeter
SullivanCory CF 1% 2% 12% 34% 50% PodsednikScott ScottTonyMcCrackenQuinton
EvansNick LF 0% 1% 3% 12% 85% MoutonLyleSmithermanStephe PoeCharles
FelicianoJesus LF 0% 1% 2% 6% 91% WanerLloyd SpanglerAl RamosKen
DavisIke 1B 0% 0% 1% 8% 91% HornSam SantosChad WhitesellJosh
CancelRobinson C 0% 3% 9% 37% 51% HemsleyRollie NorrisSteven CeroneRick
CoraAlex SS 0% 3% 19% 43% 35% BoleyJoe FletcherScott FernandezTony
HernandezAnderson 2B 0% 0% 1% 6% 93% MaciasJose InfanteAlexis VerasQuilvio
TejadaRuben SS 0% 0% 5% 24% 71% ChavezPedro DiSarcinaGary InfanteOmar
SantosOmir C 0% 0% 2% 20% 77% HeintzChris HubbardMike DalesandroMark
CastilloJavier 3B 0% 0% 0% 2% 98% SpringerSteve GonzalezEdgar ColombinoCarlo
MartinezRamon SS 0% 0% 1% 14% 85%MaranvilleRabbit LockhartKeith BluegeOssie
RiveraRene C 0% 0% 0% 8% 92% ColbertCraig ReynoldsRonn BrownJason
BowmanShawn 3B 0% 0% 0% 0% 100% CaliseMike RustBrian DuncanCarlos
Name .300 BA .375 OBP .500 SLG 140 OPS+ 45 2B 10 3B 30 HR 30 SB
WrightDavid 60% 77% 53% 47% 15% 0% 19% 24%
BeltranCarlos 40% 57% 60% 39% 3% 0% 11% 1%
ReyesJose 40% 32% 20% 12% 0% 65% 0% 95%
DelgadoCarlos 19% 21% 33% 9% 0% 0% 0% 0%
TatisFernando 20% 14% 16% 3% 0% 1% 0% 0%
PaganAngel 31% 10% 12% 5% 0% 25% 0% 0%
CarterChris 17% 4% 5% 1% 1% 0% 2% 0%
SheffieldGary 7% 18% 3% 1% 0% 0% 0% 0%
CastilloLuis 38% 40% 0% 0% 0% 2% 0% 4%
MurphyDaniel 13% 1% 1% 0% 6% 1% 0% 0%
FrancoeurJeff 8% 0% 2% 0% 8% 1% 3% 0%
DuboisJason 1% 1% 4% 1% 0% 0% 2% 0%
ReedJeremy 20% 3% 1% 0% 0% 1% 0% 0%
BrownEmil 9% 3% 3% 0% 0% 0% 0% 0%
TholeJoshua 12% 4% 0% 0% 0% 0% 0% 0%
AmbresChip 4% 2% 1% 1% 1% 0% 0% 0%
MartinezFernando 3% 0% 1% 0% 0% 0% 0% 0%
GreenAndy 4% 5% 0% 0% 0% 0% 0% 0%
LambMike 11% 1% 0% 0% 0% 0% 0% 0%
SullivanCory 8% 3% 0% 0% 0% 5% 0% 0%
EvansNick 0% 0% 0% 0% 0% 4% 0% 0%
FelicianoJesus 28% 5% 0% 0% 0% 0% 0% 0%
DavisIke 0% 0% 0% 0% 0% 0% 0% 0%
CancelRobinson 9% 1% 0% 0% 0% 0% 0% 0%
CoraAlex 7% 5% 0% 0% 0% 0% 0% 0%
HernandezAnderson 3% 0% 0% 0% 0% 5% 0% 0%
TejadaRuben 1% 0% 0% 0% 0% 1% 0% 0%
SantosOmir 3% 0% 0% 0% 0% 0% 0% 0%
CastilloJavier 1% 0% 0% 0% 0% 0% 0% 0%
MartinezRamon 10% 3% 0% 0% 0% 0% 0% 0%
RiveraRene 0% 0% 0% 0% 0% 0% 0% 0%
BowmanShawn 0% 0% 0% 0% 0% 0% 0% 0%
Pitching Statistics - Starters
Name Age ERA W L G GS INN H ER HR BB K ERA+
Johan Santana* 31 3.23 16 9 30 30 206.1 184 74 24 52 203 133
John Maine 29 4.20 9 8 23 22 122.0 113 57 14 57 103 103
Jon Niese* 23 4.57 8 9 26 26 143.2 155 73 14 55 103 95
Mike Pelfrey 26 4.86 10 14 33 33 189.0 216 102 18 69 111 89
Jenrry Mejia 20 4.90 3 5 16 16 75.1 78 41 8 44 54 88
Oliver Perez* 28 4.93 7 10 27 27 142.1 138 78 20 87 132 88
Nelson Figueroa 36 5.06 6 10 28 20 131.2 151 74 17 46 91 86
Fernando Nieve 27 5.19 3 5 24 11 67.2 75 39 9 31 50 83
Tim Redding 32 5.32 6 10 32 25 150.2 171 89 23 58 96 81
Jack Egbert 27 5.46 6 11 29 22 123.2 147 75 15 49 75 79
Tobi Stoner 25 5.58 6 11 26 25 132.1 156 82 18 56 77 78
Bobby Parnell 25 5.67 5 11 52 20 120.2 142 76 14 67 86 76
Adam Pettyjohn* 33 5.67 6 12 27 21 127.0 152 80 22 41 65 76
Bradley Holt 23 5.72 4 8 18 18 83.1 88 53 14 50 67 76
Lance Broadway 26 5.98 5 12 33 22 143.0 175 95 20 67 75 72
Pitching Statistics - Relievers
Name Age ERA W L G GS INN H ER HR BB K ERA+
Francisco Rodriguez 28 3.13 5 2 74 0 72.0 55 25 6 37 88 138
J.J. Putz 33 3.53 4 3 43 0 43.1 37 17 4 20 45 123
Pedro Feliciano* 33 3.92 4 3 83 0 57.1 54 25 6 24 56 110
Sean Green 31 4.30 3 4 75 0 75.1 75 36 6 36 61 101
Elmer Dessens 39 4.42 2 2 34 0 36.2 37 18 4 13 23 98
Brian Stokes 30 4.54 3 4 77 0 81.1 85 41 8 38 56 95
Brandon Knight 34 4.61 4 5 25 8 68.1 70 35 9 26 57 94
Kyle Snyder 32 4.84 3 4 31 9 70.2 76 38 10 27 46 89
Pat Misch* 28 4.87 5 7 44 13 114.2 127 62 15 36 70 89
Jon Switzer* 30 5.06 2 3 48 0 58.2 64 33 8 24 43 85
Eddie Kunz 24 5.43 3 5 48 0 58.0 64 35 7 33 36 80
Arturo Lopez* 27 5.48 1 2 28 2 42.2 48 26 5 25 26 79
* - Throws Left
ODDIBE (Odds of Important Baseball Events)
Player PO TOP MID BOT COMP 1 COMP 2 COMP 3
RodriguezFrancisco RP 71% 26% 3% LidgeBrad GossageRich GarreltsScott
SantanaJohan SP 92% 7% 1% ViolaFrank CandelariaJohn HubbellCarl
PutzJ.J. RP 45% 40% 15% BerenguerJuan LeskanicCurt NelsonJeff
FelicianoPedro RP 34% 50% 16% McElroyChuck LavelleGary MyersRandy
MaineJohn SP 35% 52% 13% PavlikRoger WellsKip MillerWade
GreenSean RP 15% 54% 30% BernardDwight StewartSammy PowerTed
DessensElmer RP 29% 30% 40% WhiteRick MesaJose FowlerArt
StokesBrian RP 6% 48% 46% RiedlingJohn PufferBrandon ParontoChad
NieseJon SP 15% 61% 24% SmithBud ThurmondMark TroutSteve
KnightBrandon SP 8% 41% 50% GardnerMark RasmussenEric CarraraGiovanni
SnyderKyle RP 4% 30% 66% ReedJerry BrockChris SoffRay
PelfreyMike SP 6% 56% 38% LaCossMike ScottMike FireovidSteve
MischPatrick RP 4% 31% 65% KnottEric YoungCliff CrawfordJoe
MejiaJenrry SP 9% 44% 48% DurbinJ.D. NippertDustin AlbersMatt
PerezOliver SP 7% 46% 47% ScoreHerb SearcySteve LollarTim
FigueroaNelson SP 6% 38% 57% ShowEric TomkoBrett PowerTed
SwitzerJon RP 3% 28% 69% EischenJoey HolzemerMark CrowellJim
NieveFernando SP 2% 24% 74% GrimsleyJason CressendJack ShieldsSteve
ReddingTim SP 1% 26% 73% FernandezJared OrtizRamon SpringerDennis
KunzEddie RP 1% 17% 82% EyreWillie DrummondTim BowlesBrian
EgbertJack SP 1% 21% 78% RandallScott PughTim AguileraRick
LopezArturo RP 2% 16% 82% JarvisMatt BarzillaPhilip RodriguezRich
StonerTobi SP 0% 18% 81% ArroyoBronson IrelandEric HarikkalaTim
ParnellBobby SP 0% 5% 95% ManningCharlie TorresDilson MillerKurt
PettyjohnAdam SP 2% 15% 83% MulhollandTerry MayDarrell LorraineAndrew
HoltBradley SP 2% 17% 82% KinneyMatt BonserBoof ElliottDonnie
BroadwayLance SP 0% 5% 95% BauerPeter FarnsworthJeff MendozaHatuey
Player 130 ERA+ 100 ERA+ K/9 >8 BB/9 <2 HR/9
<1
RodriguezFrancisc 65% 96% 97% 0% 84%
SantanaJohan 62% 98% 79% 30% 51%
PutzJ.J. 45% 79% 83% 1% 79%
FelicianoPedro 27% 72% 71% 1% 58%
MaineJohn 9% 59% 32% 0% 61%
GreenSean 11% 57% 19% 0% 87%
DessensElmer 23% 52% 4% 16% 59%
StokesBrian 4% 41% 2% 0% 74%
NieseJon 2% 39% 3% 0% 77%
KnightBrandon 6% 43% 29% 2% 45%
SnyderKyle 3% 28% 2% 2% 39%
PelfreyMike 1% 22% 0% 0% 77%
MischPatrick 3% 27% 2% 7% 43%
MejiaJenrry 1% 24% 6% 0% 61%
PerezOliver 0% 20% 64% 0% 28%
FigueroaNelson 1% 16% 5% 4% 42%
SwitzerJon 2% 25% 6% 1% 51%
NieveFernando 1% 17% 8% 0% 47%
ReddingTim 0% 5% 0% 0% 15%
KunzEddie 1% 14% 0% 0% 44%
EgbertJack 0% 4% 0% 0% 47%
LopezArturo 1% 13% 2% 0% 60%
StonerTobi 0% 4% 0% 0% 30%
ParnellBobby 0% 2% 1% 0% 45%
PettyjohnAdam 0% 5% 1% 6% 13%
HoltBradley 0% 6% 20% 0% 12%
BroadwayLance 0% 0% 0% 0% 24%
Extrapolated Career Statistics
Name BA OBP SLG G AB R H 2B 3B HR RBI BB SO SB CS OPS+
CastilloLuis .289 .367 .350 2341 8846 1385 2555 273 75 40 633 1093 1140 457 167 92
ReyesJose .285 .353 .434 2058 8646 1470 2460 459 158 172 804 922 1200 693 186 108
WrightDavid .292 .389 .473 2733 10245 1819 2989 618 38 387 1686 1635 2539 308 106 128
BeltranCarlos .280 .362 .485 2414 9292 1645 2603 545 78 402 1593 1204 1753 389 52 120
DelgadoCarlos .278 .379 .536 2297 8284 1391 2301 536 22 519 1701 1219 1971 14 8 135
SheffieldGary .291 .392 .512 2638 9408 1663 2737 479 28 515 1704 1506 1218 254 104 139
Player W L S ERA G GS IP H HR BB SO ERA+
SantanaJohan 242 146 1 3.44 566 489 3401 3051 411 950 3248 128
All figures in % based on projection playing time
Disclaimer: ZiPS projections are computer-based projections of performance.
Performances have not been allocated to predicted playing time in the majors -
many of the players listed above are unlikely to play in the majors at all in 2009.
ZiPS is projecting equivalent production - a .240 ZiPS projection may end up
being .280 in AAA or .300 in AA, for example. Whether or not a player will play
is one of many non-statistical factors one has to take into account when predicting
the future.
Players are listed with their most recent teams unless Dan has made a mistake.
This is very possible as a lot of minor-league signings are generally unreported in
the offseason.
ZiPS is projecting based on the AL having a 4.46 ERA and the NL having a 4.41 ERA.
Players that are expected to be out due to injury are still projected. More information
is always better than less information and a computer isn’t what should be projecting
the injury status of, for example, a pitcher with Tommy John surgery.
Positional offense is ranked by RC/27 and divided into quintiles based on what the
most frequent starting players at each position did in 2007-2009. Excellent is the top
quintile, Very Good the 2nd quintile and so on.
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.
Sorry, I'd sort of forgotten about this thread. Thanks for tackling my question and you gave reasonable answers. You didn't seem to quite address the money aspects (including possible long term implications of Polanco shifting from 1/$4.5 to 3/$19) and I'm not sure to what extent you're counting on the starters you want signing pretty cheaply but that's OK -- even you might want some time for non-BTF posting. :-) So good work.
Now, Mr JFish ... again, what do you want Dan to do? He's got a thorough complex model. You aren't suggesting any omitted variables that apply widely. You are vaguely suggesting that spike years should be taken as evidence that a player has changed -- this amounts to saying that, in such cases, older data should basically be ignored -- but other research on that question has found little/no predictive ability of spikes or trends. I think Tango once found that spikes (and collapses?) are retained slightly more than the basic model would suggest but then he was working from a simpler model than Dan's I think.
So basically you are asking Dan to go player-by-player and make specific adjustments on whether they have "changed their approach." Unfortunately, we always hear "changed their approach" stories, of one form or another, whenver a player spikes/collapses only for those seasons to be followed by normal ones.
But yes, Dan's model, like basically any model, can only provide you with a mean projection. Because of the "randomness" of baseball, these projections are not very precise for predicting individual player performance. A good but simple interpretation of the Zobrist projection is "based on past data, players with a weighted 4-year performance similar to Zobrist at Zobrist's age have averaged about a 113 OPS+." Obviously some subset of such players ended up being much better than that. You, whether based on personal qualified observation or magic 8-ball, think Zobrist is one of those guys. There's no problem with that. But unless you can come up with sufficient evidence why the rest of us should expect Zobrist to be better, then that's your personal opinion. And unless you can come up with a systematic reason (which is measurable for all players) why players like Zobrist should be expected to regularly outperform their projections, then you're not contributing to projection models and you really don't have any reason to complain.
Yes, I don't see the Mets picking up two SP, especially if Lackey is one of them. They have Niese, Nieve and Figueroa already for depth and I don't believe they would put Maine in the bullpen. I know there was talk of it at the end of the 2008 season, but it seems crazy to put the guy with 3 good pitches to the pen. If any starter should go there, it should be the guy with 1 good pitch (Pelfrey).
I would be surprised if either the Mets or Cameron was interested in a second tour of duty. Cameron seems like a great guy, but he is also going to be 37 in January. I would not want to commit either the dollars or years he is likely to get.
Just for the record, I'm against Bengie Molina and Orlando Hudson. Both of them will cost too much and add too little.
I understand a stat centric site would believe statistics tell you everything you need to know about a player but that just isn't true and, just like the people that believe sabermetrics and stats in general can't project a player, you are wrong if you think that. I also don't think it is fair to categorize any anomaly as “randomness” because there are often very good reasons for the deviation that even the best statistical analyst can’t quantify with a number. Don’t get me wrong, I’m not anti numbers but there should be a mix and I’m not seeing any weight being given to anything outside of a spreadsheet.
Treating every player the same, ignoring recent reports, and trying to strive for a league wide mean instead of accurate individual projections isn't a good system in my opinion. Sure, I'm not a stat head so most of you couldn't care less what I think (including Dan) but I think it is ridiculous for you to throw some numbers into a spreadsheet and look down on anyone that argues the values it spits out as if it is God handed dine divine providence. If I need an example of a player with numbers that exactly mirror Zobrist for you to acknowledge the fault here than that will never happen. It's pretty convenient to hide behind a stance like that since I doubt there is a player with exactly the same statistical and personal profile.
I don't expect you to agree but to just ignore the other side of player evaluation is a glaring weakness in the work being done here. You should never get so comfortable with what you are doing that you can't still strive for improvements, trying to incorporate the other side of player evaluation could be a pretty good start in my opinion. If I'm wrong about the system treating every player the same despite most players being unique or about you ignoring scouting, then feel free to set me straight. I imagine all I'm going to get is more "you don't understand" or "you don't have any evidence".
In addition to the "CTA" factor, you would also need the "BSoTL" factor.
And how would you know if the increased performance is because of "CTA" or just "Randomness As Usual", if you will? My gut feel is that you would inject more error into the system than you would fix.
I think you miss the general point of modeling, in all fields. Scouting data is inherently subjective and undependable.
I can't tell you how many times I heard you argument for some of the the projections that I'm happiest ZiPS nailed. If I had a dolla for every time someone told me "No way Nelson Cruz will hit that well, Major League curveballs will eat him alive" or "Your projection for Gary Matthews Jr. is terrible. He's clearly put it all together," I could probably buy a team.
And I'd wager you were off on just as many, you just aren't as concerned with those evidently.
Of course I already got the response I figured I would get, "you miss the general point of modeling". I repeat, you aren't trying to be accurate, you are just trying to be more accurate than the next guy by being in the ball park hoping to hit on a few. It's amazing that as long as the stat community has labeled baseball traditionalists as thickheaded, you are equally dense and stubborn.
I'm the guy who runs the CHONE projections at baseballprojection.com. I would be very happy to work in scouting reports to the projections. I don't have them though. Could you get them for me? There are about 1500 hitters on my projection pages. I would like a scouting report, showing the standard 2-8 scale for hitting, hitting for power, running, throwing, and fielding. I'll need them for the last 4 years, so if Zobrist's scouting report has changed I'll be able to model expected preformance increase from that.
I will need all of these linked to the MLBAM player ID codes. If you can get that to me by 8:00 tonight, that would be great.
No one here thinks this.
Treating every player the same, ignoring recent reports, and trying to strive for a league wide mean instead of accurate individual projections isn't a good system in my opinion. Sure, I'm not a stat head so most of you couldn't care less what I think (including Dan) but I think it is ridiculous for you to throw some numbers into a spreadsheet and look down on anyone that argues the values it spits out as if it is God handed dine divine providence.
It's just a question of objective vs. subjective. Dan's projections are objective and unbiased. If you want to insert your subjective opinion, and change them for your own use, knock yourself out. You may well be right - or you may be wrong.
Speaking of which, what do you say to a little wager to test out your theory? Pick a number of Zobrist's OPS+ for 2010, what you would change it to if you were altering Dan's projection, and I will take Dan's projected 113 OPS+. Whoever is closest to Zobrist's final OPS+ for 2010 wins, subject to a minimum of 400 PA. The loser has to sponsor a $10 baseball-reference.com player page of the winner's choosing. Do we have a deal?
Which reports, exactly? Where are these coming from?
I think it's laughable the stance you are taking and essentially mocking me for suggesting you could improve your work. I guess it is a generational thing that you feel you are superior and can't improve? Of course we are probably from the same generation so that can't be right. Strasburg is an extreme example but you make up data all the time that you don't have every time you project a minor league player so you are being pretty hypocritical to act as if you couldn't take scouting reports and incorporate them into your approach.
And no I'm not going to wager $10 on a bet, you forget that I'm not an analysts or someone pretending to be an expert. I'm simply a baseball fan that enjoys baseball analysis trying to have a conversation about the possible benefits of trying to incorporate the other half of baseball analysis into an already pretty good projection system. If you want me to pick a number and see who is closer, fine but I won’t be putting money down and that still wouldn’t change my opinion. Zobrist is one player, Dan’s whole system banks on being close more often than being far off, not being accurate. For your little test to be realistic we would both need to pick a number and both be in the same margin of error.
There are plenty of reports easily accessible Lassus and, I imagine, you probably read some yourself.
Actually, it'd be more useful if you could have 30-40 years worth (that way, you could study just how much a player with 70 power outperforms someone with 60 power but similar statistics).
Again, it's a lot like hurricane forecasting. Meteorologists come to different conclusions from the model runs, but all them want those models to be objective. When one of them is looking at, say, the latest GFDL run, they need to know that that the model is objective, even if it doesn't account for every possible variable. If the GFDL might contain some guy's opinion that the hurricane totally reminds him of Hurricane Camille or some other guy's opinion that it feels like the storm will curve, the GFDL model becomes a less valuable tool, even if the majority of random subjective opinions inserted are correct.
Now, which ones are based on objective information. And if they're based on subjective information, what is the historical success rate of these particular subjective sources?
He's not saying that.
And no I'm not going to wager $10 on a bet
That's fine, I just meant for it to be friendly. We can not bet any money or just forget the whole thing, whatever you want.
If Chone doesn't feel Beltre mirrors Zobrist than why use him as evidence Zobrist can't remain a very productive player?
I'm fine for a friendly game but the conditions you set aren't really why I even started the discssion. Trying to see which is more accurate with you taking ZiPS and me throwing a number off the top of my head when the community's evaluation of the success of ZiPS being the number of projections that are within a window of margin for error isn't going to prove anything. I'll take 123 I suppose if I had to throw a number out. :)
- JFish knows Zobrist personally.
- JFish has Zobrist in a keeper fantasy or sim league.
The amount of angst (spilling over into multiple posts!) over Zobrist "merely" being projected as a star player instead of an MVP candidate is simply unbelievable. You'd think that ZIPS projected Zobrist to have a .600 OPS or something.
It's all pointless though, very few of the posts since my initial one have been constructive on either side due to the difference of opinion and approach towards evaluating. I'd wager He's Bought a Bat Like Prince Fielder wouldn't listen to anything I have to say even if I had scouting reports for every player to play baseball in the last 40 years to back up my assertion that Dan should at least look to make some adjustments based on recent reports. :p
I don't know why it doesn't add up - it's entirely consistent. In both cases, Dan weighed data other than merely from the player's most recent season.
Actually you are saying he should basically ignore the earlier data on Zobrist and seriously underweight the earlier data on Wright.
but take into consideration extra data like scouting reports.
As soon as every team shares their scouting information on every player with Dan, he'll likely do so. Otherwise you are asking him to go player by player, do web searches for media reports of "changed approaches", assess whether he believes said media reports, then make subjective changes to player projections for those players which he can find sufficient media information on while leaving the other players unchanged. I assume you can see that's neither feasible nor desirable.
If your point is that teams should mix statistical data and scout observations in assessing a player -- well, duh.
It isn't a matter of me suggesting Zobrist made changes, it has been documented by scouts, team officials, and reporters not to mention Zobrist himself.
This is no different than all the other times this sort of thing has happened. In 2003, we read endlessly about Esteban Loaiza adding the cut fastball to his repertoire and this explaining his 159 ERA+ that year. The next year he was down to 82 and we weren't hearing about his fabulous new pitch anymore.
Look, Zobrist may have changed his approach. His new approach may be hugely successful. It's also possible that, now that pitchers have seen his new approach, they will make adjustments and he'll end up right back where Dan says. Who knows? But his mean projection is a 113 OPS+ and you have to refute thousands of data points to show otherwise.
It's very simple. The standard ZiPS weighting scheme is 8/5/4/2 with further adjustments for all sorts of things. That puts Zobrist at 113. You think Zobrist is better than that -- good for you. The only evidence you've presented in support is references to media reports that he's changed his approach. I'm sure he has changed his approach -- and this changed approach has led him from being a guy projected to be replacement level or worse as a hitter to a guy projected to be one of the better-hitting 2B in baseball, a fact picked up on by ZiPS. But you've presented no relevant evidence that he's better than that other than his recent performance -- which is accounted for in the model.
He essentially ignored Wright's most recent season giving him a projection above his career averages minus 2009 and penalized Zobrist for past seasons (partial seasons at that).
First of all you realize Dan is not making decisions on individual players right? You understand his role is to build the best model he can, right? And the model doesn't ignore Wright's most recent season at all -- of course it gives it the greatest weight of any information just as the model does for every player. And it doesn't "penalize" Zobrist for past seasons, it properly weights past seasons in its projection. Are you suggesting we ignore past seasons for every player? Or only your personal pet peeves?
That doesn't add up to me. How does Wright have numbers above his career average (projected to have a 140 OPS+ for 2010 even though his career average from '04-'08 was 136) before 2009 when his most recent season (the one that should be weighted the most, correct?) is well below in every category?
Because funny things happen on the way to the projection? Dan doesn't actually model OPS+ of course but the components. Dan doesn't model OBP and SLG directly either I don't think, but, if you apply the 8/5/4/2 weight to Wright's OBP, you get a projection of 395, above his career average. Apply it to his SLG and you get 499, which is below his career average. He's moving into his age 27 seasons when players tend to peak so there is reason to think that both of those numbers will be a little higher than that. Dan is projecting that a guy with a career 136 OPS+ through age 26 is going to improve slightly at age 27 -- nothing too surprising about that really.
Now look -- I think Wright's projection is probably too high as well. But so what? That's my opinion. Until I present solid evidence to think that about Wright but not others with Wright's projection, I'm not going to convince anybody else. Until I can present a systematic explanation for why Wright _and other players like him_ are over-projected, I am in no position to criticize the model. I don't have to believe everything the model says (I'm a sentient human being after all) but "I think you're wrong on David Wright and some bozo who writes for the NY Post agrees with me" is not valid criticism of the model.
I understand a stat centric site would believe statistics tell you everything you need to know about a player
Oh grow up.
there are often very good reasons for the deviation that even the best statistical analyst can’t quantify with a number.
Anything can be quantified. True, often only crudely. But Dan can easily build into his model an indicator variable that says "jfish thinks this guy is super-duper" and see if that improves the explanatory power of his model. He can put in an indicator variable for "some sportswriter noticed this guy was having a real good season, wrote a story in which the player gave credit for his improvement to [insert hitting coach, pitching coach, hypnotist, new training regimen, new diet, new girlfriend, new wife, new child, finally getting over the divorce, new city, returning to old city, father passing away, father recovering from illness, or -- most unbelievable of all -- changing his approach due to FIP].
Nobody who models "believes" models. There are big error bars on any projection. Dan provides rough guidelines to those errors right in his report. What is your confusion?
Meanwhile, while you accuse others of believing only in statistics, you apparently are naive enough to believe media reports. Given a choice between a good model based on tons of data vs. a series of post-hoc media explanations for something unexpected, every sentient human being should lean very heavily towards the model.
Don’t get me wrong, I’m not anti numbers but there should be a mix and I’m not seeing any weight being given to anything outside of a spreadsheet.
Because this information is not available on every current player much less players from the past. This information is highly subjective and of low reliability (i.e. very high variance, surely worse than a model) and, by the time it gets to you or me or Dan, has been filtered through half-a-dozen other perspectives and is completely worthless.
But teams? Teams can easily build scouting information into their models. I can show them how to do it in less than an hour if they don't know already. It might improve their player projections and it could also be a great piece of information in evaluating which scouts are any good.
Treating every player the same, ignoring recent reports, and trying to strive for a league wide mean instead of accurate individual projections isn't a good system in my opinion.
Propose a feasible alternative and show it works better.
I think it is ridiculous for you to throw some numbers into a spreadsheet and look down on anyone that argues the values it spits out as if it is God handed dine divine providence.
Oh grow up.
If I need an example of a player with numbers that exactly mirror Zobrist for you to acknowledge the fault here than that will never happen.
That would be useless anyway so don't waste your time. You really don't understand how this works, do you?
What you want to provide are, at a minimum, good convincing arguments that Zobrist should out-perform his projection. You don't seem to have any other than media reports of a change in his approach. You don't appear to have any claim to expertise in hitting technique and are not offering personal observations of this changed approach nor arguments for why such an approach should be so successful. You've shown no grasp of baseball history and the many, many times a player popped up with an amazing season only to fall back heavily the next year -- note in Zobrist's case he's projected to fall back to being one of the, what, 5-best hitting 2B in baseball?
For crying out loud, I think I was about 13 when I realized that one year of baseball stats is pretty much meaningless and concluded that I needed to see a player do it two years in a row before I believed in him. Older and wiser I realize it's more like 3-4 years.
I'd imagine I was 12 when I realized media reports were completely unreliable.
I don't expect you to agree but to just ignore the other side of player evaluation is a glaring weakness in the work being done here.
You realize you haven't provided a single piece of this "other side of player evaluation" right? You haven't seen Zobrist's scouting reports have you? You aren't a trained baseball scout are you? All you're basing your opinion on is 1.5 seasons of data and a few stories in the paper. Your evaluation is every bit as numbers-based as the model. You've simply chosen to ignore his earlier numbers and the justification you've offered for doing so is thoroughly unconvincing in my opinion.
You should never get so comfortable with what you are doing that you can't still strive for improvements
OK, now you're an #######. Dan works his ####### ass off improving his system. Not only does he do that, but he shares his results freely with the public. He also builds a database that other people can use to run their own simulations and use to play Diamond Mind with. Where the #### do you get off?
trying to incorporate the other side of player evaluation could be a pretty good start in my opinion.
How? How? How? For every player in professional baseball. With reliable observational evidence. How?
"you don't understand" or "you don't have any evidence".
It is clear you don't understand and you don't have any evidence. You suggest we listen to your criticism -- which amounts to "scouting information could be useful" which everyone agrees with already. Meanwhile you really might want to listen to criticism yourself. You don't understand, you are not offering evidence, your criticisms of the model are baseless because you are unable to point to any missing factors.
You are of the opinion Zobrist is better than that. Good for you. You may be right. Get back to us when you've been correct about 100 players and wrong about only 50 (warning: I will require the players to over/under-perform for at least 2 seasons, preferably 3, before I accept your evidence) and I'll actually believe that you have some insight.
Meanwhile I'll gladly make a bet with you for a $10 b-r sponsorship. If Zobrist puts up a 120 OPS+ or better in 500+ PA, you win and I sponsor the $10 page of your choice. If it's less than that, I win and you sponsor the $10 page of my choice. If he doesn't make it to 500 PA, we both sponsor a $5 page. (this way b-r wins regardless)
On Zobrist, My projection is that he will be a very productive player, but not nearly as good as he was in 2009. I did not say that he would mirror Beltre's career, but that is one of many examples to illustrate that using multiple years is better than expecting a player to repeat his career year.
I think value could be found in scouting reports, but I'm projecting 1500 players (just the hitters alone). My guess is Dan's list is about the same size. I want YOU to take some time to think about what kind of work would be involved in adjusting that many projections based on some kind of consensus on scouting opinions.
I don't think you have the slightest clue what kind of time would be involved. This is really rich: "I think it's laughable the stance you are taking and essentially mocking me for suggesting you could improve your work. I guess it is a generational thing that you feel you are superior and can't improve?"
So you suggest I spend a few thousand hours to improve my work, I say no, and you think it's because I have some false sense of superiority?
So is this NEVER done? Players are simply put into the formula, and whatever it spits out is the projection we see here? From most of what I read (including the interview over at FanGraphs), this is correct and makes sense.
But, I ask only because statements that Dan has made himself have made me wonder. Such as:
When asked about Brad Hawpe's projection: "Hawpe's just not the type that ages well. When his bat speed slows just a tinge, his $H is going to drop like a rock. He was 15 points above his normal as it is."
Is there something in Zips that says all players at age 30(31ish) with a body/swing/etc like Hawpe will show this type of digression and post numbers under anything in his last 3 or 4 years? Or are there some red flag type things that tag him as a 'type' of player and the system recognizes that...warranting a certain drop in Avg., OBP, and SLG (~20 points in each for Hawpe)? Are there other examples we can see showing similar results?
I should say, I actually like the projection for Wright showing improvement at an age of progession/peak, especially with his seasons prior to 2009.
When asked about Brad Hawpe's projection: "Hawpe's just not the type that ages well. When his bat speed slows just a tinge, his $H is going to drop like a rock. He was 15 points above his normal as it is."
I didn't mean to send mixed messages. I don't rely solely on ZiPS to evaluate things.
In Hawpe's case, ZiPS seems High BABIP, Mid BA, Low Singles, Fairly Slow Batters in Early 30s as big risks for dropoff.
When I'm referring to his bat speed slowing, that's not part of any projection system, simply my observation as to why it's the case.
I can give you more examples in a bit - I have to catch up on the news of the day.
I tried this already Walt - I would take the bet with you myself, just for the fun of it, but I agree with every last thing you said, so...
Come on, AROM, we pay ALOT of money for those projections you post. It's literally the least you could do.
I think you're giving in too much - he's suggesting you spend a few thousand hours to alter your work. He has no idea if it would improve your work, because he has no idea how you should quantify the various scouting and media reports he wants you to incorporate into your model so as to actually make them predictive.
I'm just trying to have a discussion, why are you being ridiculous Walt? I must have struck a nerve and said something right about the superiority complex going on here. 140 or so posts and you have to go on a expletive tirade? That isn't doing anything, perhaps you are the one that needs to grow up? Haha
When I have some time, I'm going to take Dan's disk and compare their OPS+ and ERA+ (since that is sort of how he sorts them by here) and see how many he actually hits. You say he is right 100 times for every 50 he is wrong, we will find out. ;)
Edit: By the way, I've never said Zobrist will duplicate the season he just had but it's quite a drop off being projected. I think you all took a couple lines, made up your minds about what I was trying to say, and ran with it. :( I'll work up a spreadsheet some time in the next month with 08, 07 and whatever disks I can find (and have the time to do) to see.
Heyman also says the Mets are suffering "sticker shock" over what some of that tier might sign for. Wasn't it common knowledge that Wolf would get something like 3/27?
Or, let me just sum the whole thing up in five words for you...That's because it's a spreadsheet.
Still waiting on the Valtrex to kick in.
Exactly. And...
So, JFish, that is that. Nobody looks at these as gospel. In fact, I LIKE the fact that they don't know what I know. It gives me a reasonable basis to start with, and I can tick up or down from there based on what I think I know about the player that the model doesn't. The important part is that the original projections are just that - simple, mathematical regressions w/o contamination.
That said, I got through looking at the Mets pitchers annnnnnnnnd it's time to do another round of analysis, Goldilocks-style:
Too Optimistic
Francisco Rodriguez: 3.13 ERA, 72 IP, 37 bb, 88 k
John Maine: 4.20 ERA, 122 IP, 57 bb, 103 k
Jennry Mejia: 4.90 ERA, 75 IP, 44 bb, 55 k
Too Pessimistic
Oliver Perez: 4.93 ERA, 142 IP, 87 bb, 132 k
Nelson Figueroa: 5.06 ERA, 131 IP, 46 bb, 91 k
Brian Stokes: 4.54 ERA, 81 IP, 38 bb, 56 k
Mike Pelfrey: 4.86 ERA, 189 IP, 69 bb, 111 k
Just Right
Johan Santana: 3.23 ERA, 206 innings, 52 bb, 203 k
Jon Niese: 4.57 ERA, 143 IP, 55 bb, 103 k
Any thoughts?
What constitutes being right or wrong? If you mean exact match (OPS+ projected: 120, actual 119) then he'll be right about 2-3% or the time I guess. Give him 50 points either way and he's probably right 95% of the time. Kind of meaningless without first declaring your criteria.
But that's not what Walt was talking about. He's suggesting you take 150 players. Tell us which ones are better than ZIPS says and which are worse. And if you're right 2/3 of the time at the end of the year (or 2 years to satisfy Walt) then you've provided some evidence that you know things that the computer doesn't, that can improve projections.
Agreed. I was starting to feel like I was the only one crying Wolf (sorry).
It's particularly interesting to me that you believe the SP is in real trouble despite believing, unlike most, that Ollie has a pretty good chance to be a pretty good pitcher in 2010.
Santana, 133 ERA+
Maine, 103 ERA+
Pelfrey, 89 ERA+
Perez, 88 ERA+
Niese, 95 ERA+
Then, of course, SOMEONE is going to get injured. At least one person. Probably a couple. Every replacement guy on the list below is projected to an ERA above 5 (except Mejia, who won't see Citifield this year).
SP6: Figueroa, 86 ERA+
SP7: Nieve, 83 ERA+
SP8: Stoner, 78 ERA+
SP9: Parnell, 76 ERA+
SP10: Mejia, 88 ERA+
The Phillies have TEN pitchers projected to have an ERA better than 5.00 (incl. Moyer, Pedro, and even Rodrigo Lopez). Even the Nationals have five -- and their starters 6-10 have ERA+ marks projected around 85, 83, 81, 81, and 80. Yes... better than ours.
I say this every year - but there are going to be at least eight guys starting for the Mets this year, and probably closer to ten or twelve. In order to be competitive, I think we're going to need one of Pelfrey/Perez to beat their projection significantly, AND to add a starter, AND for Maine to be healthy (or to add a second starter). But that's just me.
Basically, you think the Mets should add two starters because there's basically no chance of all those things happening. I agree.
Agree with this.
Basically, you think the Mets should add two starters because there's basically no chance of all those things happening. I agree.
No chance? Disagree. But to count on it happening would be a mistake. Unfortunately, the alternative involves signing Joel Pineiro and Jason Marquis. So I'm not sure the aim shouldn't be to hope that fortunate scenario happens, at least get a better sense of what you have in Perez/Pelfrey/Maine, and head into 2011 either bolstered by their emergence, Niese/Nieve surprising, Holt/Mejia taking leap forward, or getting someone/someones from what has to be a more inspiring free agent class.
Also, I had no idea Dan hated Ben Zobrist so much.
How about something along the lines of Jon Garland and Pedro Martinez? You'd probably be able to get those guys for less than 15 million altogether. Garland is as good a bet for 175 innings as anyone and I think he can be counted on for a 95+ ERA+.
I think Pedro's 2008 season was something of fluke (his father's health issues might have affected his game more than he let on) so I think he'll be good for whatever number of innings he throws.
That's something that allows the Mets to "go" for it this year without giving up too much of the future, and not locking themselves into long term deals.
I think it's a little sad how often the long baseball related threads consist of lots of sensible people trying to set some moron straight.
Oh yes, and of course I add the obligatory thanks to Dan for putting in the time to provide us all with such great info.
Arroyo is about the same pitcher as Jon Garland value wise although they get things done differently. If you think Garland is going to collapse because of the lack of strikeouts, Arroyo is a better option than Garland. Arroyo will give you 200 averagish innings and he's signed at 1y/13m or 2y/22m if you pick up the option.
Harang put up 142/43 k/bb ratio in 162 IP last year. He didn't keep the ball in the yard but a guy with those K/BB ratios is always interesting. He is owed 1y/14.5 or 2y/25 if you pick up the option.
If the Reds truly want to dump them and are willing to eat some money, I wouldn't hate either move. That's a medium-sized if and even if they were, they might not trade them to the Mets. But there's reason to believe they'd be good moves.
That said, I wouldn't mind just signing Garland.
I think Perez beats that projection (although Pelfrey may not). I also think Maine beats that, but you have to balance that out by wondering how many innings he can be counted on. But if the Mets add Lackey and have three guys (Niese, Figueroa and Nieve) ready to step in and give respectable numbers, I would be okay with that. I'd rather have Lackey's superior talent than two guys who are likely to be league average (or slightly above) but who can deliver 350-400 IP. Adding two starters should be the backup plan if they can't sign Lackey.
I'm not worried about the Phillies having 10 reliable pitchers. There's no guarantee that they re-sign Pedro and counting on 47-year-old Jamie Moyer (coming off an injury) for any quality innings seems unwise.
Except that Moyer is one of several options for a #5 starter, which isn't quite the same as "counting on".
I'd love to see the Phils make a pitch for Martinez with the idea that he and Moyer would be spot starters, maybe 20-22 starts each and 12-15 relief appearances (all the long relief needed plus an occasional inning) to the tune of 125-150 innings each. Both seemed to do best last year when they got extra rest. The Phils would use a 4+ pitcher rotation, with M&M;sharing the 5th slot and filling in for the inevitable injury.
It might be divisive in a "everybody knows their role" mindset, but if any team could handle it, I think the Phils players could and Manuel and Dubee could manage it.
EDIT: And if I were the Mets, Martinez would seem to be a good fit, as long as you know you won't get 30 starts or 200 innings.
Yeah, but only 1 year. Even if he's a complete bust, he doesn't hamper the Mets ability to retool in 2011.
Granted, I don't think any of you are necessarily penciling in 180 innings of 4 ERA ball from Perez, and no one is making any grandiose, Megdalesque predictions, but if I were running the Mets, I would consider anything positive I got from Perez next year as a bonus, and no way I would enter the season with him as anything but my fifth starter.
Obviously, last year was screwed up by the WBC and being out of shape. But he was already showing signs of improvement before getting hurt in his next to last start. In 7 starts after the All-Star break, Perez had a 4.19 ERA. Then he got hurt and tried to pitch and had the disastrous start against Philly before they shut him down for the year.
There have been no reports of any setbacks from his knee surgery and when he had the operation the prognosis was that he would be ready for the start of Spring Training. The only thing we've heard is that for the first time he is staying in this country during the off-season, where the club can do a better job of monitoring his progress.
I think 175 IP and a 4.50 ERA are reasonable projections for a healthy Perez. That combination of IP and ERA in the National League last year gave you something along the likes of Johnny Cueto, Paul Maholm and Derek Lowe, or a pretty good 4th SP.
Those are lame ass excuses. He sucked royally last year and he might do the same in 2010.
Honestly, there is no way to make a thoughtful prediction for Ollie. He is a nut and he is also a huge question mark in terms of health and stuff.
not getting on you, PF, just reflecting.
Yes, and this is why he's better than Harang, who's guaranteed to suck. He'll bounce like a dead arm dropped from six feet on a pile of gravel.
Even this is optimistic, imo, but with the plethora of 5th starters the Mets have available (Figueroa, Niese, Nieve, Maine) the backup plan's already in place in case Ollie can't make it out of ST.
With a year as utterly godawful as Perez had in 2009, and with his long erratic history, doesn't the default position of any FO that doesn't want to wind up with a 180 inning hole in its rotation have to be that anything he contributes is gravy?
That's why, given the choice, I'd go with Arroyo. Harang will never again be the pitcher he was from 2005-2007, so why chase a ghost that will never materialize?
Mr. Smith, not sure but are suggesting a projection should be considered accurate if it is within 50 points of OPS+? Being within 2 standard deviations of the actual right number isn't accurate to me, I wouldn't consider a projection within half of that accurate. Being labeled "projections" infers that these numbers are what Dan (or at least his system) thinks the player will do next year so the actual result should be very close to the projection. Less than one standard deviation, if Dan knows that maybe I can give a criteria and try putting something together if I have time.
You keep missing the point. It's not that people think this is a bad idea. It's that neither you nor anybody else has any good idea as to how to gather and incorporate this information in a systemic way that actually improves the accuracy of the predictions.
You take the values it spits out and make small adjustments. I'm not saying take someone like Zobrist or Wright and increase/decrease their OPS+ by 20-30 points, even the number I threw out was only 10 points higher. Just suggesting small adjustments on numbers the clearly look too low/high based on known factors like changes in park (which may be incorporated already?), league (from AL to NL which may also be incorporated?), injuries, trends, etc. I fully understand it is less objective then letting a computer tell you what the player will do but the stuff does matter in projecting players future performance.
That's what people are saying.
Ah, my 16 wins and 3.5-3.7 ERA prediction for 2007 haunts me still. What did he end up with that year, anyway?
Then the model is obviously flawed not being able to take into account something as important as scouting into the projection. If everyone agrees scouting is important, why the hell would you put any stock in a projection system that ignores it?
Sorry, Howard. It's less that prediction than the Koufax thing (which, admittedly, some of us took out of context a little).
Anyway, don't make me bust out his xFIP or RA+ for 2007. ;-)
Just suggesting small adjustments on numbers the clearly look too low/high based on known factors like changes in park (which may be incorporated already?), league (from AL to NL which may also be incorporated?), injuries, trends, etc.
Again, not to speak for Dan, but Park and League areincorporated already. Don't know about injuries, but doubt it. Trends? Well, besides the fact that this is already implicity incorporated in the projection, insofar as the main input is stats from the past four years, but three year trends (e.g. Derek Jeter's declining OPS from 2000-2002) are generally NOT predictive, unless there is a *clear* underlying reason, such as age, and that would already be incorporated as well.
Here's a link to an article that talks about Zobrist.
Within the article, the interviewee gives a scouting opinion on Zobrist, saying that he sees his power dipping next year, because he likely doesn't have the physical stregth to maintain that from year to year. This is "scouting" information as you're defining it, right? So how do we account for it? Should I adjust Zobrist's ISO down a bit? How much? 5%? Dave Cameron's more of an analyst than a scout, so maybe I discount it? So maybe 1 or 2%? Or do I ignore it because it went against my preconceived notion of what Zobrist's 2010 will look like? And how do I consistently incorporate it? Do I tag anything from "Dave Cameron" a certain way? Do I give him more weight if he gives an opinion on the Mariners, because that's the team he follows most closely? Or maybe I don't, because he'd be biased?
Then the model is obviously flawed not being able to take into account something as important as scouting into the projection.
All models are wrong, some models are useful. Dan (and Sean's, and others) projection systems will always be wrong sometimes. If they were able to get (as Dan, I think, noted) *actual* scouting reports, you know, from scouts, not something mentioned offhand to a reporter that makes its way into a Notes column and may or may not have been taken completely out of context or changed slightly, AND you could come up with a framework for how to numerically quantify said scouting reports, then take the time to incorporate this into your process, well, sure, you could probably improve projections this way.
However, given the hurdles in properly, scientifically incorporating that data, not to mention getting information that isn't filterted through the press and for all we know was only released to the reporter in question because they are posturing for a trade or a contract negotiation or something, why is it so hard for you to understand why these systems don't incorporate this stuff.
But models are objective. They take into account proven factors that improve the accuracy of the projections. The problem with subjective observations is that we don't know that they will make the projections better. If you go through the model and apply adjustments based on one's subjective opinions on certain players, it ceases to be a model.
The value of a model is that it gets rid of the very noise that you're trying to introduce. It says here's our best guess as to what the player will do based on quantifiable factors. You think Zobrist will maintain his performance because he changed his approach. But maybe there's a hole is his new approach that pitchers haven't yet figured out how to exploit. We just don't know the answer yet. A model relies on what is known to be true. You can make your own adjustments based on what you think to be true. But what you think has a pretty fair chance of being wrong.
I'd still like to know what Dan/Chone think is acceptable to still be considered accurate because I'm curious just how accurate/inaccurate it is. If we are using OPS+, I think within 8-10 is fair. 50 points is the difference between an average hitter and a MVP type hitter...even 20-30 is the difference between an average hitter and an All-Star.
Again, the short answer is that if you follow your method here, you don't know if you're making sense of the noise, or just adding more of it. You just. don't. know.
I think the value is that it gives you the best guess of a player's projections based on what is known to be true. Then I can take my own subjective opinions and make adjustments. If the model incorporated Dan's subjective opinions, I'm less able to do that because I wouldn't know what projections are based solely on objective data and what projections are based on Dan's subjective opinions of unquantifiable data. I may trust my own subjective opinions, but I have no idea of the value of Dan's subjective opinions. If I'm putting together a model just for me, I can incorporate subjective elements and be perfectly happy. But if I'm evaluating a model done by somebody else, I want it to be purely objective because I don't know if somebody else's subjective opinions actually do improve the projections.
This is not saying the system is inaccurate, because we don't know if anybody's subjective opinions actual help the model. We're saying the system is as accurate as it can be given what we know.
If he pitches anyhting like he did for the Tokyo Yakult Swallows, the Mets will be OK.
If the Mets are signing him, he'll be a backup catcher in 2010.
Nice.
I'm worried about the unknown knowns. These are things we know, but don't realize we know. Like how uncool we really are.
And the Mets landed him! AP LINK
I think he needs a "Japanese _____" in order to identify him. Omar says he has a splitter. I suggest "the Japanese JJ Putz."
If you had only listened to those scouts, Dan.
You must be Registered and Logged In to post comments.
<< Back to main