This is our old blog. It hasn't been active since 2011. Please see the link above for our current blog or click the logo above to see all of the great data and content on this site.
«
»

Can SPM Predict the Playoffs?

Posted by Neil Paine on May 4, 2009

A week ago, we delved deeper into our new Statistical Plus/Minus results and tried to retrodict the standings for the regular season, coming within a healthy +/- 6.6 wins per team. Now the question is, can we put the same method to the test for a playoff series? Let's use the recently-completed first round as a test case...

(I should point out that I fixed the Derek Fisher/D.J. Mbenga error that had plagued last week's release, as well as crafting new version of the projection formula by throwing out incoming 1st-year players, which significantly improved predictive accuracy for each player. The new formula and its results can be found on the new spreadsheet, available for download.)

The main exercise here will be using projections for 2010 to form a measure of "true SPM skill" in 2009. Borrowing a concept from the White Sox chapter of Baseball Prospectus 2006, we could describe the simple SPM projection system like this:

2010 Forecast = 2009 True Ability + Age Adjustment

or

2009 True Ability = 2010 Forecast - Age Adjustment

Look at that! We all can do elementary algebra! Anyway, the premise is very basic: a player's "true skill" this season is equal to his projection for next season (which is in turn made up of a weighted average of the past 3 seasons' worth of data), but before you adjust up or down for age. Since the simple projection system handles the age adjustment in such a, well, simple fashion, it's pretty easy to find each player's "true skill" level in this manner.

Now it's time to put them to use in retrodicting the first round of the playoffs. Remember, the team minute totals are the only things known to us after the fact, because projecting playing time is another laborious endeavor for another post. But we mainly want to see how the true skill levels (in terms of a per-minute rate) did at predicting the performance of this year's playoff teams (converting predicted +/- to Win%, and running log5 projections for every game, assuming a constant 60% home-court advantage):
East
1. Cleveland vs. 8. Detroit

Player Ag Tm Ht Pos Skill Min
DelonteWest 25 CLE 76 G 0.51 164
LeBronJames 24 CLE 80 F 12.76 162
MoWilliams 26 CLE 73 G 0.79 148
AndersonVarejao 26 CLE 82 F 0.64 133
ZydrunasIlgauskas 33 CLE 87 C -0.34 111
JoeSmith 33 CLE 82 F -1.91 81
DanielGibson 22 CLE 74 G -1.25 60
BenWallace 34 CLE 81 F 1.15 45
WallySzczerbiak 31 CLE 79 F -1.53 38
DarnellJackson 23 CLE 81 F -3.35 10
SashaPavlovic 25 CLE 80 G -2.94 4
TarenceKinsey 24 CLE 78 G -2.97 3
Cleveland 10.57 0.845
RichardHamilton 30 DET 78 G 0.30 154
AntonioMcDyess 34 DET 81 F 0.74 136
TayshaunPrince 28 DET 81 F -0.35 129
RodneyStuckey 22 DET 77 G -0.41 128
RasheedWallace 34 DET 82 F 3.33 122
WillBynum 26 DET 72 G -1.06 78
ArronAfflalo 23 DET 77 G -2.87 66
JasonMaxiell 25 DET 79 F -1.09 64
KwameBrown 26 DET 83 F -1.08 48
WalterHerrmann 29 DET 81 F -2.60 22
AmirJohnson 21 DET 81 F 0.84 13
Detroit 0.08 0.501
*Cleveland in 4* 48.6%
Detroit in 4 0.1%
Cleveland in 5 34.6%
Detroit in 5 0.1%
Cleveland in 6 11.1%
Detroit in 6 0.5%
Cleveland in 7 4.4%
Detroit in 7 0.5%

2. Boston vs. 7. Chicago

RajonRondo 22 BOS 73 G 3.87 317
PaulPierce 31 BOS 78 F 3.66 311
GlenDavis 23 BOS 81 F -1.43 291
RayAllen 33 BOS 77 G 2.34 280
KendrickPerkins 24 BOS 82 C -0.76 265
EddieHouse 30 BOS 73 G 1.17 103
StephonMarbury 31 BOS 74 G -2.38 82
BrianScalabrine 30 BOS 81 F -2.24 80
MikkiMoore 33 BOS 83 C -2.11 49
TonyAllen 27 BOS 76 G -0.81 48
LeonPowe 25 BOS 80 F 0.80 24
BillWalker 21 BOS 78 F -1.94 3
GabePruitt 22 BOS 76 G -2.16 3
Boston 5.42 0.676
DerrickRose 20 CHI 75 G -2.21 313
JohnSalmons 29 CHI 79 G -0.64 313
BenGordon 25 CHI 75 G 0.06 304
JoakimNoah 23 CHI 83 C 2.20 271
KirkHinrich 28 CHI 75 G 1.10 210
TyrusThomas 22 CHI 81 F 0.24 195
BradMiller 32 CHI 83 C 1.98 186
LindseyHunter 38 CHI 74 G -1.67 24
TimThomas 31 CHI 82 F -0.69 15
LintonJohnson 28 CHI 80 F -2.19 10
AaronGray 24 CHI 84 C -1.05 9
AnthonyRoberson 25 CHI 74 G -2.25 4
Chicago 0.75 0.523
Boston in 4 17.2%
Chicago in 4 1.3%
Boston in 5 28.9%
Chicago in 5 2.8%
Boston in 6 18.8%
Chicago in 6 7.1%
*Boston in 7* 17.7%
Chicago in 7 6.2%

3. Orlando vs. 6. Philadelphia

RashardLewis 29 ORL 82 F 3.38 238
HedoTurkoglu 29 ORL 82 F 2.44 223
RaferAlston 32 ORL 74 G 1.38 198
DwightHoward 23 ORL 83 C 5.39 194
CourtneyLee 23 ORL 77 G -1.46 167
MickaelPietrus 26 ORL 78 G -0.39 125
AnthonyJohnson 34 ORL 75 G -2.08 96
MarcinGortat 24 ORL 84 F -0.50 86
J.J.Redick 24 ORL 76 G -2.55 73
TonyBattie 32 ORL 83 C -3.47 40
Orlando 6.27 0.704
AndreIguodala 25 PHI 78 G 4.28 269
AndreMiller 32 PHI 74 G 1.97 258
ThaddeusYoung 20 PHI 80 F -1.16 229
LouisWilliams 22 PHI 74 G 0.67 149
WillieGreen 27 PHI 76 G -3.32 147
SamuelDalembert 27 PHI 83 C -2.26 133
TheoRatliff 35 PHI 82 C -0.27 94
DonyellMarshall 35 PHI 81 F 0.65 50
RoyalIvey 27 PHI 75 G -2.61 45
ReggieEvans 28 PHI 80 F -1.08 36
MarreeseSpeights 21 PHI 82 F -1.32 29
Philadelphia 1.80 0.558
Orlando in 4 16.9%
Philly in 4 1.3%
Orlando in 5 28.8%
Philly in 5 2.9%
*Orlando in 6* 18.8%
Philly in 6 7.3%
Orlando in 7 17.8%
Philly in 7 6.3%

4. Atlanta vs. 5. Miami

JoeJohnson 27 ATL 80 G 1.63 270
JoshSmith 23 ATL 81 F 2.35 266
MikeBibby 30 ATL 73 G 1.10 255
RonaldMurray 29 ATL 76 G -1.58 213
MauriceEvans 30 ATL 77 G -1.86 180
AlHorford 22 ATL 82 C 0.69 171
ZazaPachulia 24 ATL 83 C -1.17 166
SolomonJones 24 ATL 82 F -2.29 50
MarvinWilliams 22 ATL 81 F -0.62 41
MarioWest 24 ATL 77 G -0.55 34
AcieLaw 24 ATL 75 G -4.39 16
RandolphMorris 23 ATL 82 C -3.97 8
SpeedyClaxton 30 ATL 71 G -2.61 3
ThomasGardner 23 ATL 77 G -4.01 3
OthelloHunter 22 ATL 80 F -2.58 2
Atlanta 0.92 0.529
DwyaneWade 27 MIA 76 G 9.04 285
JamesJones 28 MIA 80 F -1.71 235
MarioChalmers 22 MIA 73 G 1.72 231
UdonisHaslem 28 MIA 80 F -2.93 204
MichaelBeasley 20 MIA 82 F -3.00 179
JermaineO'Neal 30 MIA 83 F -1.54 162
DaequanCook 21 MIA 77 G -3.32 161
JoelAnthony 26 MIA 81 C -2.31 89
JamaalMagloire 30 MIA 83 C -3.67 48
JamarioMoon 28 MIA 80 F 1.11 40
ChrisQuinn 25 MIA 74 G -1.37 24
YakhoubaDiawara 26 MIA 79 F -4.23 21
DorellWright 23 MIA 79 G -2.25 3
Miami 0.55 0.517
Atlanta in 4 6.4%
Miami in 4 5.2%
Atlanta in 5 16.0%
Miami in 5 9.2%
Atlanta in 6 13.9%
Miami in 6 17.3%
*Atlanta in 7* 19.6%
Miami in 7 12.4%

West
1. Los Angeles vs. 8. Utah

KobeBryant 30 LAL 78 G 6.44 203
PauGasol 28 LAL 84 F 2.99 194
LamarOdom 29 LAL 82 F 2.42 183
TrevorAriza 23 LAL 80 F 2.45 159
DerekFisher 34 LAL 73 G -0.34 146
ShannonBrown 23 LAL 76 G -3.50 87
AndrewBynum 21 LAL 84 C 0.69 77
SashaVujacic 24 LAL 79 G 1.01 77
LukeWalton 28 LAL 80 F -1.76 49
JoshPowell 26 LAL 81 F -5.68 18
JordanFarmar 22 LAL 74 G -2.10 8
Los Angeles 9.54 0.811
DeronWilliams 24 UTA 75 G 1.94 211
CarlosBoozer 27 UTA 81 F 2.04 186
RonnieBrewer 23 UTA 79 G 1.02 158
PaulMillsap 23 UTA 80 F 2.42 155
AndreiKirilenko 27 UTA 81 F 2.63 136
KyleKorver 27 UTA 79 F -0.41 136
C.J.Miles 21 UTA 78 G -1.68 58
MattHarpring 32 UTA 79 F -2.16 49
MehmetOkur 29 UTA 83 F 1.70 43
JarronCollins 30 UTA 83 F -3.20 35
BrevinKnight 33 UTA 70 G -1.02 17
RonniePrice 25 UTA 74 G -2.41 16
Utah 5.55 0.680
L.A. in 4 18.5%
Utah in 4 1.1%
*L.A. in 5* 29.9%
Utah in 5 2.5%
L.A. in 6 18.9%
Utah in 6 6.4%
L.A. in 7 17.0%
Utah in 7 5.6%

2. Denver vs. 7. New Orleans

CarmeloAnthony 24 DEN 80 F 2.17 183
ChaunceyBillups 32 DEN 75 G 4.29 178
KenyonMartin 31 DEN 81 F -0.25 163
NeneHilario 26 DEN 83 F 1.70 148
J.R.Smith 23 DEN 78 G 1.78 125
ChrisAndersen 30 DEN 82 F 1.36 112
DahntayJones 28 DEN 78 G -3.23 103
AnthonyCarter 33 DEN 73 G -1.18 75
LinasKleiza 24 DEN 80 F -2.16 69
JohanPetro 23 DEN 84 C -4.12 17
RenaldoBalkman 24 DEN 80 F 0.92 15
JasonHart 30 DEN 75 G -3.33 13
Denver 4.48 0.645
ChrisPaul 23 NOH 72 G 10.93 201
DavidWest 28 NOH 81 F 0.44 178
PejaStojakovic 31 NOH 81 F -0.28 162
RasualButler 29 NOH 79 G -2.15 158
JamesPosey 32 NOH 80 F 1.11 123
TysonChandler 26 NOH 85 C -0.70 94
SeanMarks 33 NOH 82 F -4.16 80
AntonioDaniels 33 NOH 76 G -1.96 64
HiltonArmstrong 24 NOH 83 F -3.53 53
DevinBrown 30 NOH 77 G -2.21 33
JulianWright 21 NOH 80 F -2.51 32
MorrisPeterson 31 NOH 79 F -1.52 21
RyanBowen 33 NOH 79 F -0.41 2
New Orleans 4.70 0.652
Denver in 4 5.4%
New Orleans in 4 6.1%
*Denver in 5* 14.4%
New Orleans in 5 10.5%
Denver in 6 12.9%
New Orleans in 6 18.8%
Denver in 7 19.0%
New Orleans in 7 13.0%

3. San Antonio vs. 6. Dallas

TonyParker 26 SAS 74 G 1.53 181
TimDuncan 32 SAS 83 F 5.93 164
MichaelFinley 35 SAS 79 G -2.03 143
BruceBowen 37 SAS 79 F -2.17 130
RogerMason 28 SAS 77 G -1.44 108
ImeUdoka 31 SAS 78 F -0.77 104
MattBonner 28 SAS 82 F 1.39 100
KurtThomas 36 SAS 81 F 0.29 80
GeorgeHill 22 SAS 74 G -0.83 76
DrewGooden 27 SAS 82 F -2.27 71
FabricioOberto 33 SAS 82 F -0.33 22
JacqueVaughn 33 SAS 73 G -4.37 21
San Antonio 1.16 0.537
DirkNowitzki 30 DAL 84 F 4.42 186
JasonKidd 35 DAL 76 G 5.36 186
JoshHoward 28 DAL 79 F 0.91 159
JasonTerry 31 DAL 74 G 1.92 156
ErickDampier 33 DAL 83 C 0.83 149
JoseBarea 24 DAL 72 G -2.76 132
BrandonBass 23 DAL 80 F -3.24 85
AntoineWright 24 DAL 79 G -3.08 67
RyanHollins 24 DAL 84 C -2.01 40
JamesSingleton 27 DAL 80 F -1.15 18
GeraldGreen 23 DAL 80 F -4.13 12
MattCarroll 28 DAL 78 G -3.21 11
Dallas 5.63 0.683
San Antonio in 4 1.4%
Dallas in 4 16.6%
San Antonio in 5 5.0%
*Dallas in 5* 21.4%
San Antonio in 6 5.4%
Dallas in 6 25.8%
San Antonio in 7 10.9%
Dallas in 7 13.5%

4. Portland vs. 5. Houston

BrandonRoy 24 POR 78 G 4.82 238
LaMarcusAldridge 23 POR 83 F 0.78 237
SteveBlake 28 POR 75 G 0.36 231
TravisOutlaw 24 POR 81 F -0.78 170
JoelPrzybilla 29 POR 85 C -0.34 162
RudyFernandez 23 POR 78 G 2.05 162
GregOden 21 POR 84 C 1.06 96
NicolasBatum 20 POR 80 F 0.55 63
ChanningFrye 25 POR 83 F -3.08 36
SergioRodriguez 22 POR 75 G -1.24 27
JerrydBayless 20 POR 75 G -4.01 11
MichaelRuffin 32 POR 81 F -0.50 5
Portland 5.23 0.670
ShaneBattier 30 HOU 80 F 1.78 236
RonArtest 29 HOU 78 F 3.35 221
YaoMing 28 HOU 90 C 2.50 216
LuisScola 28 HOU 81 F 0.62 200
AaronBrooks 24 HOU 72 G -1.55 188
KyleLowry 22 HOU 72 G 0.94 115
VonWafer 23 HOU 77 G -2.16 107
CarlLandry 25 HOU 81 F -0.41 93
ChuckHayes 25 HOU 78 F 0.38 33
DikembeMutombo 42 HOU 86 C -1.92 20
BrianCook 28 HOU 81 F -3.14 5
BrentBarry 37 HOU 78 G 0.02 4
JamesWhite 26 HOU 79 G -2.32 2
Houston 4.60 0.649
Portland in 4 6.9%
Houston in 4 4.8%
Portland in 5 16.9%
Houston in 5 8.6%
Portland in 6 14.4%
*Houston in 6* 16.5%
Portland in 7 19.8%
Houston in 7 12.0%

Hey, that's not very bad, to be honest. I should have used this method for the Stat Geek Smackdown! For the curious, here's what it predicts for Round 2, assuming the same distribution of minutes for each team as in Round 1:
East
1. Cleveland vs. 4. Atlanta

Cleveland in 4 45.0%
Atlanta in 4 0.1%
Cleveland in 5 35.4%
Atlanta in 5 0.2%
Cleveland in 6 12.4%
Atlanta in 6 0.7%
Cleveland in 7 5.4%
Atlanta in 7 0.7%

2. Boston vs. 3. Orlando

Boston in 4 4.4%
Orlando in 4 7.4%
Boston in 5 12.3%
Orlando in 5 12.2%
Boston in 6 11.5%
Orlando in 6 20.5%
Boston in 7 18.0%
Orlando in 7 13.7%

West
1. Los Angeles vs. 5. Houston

L.A. in 4 22.2%
Houston in 4 0.8%
L.A. in 5 32.2%
Houston in 5 1.7%
L.A. in 6 18.7%
Houston in 6 4.8%
L.A. in 7 15.2%
Houston in 7 4.4%

2. Denver vs. 6. Dallas
(Denver already won Game 1)

Denver in 4 7.2%
Denver in 5 17.6%
Dallas in 5 8.0%
Denver in 6 13.7%
Dallas in 6 18.7%
Denver in 7 19.5%
Dallas in 7 15.4%

Related posts:

    Predicting Round 3 With SPM
    Predicting the Finals With SPM
    Predicting with Statistical Plus/Minus
    Which Single-Game Playoff Outcomes Were the Most Surprising?
    2010-11 Surprise Teams (11-22-2010 edition)

This entry was posted on Monday, May 4th, 2009 at 4:20 am and is filed under Analysis, Playoffs, Projections, Statistical +/-. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

8 Responses to “Can SPM Predict the Playoffs?”

  1. Jason J Says:
    May 4th, 2009 at 9:18 am

    Those are interesting results. I actually think the method got a little undermined by players trying to fight through injuries and simple matchups.

    Metrics don't account for Houston being a matchup nightmare for Portland. Metrics don't account for Kenyon Martin single covering David West or for Tyson Chandler's production being impinged by injury. Metrics can't account for Gordon finding 7th gear against Boston or to understand what exactly it means to the Celts to only have 2 legitimate bigs left in their line-up.

    Considering all the factors that SPM can't anticipate, those results are very solid.

  2. Guy F Says:
    May 4th, 2009 at 2:17 pm

    How do you convert +/- to win%?

  3. Neil Paine Says:
    May 4th, 2009 at 4:35 pm

    For 2009, WPct ~ .49869 + (.0327 * Efficiency Differential).

  4. Gerrit Says:
    May 4th, 2009 at 7:31 pm

    Interesting that the team without HCA is more likely to win in six than seven. I'm curious if that matches real world results.

  5. Jason J Says:
    May 5th, 2009 at 10:57 am

    Gerrit - That matches the conventional wisdom regarding playoff upsets, but I don't know if that's actually how it works out in the real world.

    Top of my head that does tend to be true though. Lakers beat the the #1 seed Blazers in 6 in 1991. Bulls beat the East leading Knicks in 6 in 1993. Cavs beat the #1 seed Pistons in 6 to make the finals a couple of years ago.

  6. Neil Paine Says:
    May 5th, 2009 at 1:11 pm

    Yeah, I don't have the data on-hand, but the underlying formula can be found here:

    http://www.whowins.com/formulae/probformulae.html

    Intuitively it makes sense -- given that HCA is so important, if an underdog is going to win a series, they're probably going to steal a road game and make the most of their games at home. When it comes down to a Game 6, up 3-2 at home, that's obviously your best chance to take the series; lose there, and you're back on the road for a do-or-die Game 7 with the odds heavily stacked against you.

  7. biggles Says:
    May 5th, 2009 at 1:15 pm

    "Hey, that’s not very bad, to be honest."

    You almost did too well!

    My test of the hypothesis that your distributions were the "true" ones produced a p-value of 0.945 (where 0 = your distributions are crap and 1 = you were incredibly lucky or you cheated). Which, if my code is right, suggests your predictive success may be unsustainable... (Or maybe I just chose a test statistic flukily friendly to you.)

    Unrealistic wish: that the Stat Geek Smackdown required you to submit probability distributions -- then it would be a much truer test of predictive skill.

  8. Eddy Says:
    May 5th, 2009 at 9:33 pm

    Doing some simple arithmetic shows the Magic have a 53.8% of beating the Celtics in the Semifinals, which makes sense, given that conventional wisdom and evaluation of different metrics supported the notion the series would be a toss-up. Good stuff, Neil.

«
»