2015-16 Team win projections

[Crow]

My thinking is scale them. And / or give them a bit less regard.

[Crow]

Scaling has been done before, so I leaned that way. But it is sort of a favor that they didn't ask for and don't "deserve", so yeah it is not clear what to do. Do what you want.

[Mike G]

Again relative to the b-r.com forecast, avg errors:

Code: Select all

KF   6.06      fpli  6.59      BD    7.39
AJ   6.11      Crow  6.73      nr    7.54
km   6.21      MG    6.90      EZ    7.91
DF   6.26      rsm   6.91      Dan   7.98
bbs  6.28      yoop  6.97      DrP   8.16
Cal  6.31      snd   6.99      taco  8.20
tzu  6.40      itca  7.16      15py  9.51
DSM  6.45                  

Season is 31% over.

[nbacouchside]

Again relative to the b-r.com forecast, avg errors:

Code: Select all

KF   6.06      fpli  6.59      BD    7.39
AJ   6.11      Crow  6.73      nr    7.54
km   6.21      MG    6.90      EZ    7.91
DF   6.26      rsm   6.91      Dan   7.98
bbs  6.28      yoop  6.97      DrP   8.16
Cal  6.31      snd   6.99      taco  8.20
tzu  6.40      itca  7.16      15py  9.51
DSM  6.45                  

Season is 31% over.

Clearly, the b-r.com forecast and average errors should be the measuring stick. Clearly.

[tacoman206]

Ugh, Indy is killing me this year by being so good. Last time I use an all-human forecasting system.

[Dr Positivity]

Too many triple bogeys on my card also, just cheering to beat Vegas now. Will look at using prior seasons more next year

[RoyceWebb]

Two more from October.

http://insider.espn.go.com/nba/insider/ ... -races-nba

Pelton's preseason projections based on RPM:

TEAM WINS SIM
Golden State 59.8 58.5
San Antonio 58.3 57.2
Cleveland 55.1 55.4
Houston 54.6 53.0
Okla City 54.0 53.2
L.A. Clippers 53.3 51.2
Boston 47.8 48.5
Memphis 46.7 46.0
Chicago 44.8 45.4
Toronto 44.2 44.2
Washington 43.6 43.6
New Orleans 43.5 42.9
Indiana 42.9 43.4
Atlanta 42.1 42.6
Utah 41.6 40.5
Portland 40.7 39.7
Dallas 39.7 38.7
Miami 38.6 40.0
Detroit 38.2 39.0
Phoenix 38.1 37.6
Milwaukee 37.4 38.0
Sacramento 35.6 35.4
Charlotte 35.2 36.5
Orlando 33.7 34.7
Minnesota 29.8 29.8
New York 28.7 30.1
L.A. Lakers 28.7 28.8
Denver 27.6 27.7
Brooklyn 22.9 24.7
Philadelphia 22.7 23.8

Note: Projections are based on estimated playing time and the multi-year version of ESPN's RPM, adjusted for age, along with my projections for rookies and other players without RPM projections. Simulation results reflect 1,000 simulations of the regular season, factoring in schedule and randomness.

[bchaikin]

Simulation results reflect 1,000 simulations of the regular season, factoring in schedule and randomness.

i'd be interested in hearing how your simulation works...

[Mike G]

At about 32% of the season:

Code: Select all

KF   6.03      Crow  6.62      BD   7.36
AJ   6.17      fpli  6.69      nr   7.55
tzu  6.25      MG    6.93      EZ   7.89
DF   6.26      rsm   6.93      DrP  7.96
Cal  6.26      snd   6.93      Dan  8.03
bbs  6.27      itca  7.12      taco 8.27
km   6.27      yoop  7.14      15py 9.50
DSM  6.33                  

Ugh, Indy is killing me this year by being so good. Last time I use an all-human forecasting system.

Since everyone expected Ind to not be this good, it doesn't matter how much better they do, beyond 46 wins. That's the top guess (fpliii and Statman Dan); so even if you have the lowest guess, the separation between projections is constant.

What bugs the heck out of me is that as a fan of the Pacers -- and the Bulls -- I find myself wishing they'd lose games. If I were a 'true fan', I'd predict they'll do better than my numbers suggest.

[kmedved]

I've added Pelton's projections to the list - they perform extremely well:

[caliban]

Sunday Update week 2; and holy simulation from Pelton. Used half round down for the transfer to the template to close in on 1230, idk.

Basketball Dork with a solid week, Vegas Opening line with the dud & AJbaskets back on top

[Statman]

What bugs the heck out of me is that as a fan of the Pacers -- and the Bulls -- I find myself wishing they'd lose games. If I were a 'true fan', I'd predict they'll do better than my numbers suggest.

You can't care at all about your projections. I'm a Mavs fan - I think I projected them worse than about anybody (just how my model worked out). I'd much rather see them greatly exceed my projections than be right in this case.

[Mike G]

After a couple days of teams mostly doing what they're supposed to, everyone looks better, and there's a new leader, in avg error vs b-r.com's projections.

Code: Select all

tzu  5.82      fpli  6.52      BD    7.03
KF   5.86      Crow  6.53      nr    7.54
AJ   5.86      MG    6.60      EZ    7.70
Cal  5.91      rsm   6.68      Dan   7.72
DF   5.96      yoop  6.82      DrP   7.96
km   6.05      itca  6.84      taco  7.96
DSM  6.20      snd   6.88      15py  9.58
bbs  6.24                  

tarrazu shaved .43 off his error from 2 days ago.

Meanwhile, how strange would these projections have seemed, before the season?

Code: Select all

west   W      east   W
Hou   72      Cle   60
LAC   64      Atl   52
Mem   58      Was   50
NOP   58      Chi   49
OKC   56      Mil   48
GSW   55      Tor   45
SAS   49      Bos   42
Uta   40      Mia   40
Dal   37      Ind   32
Phx   37      Phl   30
Por   37      Det   28
Sac   36      Cha   24
LAL   31      Brk   23
Den   25      Orl   18
Min   18      NYK   15

[Statman]

Meanwhile, how strange would these projections have seemed, before the season?

Code: Select all

west   W      east   W
Hou   72      Cle   60
LAC   64      Atl   52
Mem   58      Was   50
NOP   58      Chi   49
OKC   56      Mil   48
GSW   55      Tor   45
SAS   49      Bos   42
Uta   40      Mia   40
Dal   37      Ind   32
Phx   37      Phl   30
Por   37      Det   28
Sac   36      Cha   24
LAL   31      Brk   23
Den   25      Orl   18
Min   18      NYK   15

Where did those projections come from? No one had Houston that good, I believe I was the biggest over projection at 61.

[Mike G]

Those are the Bizarro projections: Houston is currently headed for 36.4 wins, which would be 18 under our avg guess of 54.4
Adding 18 to 54.4, we get 72.4
The theory is that being off by 18 wins, on avg, shouldn't be more surprising in one direction than in the other.

Of course, after 1/3 of the season has been played, it's hard to imagine that the Rockets really should be as good as 54 or 55 wins.
But before the season, a guess of 72 shouldn't seem any stranger than saying they drop to 36 -- should it?

And yeah, they are 25 under your guess; if they were 25 over, they'd win 86. Averaging a bunch of regressed predictions, we don't get that kind of unreal number.


In Bizarro world, the east-west disparity has nearly doubled from last season. In real life, it's almost vanished.