2024-25 team win projection contest

[Crow]

5th and 1st place have gotten closer. 40-50% closer in 8 days.

[Mike G]

Adding the 2 that Crow suggested may represent 'Vegas', though only one seems very legit.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.53   7.3   .40      perW   7.15   8.9   .18
bpmW   5.58   7.2   .42      emin   7.22   9.3   .36
24pr   5.90   7.8   .28      4141   7.23   9.0   
avgA   6.20   7.8   .40      Crow   7.25   9.2   .38
TmTj   6.61   8.4   .42     vegas   7.34   9.2   .35
eWin   6.88   8.5   .23      medi   7.50   9.5   .36
DQin   6.95   8.8   .40      24py   8.52  10.7   .28
dtka   6.96   8.2   .40      bmgm   9.66  11.4   .28

A doormat entry isn't a bad thing.

[TeemoTeejay]

I’d have sworn that Vegas usually wins these things right?

[Crow]

No. It has I think, but not "usually".

[DQuinn1575]

Adding the 2 that Crow suggested may represent 'Vegas', though only one seems very legit.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.53   7.3   .40      perW   7.15   8.9   .18
bpmW   5.58   7.2   .42      emin   7.22   9.3   .36
24pr   5.90   7.8   .28      4141   7.23   9.0   
avgA   6.20   7.8   .40      Crow   7.25   9.2   .38
TmTj   6.61   8.4   .42     vegas   7.34   9.2   .35
eWin   6.88   8.5   .23      medi   7.50   9.5   .36
DQin   6.95   8.8   .40      24py   8.52  10.7   .28
dtka   6.96   8.2   .40      bmgm   9.66  11.4   .28

A doormat entry isn't a bad thing.

Hi, can you do me a favor and post like who each entry is - like I see WShr, and think it's a WinShare entry, but dont know whose, and am not sure who avgA or 4141 are
Thanks
DQin (DQuinn1575)

[TeemoTeejay]

Adding the 2 that Crow suggested may represent 'Vegas', though only one seems very legit.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.53   7.3   .40      perW   7.15   8.9   .18
bpmW   5.58   7.2   .42      emin   7.22   9.3   .36
24pr   5.90   7.8   .28      4141   7.23   9.0   
avgA   6.20   7.8   .40      Crow   7.25   9.2   .38
TmTj   6.61   8.4   .42     vegas   7.34   9.2   .35
eWin   6.88   8.5   .23      medi   7.50   9.5   .36
DQin   6.95   8.8   .40      24py   8.52  10.7   .28
dtka   6.96   8.2   .40      bmgm   9.66  11.4   .28

A doormat entry isn't a bad thing.

Hi, can you do me a favor and post like who each entry is - like I see WShr, and think it's a WinShare entry, but dont know whose, and am not sure who avgA or 4141 are
Thanks
DQin (DQuinn1575)

The top 4 are all other stuff

4141 = all teams 41-41

[Mike G]

I took WS/48, PER, BPM, and my own eWins player rates from last year and gave them all the minutes from Kevin Pelton.
Even though these stats are unrelated, they seem to congregate near one end of the Wins spectrum -- they are the 4 lowest on Bos, the 4 highest on Brk, etc. This suggests that Minutes are a big part of the predictive value.

TmTj = TeemoTeejay
DQin = DQuinn1575
dtka = dtkavana
emin = eminence
Crow = Crow
medi = mediocre (last year's winner)
vegas = https://www.covers.com/nba/win-totals-odds
bmgm = https://sports.betmgm.com/en/blog/nba/n ... tals-bm23/

those last 2 :

Code: Select all

west  bmgm vegas      east   bmgm vegas
Dal    50    49        Atl    36    36
Den    57    50        Bos    64    58
GSW    46    43        Brk    32    19
Hou    41    43        Cha    21    30
LAC    51    35        Chi    39    28
LAL    47    43        Cle    48    48
Mem    27    47        Det    14    25
Min    56    52        Ind    47    46
NOP    49    46        Mia    46    44
OKC    57    56        Mil    49    50
Phx    49    47        NYK    50    54
Por    21    22        Orl    47    47
Sac    46    46        Phl    47    49
SAS    22    36        Tor    25    29
Uta    31    28        Was    15    20

I don't know who would bet the Spurs win <22 or the Clips >51

Last night was bad for everyone but especially at the top.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.95   7.5   .37      DQin   7.31   9.0   .37
bpmW   5.97   7.4   .39      Crow   7.31   9.3   .37
24pr   5.98   7.9   .26      perW   7.38   9.1   .15
avgA   6.59   8.0   .37      vegas  7.53   9.4   .32
TmTj   6.90   8.6   .40      medi   7.63   9.6   .34
eWin   6.96   8.7   .20      emin   7.68   9.6   .32
dtka   7.23   8.4   .38      24py   8.64  10.8   .26
4141   7.30   9.0            bmgm   9.75  11.6   .26

[TeemoTeejay]

I took WS/48, PER, BPM, and my own eWins player rates from last year and gave them all the minutes from Kevin Pelton.
Even though these stats are unrelated, they seem to congregate near one end of the Wins spectrum -- they are the 4 lowest on Bos, the 4 highest on Brk, etc. This suggests that Minutes are a big part of the predictive value.

TmTj = TeemoTeejay
DQin = DQuinn1575
dtka = dtkavana
emin = eminence
Crow = Crow
medi = mediocre (last year's winner)
vegas = https://www.covers.com/nba/win-totals-odds
bmgm = https://sports.betmgm.com/en/blog/nba/n ... tals-bm23/

those last 2 :

Code: Select all

west  bmgm vegas      east   bmgm vegas
Dal    50    49        Atl    36    36
Den    57    50        Bos    64    58
GSW    46    43        Brk    32    19
Hou    41    43        Cha    21    30
LAC    51    35        Chi    39    28
LAL    47    43        Cle    48    48
Mem    27    47        Det    14    25
Min    56    52        Ind    47    46
NOP    49    46        Mia    46    44
OKC    57    56        Mil    49    50
Phx    49    47        NYK    50    54
Por    21    22        Orl    47    47
Sac    46    46        Phl    47    49
SAS    22    36        Tor    25    29
Uta    31    28        Was    15    20

I don't know who would bet the Spurs win <22 or the Clips >51

Last night was bad for everyone but especially at the top.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.95   7.5   .37      DQin   7.31   9.0   .37
bpmW   5.97   7.4   .39      Crow   7.31   9.3   .37
24pr   5.98   7.9   .26      perW   7.38   9.1   .15
avgA   6.59   8.0   .37      vegas  7.53   9.4   .32
TmTj   6.90   8.6   .40      medi   7.63   9.6   .34
eWin   6.96   8.7   .20      emin   7.68   9.6   .32
dtka   7.23   8.4   .38      24py   8.64  10.8   .26
4141   7.30   9.0            bmgm   9.75  11.6   .26

Do Kevin pelton wins do better than numberfire or rotowire?

[DQuinn1575]

I took WS/48, PER, BPM, and my own eWins player rates from last year and gave them all the minutes from Kevin Pelton.
Even though these stats are unrelated, they seem to congregate near one end of the Wins spectrum -- they are the 4 lowest on Bos, the 4 highest on Brk, etc. This suggests that Minutes are a big part of the predictive value.

TmTj = TeemoTeejay
DQin = DQuinn1575
dtka = dtkavana
emin = eminence
Crow = Crow
medi = mediocre (last year's winner)
vegas = https://www.covers.com/nba/win-totals-odds
bmgm = https://sports.betmgm.com/en/blog/nba/n ... tals-bm23/

those last 2 :

Code: Select all

west  bmgm vegas      east   bmgm vegas
Dal    50    49        Atl    36    36
Den    57    50        Bos    64    58
GSW    46    43        Brk    32    19
Hou    41    43        Cha    21    30
LAC    51    35        Chi    39    28
LAL    47    43        Cle    48    48
Mem    27    47        Det    14    25
Min    56    52        Ind    47    46
NOP    49    46        Mia    46    44
OKC    57    56        Mil    49    50
Phx    49    47        NYK    50    54
Por    21    22        Orl    47    47
Sac    46    46        Phl    47    49
SAS    22    36        Tor    25    29
Uta    31    28        Was    15    20

I don't know who would bet the Spurs win <22 or the Clips >51

Last night was bad for everyone but especially at the top.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.95   7.5   .37      DQin   7.31   9.0   .37
bpmW   5.97   7.4   .39      Crow   7.31   9.3   .37
24pr   5.98   7.9   .26      perW   7.38   9.1   .15
avgA   6.59   8.0   .37      vegas  7.53   9.4   .32
TmTj   6.90   8.6   .40      medi   7.63   9.6   .34
eWin   6.96   8.7   .20      emin   7.68   9.6   .32
dtka   7.23   8.4   .38      24py   8.64  10.8   .26
4141   7.30   9.0            bmgm   9.75  11.6   .26

Thanks for posting this.
If you want it, here is the link for the win totals that basketball-reference.com uses

https://www.sportsoddshistory.com/nba-r ... s-by-team/

[Mike G]

I did not know b-r.com showed Vegas predictions; but they are the same as I am using as 'vegas', once I removed .38 wins from each team, so they total 1230.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
WShr   5.85   7.4   .38      4141   7.18   9.0   
bpmW   5.85   7.2   .41      Crow   7.29   9.1   .40
24pr   5.92   7.8   .28      vegas  7.40   9.3   .34
avgA   6.55   7.9   .39      perW   7.47   9.0   .17
TmTj   6.76   8.4   .42      medi   7.57   9.5   .35
eWin   6.95   8.7   .21      emin   7.60   9.5   .33
DQin   7.03   8.8   .40      24py   8.54  10.7   .28
dtka   7.12   8.3   .40      bmgm   9.64  11.4   .28

Leaders at various exponents (excluding avgA)

Code: Select all

e = 0.5          e = 1 = MAE     e = 2 = RMSE
24pr   4.74      WShr   5.846     bpmW   7.19
WShr   4.94      bpmW   5.852     WShr   7.37
bpmW   5.09      24pr   5.92      24pr   7.81
TmTj   5.79      TmTj   6.76      dtka   8.26
DQin   5.90      eWin   6.95      TmTj   8.37
                     
e = 4             e = 8             e = 16   
bpmW   9.51      WShr   12.29      WShr   14.3
WShr   9.71      bpmW   12.30      bpmW   14.5
dtka  10.07      dtka   12.37      dtka   14.6
24pr  10.39      24pr   13.20      DQin   15.3
DQin  11.07      DQin   13.35      4141   15.3

[Crow]

BRef projection is for 12 team fights for conference top 10s. I think it will shake out to be a bit less in at least the east.

[Mike G]

Over and under-achieving teams relative to the avg of our predictions.
I've inserted current SRS, which along with current W-L (and some regression to .500) leads to the b-r.com projection.

Code: Select all

over  tm   avgA  b-r  srs    over   tm   avgA  b-r  srs
12.7  Brk   27   40  -0.8   -20.3   Phl   49   28  -8.0
12.0  Cle   48   60  10.0   -17.1   NOP   45   28  -7.4
11.2  GSW   46   57  11.3   -10.4   Min   51   41   0.3
10.3  Hou   42   52   8.1    -8.5   Mil   46   37  -1.7
8.4   Det   30   38  -2.3    -7.4   Ind   44   37  -3.2
5.8   SAS   34   39  -0.3    -5.6   Phx   47   42  -0.9
5.5   Cha   30   35  -4.5    -5.4   Sac   47   41   0.2
5.4   Por   26   31  -7.0    -5.3   Dal   50   44   3.7
5.1   Chi   31   36  -4.1    -4.8   Den   49   44   2.0
4.8   LAC   37   42   1.5    -4.1   Atl   39   35  -5.1
4.6   Mem   43   47   5.6    -3.2   Bos   58   54   7.9
4.3   Orl   43   47   3.4    -2.9   Uta   30   27  -9.6
4.2   Was   23   27 -10.4    -2.6   Mia   43   40  -1.1
4.1   OKC   56   60  13.3    -1.7   Tor   30   28  -6.8
2.5   LAL   41   43   0.0    -1.2   NYK   47   46   3.2

On avg, errors are worse by about 0.55 in the last week.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
bpmW   6.11   7.4   .40      DQin   7.21   8.9   .39
WShr   6.23   7.6   .37      Crow   7.35   9.0   .41
24pr   6.26   8.0   .27      vegas  7.44   9.4   .33
avgA   6.71   8.0   .38      medi   7.64   9.6   .36
TmTj   6.79   8.4   .42      perW   7.71   9.3   .15
4141   7.17   9.2            emin   7.76   9.6   .33
eWin   7.18   8.8   .20      24py   8.78  10.8   .27
dtka   7.19   8.4   .39      bmgm   9.79  11.5   .27

Those towards the top again have been hit harder.
The spread from #1 to #14 is season-low 1.65
Last year at this time it was 3.03, from 4.30 to 7.33

[Crow]

10 teams with 7+ variance, only 6 within 4.


Lower threshold, I'm .19 from 6th.

[Mike G]

Big shakeup overnight, avg .15 worse but with wide variation, again hit harder toward the top.
First column is MAE change from yesterday.

Code: Select all

chg   .  avg err   rmse   r^2
.31   bpmW   6.41   7.6   .40
.31   WShr   6.54   7.9   .36
.41   24pr   6.67   8.3   .25
.02   avgA   6.74   8.1   .38
.04   TmTj   6.83   8.4   .43
.01   dtka   7.20   8.5   .38
-.14  Crow   7.21   8.9   .42
.17   DQin   7.38   9.0   .39
.27   eWin   7.46   9.0   .21
.02  vegas   7.46   9.3   .35
.30   4141   7.46   9.5   
.11   medi   7.76   9.6   .36
.05   emin   7.81   9.7   .33
.26   perW   7.98   9.5   .15
.16   24py   8.94  11.1   .25
.14   bmgm   9.93  11.8   .25

And how much do each of us agree with "Vegas"? or with one another?
What's the correlation between our current "error" averages and our "agree" levels?

Code: Select all

"vegas" diff      avgA diff   
Crow   2.26      bpmW   1.94
dtka   2.37      dtka   1.98
TmTj   2.91      TmTj   2.09
emin   2.93      WShr   2.29
medi   3.14      DQin   2.65
DQin   3.49      medi   2.91
bpmW   3.81      Crow   3.13
WShr   4.15      eWin   3.41
eWin   4.74      perW   3.59
perW   5.05      emin   3.65
correl .115     correl  .82

So far, there's not much to be said for going with vegas with slight variations.
The high success rate of being close to our avg guesses is corroborated by the (always) high position of that avg.

[Crow]

Moved up to 7th. First is 0.81 away.

Interesting that I am closest to Vegas and pretty far from other entries. Long way to go but rate of change may slow down.

Leading 3 sets fell twice as far or more from previous up date as average.