2023-24 team win projection contest

Home for all your discussion of basketball statistical analysis.
Post Reply
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

So the Thunder are "expected" to improve based on average results for average year to year aging?

But they are currently falling short of that... maybe because they got 0.1 years younger instead of average aging?
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

Age and wins with only a modest correlation, you indicate.

Wonder about size of correlations for age & salary, age & playoff advancement, salary & playoff advancement, age & wins / salary...
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2023-24 team win projection contest

Post by Mike G »

Crow wrote: Sun Nov 05, 2023 10:08 pm .. maybe because they got 0.1 years younger instead of average aging?
Because everyone but Dort (and the rookies) has fallen off precipitously from last year.

Code: Select all

OKC Thunder    age     PER          ws/48         BPM        mpg  
Jal. Williams  21   15.6  14.2   .119  .072    0.3   -3.0   28  33
Luguentz Dort  23   11.3  17.0   .062  .159   -3.0    1.2   28  30
Chet Holmgren  21         24.3         .239           6.3       29
Josh Giddey    20   17.1  10.6   .081 -.064    1.2   -6.1   29  29
Shai G-A       24   27.2  23.9   .226  .131    7.3    3.2   29  29

Cason Wallace  20         14.0         .153          -0.6       23
Isaiah Joe     23   14.5  11.9   .131  .047    0.9   -2.6   17  20
Ousmane Dieng  19    9.3   9.3   .038  .024   -3.6   -4.0    7  13
Aaron Wiggins  24   12.2   9.9   .097  .042   -2.1   -5.3   16  11
Olivier Sarr   23   11.5  11.7   .094  .134   -4.9   -2.2    1   7

OKC Thunder   age      PER          ws/48         BPM         mpg
A Pokusevski   21   13.4 -13.5   .066 -.428    0.0  -26.9    9   3
Jayl Williams  20   11.2  13.6   .099  .068   -1.7   -1.7   11   3
Dāvis Bertāns  30   11.9  15.4   .090  .227   -1.2   -0.4    6   2
Tre Mann       21   10.0  38.7   .021  .355   -3.9   17.5   14   2
Lindy Waters   25   11.8  -9.2   .101 -.296    0.1  -18.9    6   1
Kenr Williams  28   14.4         .133          0.5          15   0
Whatever one thinks of these stats, a dropoff is still a dropoff; and the stats tend to concur.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

The actual stats explain the 2nd column, not the fourth.
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2023-24 team win projection contest

Post by Mike G »

Code: Select all

.   avg err   rmse   r^2      .   avg err   rmse   r^2
avgA   4.94   6.25   .35      DRKO   4.10   5.13   .48
ncs.   5.07   6.30   .32      LEBR   4.46   5.70   .34
eWin   5.14   6.45   .13      KPel   4.76   6.39   .24
dtka   5.17   6.62   .29      EBPI   4.92   6.32   .32
trzu   5.18   6.45   .31      4141   5.38   6.48   
Crow   5.43   6.73   .29     vegas   6.01   7.03   .30
vzro   5.95   7.57   .28      EExp   6.43   7.52   .29
medi   6.08   7.50   .37      ChKl   6.50   8.12   .22
DQin   6.22   7.39   .28      NuFi   6.87   8.78   .21
emin   6.43   7.72   .37      IanL   7.20   8.34   .27
https://www.basketball-reference.com/fr ... _prob.html
Those on the right avg 5.69 / 7.04; on the left 5.63 / 6.97
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

At this point, almost everyone in "contest" has an estimated error less than ESPN "expert" group, except for one tie.

APBRmetrics average doing well, better than any single part.

Darko and LeBron doing very well.

( I cite Darko pretty frequently. Might "should" use it exclusively for quick takes.)
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2023-24 team win projection contest

Post by Mike G »

Extreme overnight recalibration . East v West went 4-4, and projections now avg East 41.7 - 40.3 West.

Code: Select all

.   avg err   rmse   r^2      .   avg err   rmse   r^2
dtka   4.56   5.68   .46      DRKO   4.09   4.81   .56
trzu   4.85   5.72   .45      LEBR   4.52   5.46   .42
avgA   4.88   5.74   .45      KPel   4.57   5.73   .39
ncs.   4.99   5.79   .43      EBPI   5.26   5.90   .41
Crow   5.00   6.23   .39     vegas   5.57   6.55   .39
medi   5.39   6.68   .49      4141   5.58   6.91   
eWin   5.39   6.65   .15      ChKl   5.74   7.40   .34
vzro   5.68   6.72   .42      EExp   5.96   7.12   .37
emin   6.16   7.47   .41      IanL   6.29   7.97   .33
DQin   6.36   7.59   .27      NuFi   6.34   8.42   .27
The avg's of each side are now equal.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

70% in contest ahead of Vegas.

2nd place is pretty contested.
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2023-24 team win projection contest

Post by Mike G »

No games last night.
DARKO currently leads our Average entry by .79 in mean avg error. Here's a team by team breakdown:

Code: Select all

diff   tm   proj   DRKO   avgA      diff    tm   proj   avgA   DRKO
5.9   Mil   42.8   44.4   50.3      -3.8   OKC   42.8   40.3   36.5
4.4   NOP   38.7   40.8   45.2      -3.2   Tor * 41.8   40.4   46.4
4.0   Hou   38.7   35.9   31.9      -2.6   GSW   45.5   45.8   48.4
3.1   Sac   33.9   39.4   42.5      -1.8   Den   53.6   50.7   48.9
2.9   Cle   40.3   46.4   49.3      -1.3   Cha   35.6   32.3   31.0
2.7   LAL   38.5   40.5   43.2      -1.1   Ind   42.3   35.0   33.9
2.7   Min   53.1   48.7   46.0      -0.9   Chi   35.1   37.7   38.6
2.5   Mem   30.1   42.2   44.7      -0.8   Bos   56.1   56.5   57.3
2.1   Was   32.6   31.9   29.8      -0.5   Dal   47.7   41.0   40.5
1.9   Det   32.7   28.0   26.1      -0.2   LAC   45.0   42.6   42.4
1.9   Phl   51.9   49.8   47.9      -0.0   NYK   45.4   47.2   47.2
1.3   Brk * 41.3   41.4   39.9                  
1.2   Por   34.9   31.3   30.1                  
1.1   Atl   47.0   43.1   42.0                  
1.0   Uta   31.4   36.4   37.4                  
0.7   Mia   38.9   41.7   42.4                  
0.4   Orl   42.4   36.7   36.3                  
0.1   SAS * 31.0   33.2   28.7                  
0.0   Phx   39.1   46.9   46.9                                
Projection by b-r.com is closer to DARKO's guess for teams on the left; closer to 'our' collective guesses on the right.
DARKO had the lowest guess on Mem, Mia, and (esp.) Mil, all of which look pretty good right now. Also highest on Tor.
For all but 3 * teams * DARKO and AvgA are on the same side -- both high or both low -- of the current projection. For those 3, any change will affect the lead.

UPDATE Nov. 9 -- 28 of 30 teams in action.

Code: Select all

.   avg err   rmse   r^2      .   avg err   rmse   r^2
dtka   4.99   6.34   .40      DRKO   4.41   5.29   .54
ncs.   5.26   6.44   .38      LEBR   5.06   6.03   .40
trzu   5.27   6.45   .38      KPel   5.24   6.39   .35
avgA   5.36   6.47   .38      EBPI   5.83   6.67   .34
Crow   5.44   6.75   .36      ChKl   6.07   7.93   .29
medi   5.89   6.98   .46     vegas   6.13   7.37   .31
vzro   6.13   7.11   .39      4141   6.24   7.60   
eWin   6.35   7.59   .09      EExp   6.45   7.88   .29
emin   6.83   8.05   .35      IanL   6.67   8.50   .28
DQin   6.86   8.57   .18      NuFi   7.22   9.24   .19
Both 'sides' still tied, though everyone is worse by an avg of .50
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

Yeah it was a rough night apparently.

I went down .44. Pelton, .67. Ncs just .27.

Several higher remain within reach. Conceivably in a night. More possible in a week. With 5 months to go.
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2023-24 team win projection contest

Post by Mike G »

How does this read?

Code: Select all

 Atl      Bos      Brk      Cha      Chi      Cle      Det      Ind  
b-r 47   me  60   nc  45   em  36   DQ  42   me  53   eW  35   b-r 44
DQ  46   vz  59   b-r 42   eW  36   nc  40   DQ  51   b-r 32   eW  39
Cr  44   nc  57   Cr  42   vz  33   vz  39   dt  50   nc  28   dt  38
vz  43   em  57   vz  41   nc  33   Cr  39   tz  49   tz  28   tz  37
dt  43   Cr  57   dt  40   dt  33   eW  39   vz  49   Cr  27   Cr  35
nc  42   b-r 56   me  40   b-r 33   tz  37   em  49   dt  26   vz  34
tz  42   DQ  56   tz  40   tz  31   dt  36   Cr  49   DQ  25   nc  34
em  40   dt  56   em  39   Cr  31   me  36   nc  48   me  24   em  33
me  39   tz  55   DQ  38   DQ  30   b-r 34   eW  44   vz  22   me  33
eW  39   eW  52   eW  36   me  27   em  32   b-r 40   em  21   DQ  32
                                             
 Mia      Mil      NYK      Orl      Phl      Tor      Was        
DQ  45   em  56   vz  56   me  41   b-r 55   b-r 44   b-r 35      
tz  44   eW  53   me  51   b-r 41   DQ  54   dt  42   nc  34      
nc  43   vz  52   nc  48   eW  40   vz  52   Cr  42   DQ  34      
vz  43   DQ  51   tz  47   DQ  40   nc  51   DQ  42   eW  33      
dt  42   dt  50   dt  47   em  37   me  50   vz  41   vz  33      
Cr  42   me  49   Cr  46   tz  37   tz  48   nc  41   Cr  31      
eW  42   tz  48   b-r 46   dt  34   em  48   eW  40   tz  28      
em  41   nc  47   DQ  45   Cr  33   dt  46   tz  40   dt  28      
me  41   Cr  47   em  44   nc  33   Cr  44   em  38   me  28      
b-r 39   b-r 43   eW  42   vz  33   eW  38   me  38   em  20      
                                             
 Dal      Den      GSW      Hou      LAC      LAL      Mem      Min  
b-r 46   em  60   em  52   b-r 43   DQ  45   DQ  52   dt  47   b-r 55
dt  44   me  54   nc  50   me  39   eW  44   eW  45   tz  46   em  53
nc  43   b-r 54   dt  48   eW  36   me  44   tz  44   Cr  45   me  48
DQ  42   Cr  53   vz  47   dt  34   Cr  43   me  43   eW  45   Cr  48
tz  41   tz  51   b-r 47   Cr  32   vz  43   em  43   DQ  45   dt  47
me  41   vz  49   tz  46   tz  31   b-r 43   nc  42   em  44   nc  47
eW  40   DQ  48   Cr  46   vz  31   dt  42   dt  41   me  44   vz  46
vz  40   dt  48   me  45   nc  31   tz  41   Cr  40   nc  43   tz  46
em  39   nc  48   eW  41   em  29   nc  41   vz  39   vz  43   DQ  41
Cr  39   eW  45   DQ  37   DQ  25   em  40   b-r 34   b-r 29   eW  39
                                             
 NOP      OKC      Phx      Por      Sac      SAS      Uta        
DQ  48   me  48   dt  49   eW  39   em  49   em  36   DQ  41      
em  48   b-r 44   DQ  48   b-r 34   Cr  47   eW  35   Cr  40      
Cr  46   tz  44   tz  48   nc  33   me  45   b-r 30   eW  39      
eW  45   eW  44   me  48   dt  32   vz  43   nc  30   vz  39      
vz  45   dt  41   vz  47   tz  31   nc  41   dt  28   nc  38      
tz  45   em  41   Cr  46   Cr  29   eW  41   tz  28   dt  37      
dt  44   Cr  40   eW  46   DQ  29   tz  41   DQ  27   tz  37      
me  43   vz  37   nc  45   vz  28   DQ  39   Cr  27   me  34      
nc  43   DQ  35   em  45   em  28   dt  37   me  24   em  32      
b-r 38   nc  33   b-r 39   me  21   b-r 35   vz  24   b-r 30      
b-r is the daily projection; other numbers are set.
v-zero
Posts: 520
Joined: Sat Oct 27, 2012 12:30 pm

Re: 2023-24 team win projection contest

Post by v-zero »

Looks good, easy to follow. Could also take out the b-r value and simply sort them by the absolute value of the difference between the entrant's prediction and the b-r value.
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: 2023-24 team win projection contest

Post by Mike G »

Well, this way it's clear whether you wish, for example, the Knicks would do better (you do) or worse (not).

Last year I threw Pelton in with 'our' guesses, because he was the one to beat for quite a while. I could still do that with DARKO.
Sometimes a team's success will help your avg error but not help you with regard to catching the leader.

OK, best to worst, + drko:

Code: Select all

tm    DKO  dtk  trz  ncs  Cro  med  vzr  eW   emi  DQ
Atl    4    4    6    5    3    8    4    8    7    2
Bos    1    0    2    1    1    4    2    5    1    0
Brk    1    2    3    3    0    2    1    7    3    5
Cha    2    0    2    0    2    6    1    3    3    3
Chi    4    2    3    6    5    2    5    5    2    7
Cle    7   10   10    9    9   14   10    5    9   12
Det    4    6    5    4    5    8   11    3   11    7
Ind   10    6    7   10    9   11   10    5   11   12

tm    DKO  dtk  trz  ncs  Cro  med  vzr  eW   emi  DQ
Mia    3    3    5    4    3    2    4    3    2    6
Mil    2    7    5    5    4    6   10   10   13    8
NYK    1    1    2    2    0    5   10    4    2    1
Orl    4    7    4    8    8    0    8    1    4    2
Phl    5    9    7    4   11    5    3   17    7    1
Tor    2    2    4    4    2    6    3    4    6    3
Was    3    7    7    1    4    8    2    2   15    2

tm    DKO  dtk  trz  ncs  Cro  med  vzr  eW   emi  DQ
Dal    5    2    4    3    7    4    6    5    7    4
Den    5    6    3    6    1    1    5    8    6    6
GSW    2    1    1    3    1    2    0    6    5   10
Hou    7    9   12   13   11    4   13    7   14   18
LAC    0    1    2    2    0    1    0    1    3    3
LAL    7    8   10    9    7   10    6   11   10   18
Mem   13   18   17   14   16   15   14   16   15   16
Min    6    8    9    8    7    7    9   16    2   14

tm    DKO  dtk  trz  ncs  Cro  med  vzr  eW   emi  DQ
NOP    3    6    7    5    8    6    7    7   10   10
OKC    8    3    1   11    4    3    7    1    3   10
Phx    8   10    9    6    7    9    8    7    6    9
Por    3    2    2    1    5   13    6    6    6    5
Sac    4    2    5    6   12   10    8    6   14    4
SAS    4    2    2    0    3    6    6    5    6    2
Uta    7    8    7    9   11    4   10   10    3   12
DARKO has only a couple of best guesses for teams, but nowhere is he worst; very few are worse than avg.
His biggest error is also the best guess for Memphis.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

Thanks for those tables. Was wanting but didn't want to rush.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2023-24 team win projection contest

Post by Crow »

I am tied with Darko for variances from BRef current projection by 2 or less but have 5 more of 7 or greater.
Post Reply