2023-24 team win projection contest

[Crow]

With the departure of stars, there were more opportunities and stats for the others to pick-up.

[Crow]

For Cavs, it is mainly a collapse of the offense but the former #1 defense also lost considerable ground from last season.

Own FG% and turnovers way worse. More fouls / opponent FTM and fewer opponents turnovers .

[Mike G]

Code: Select all

Atl       Bos      Brk      Cha      Chi      Cle      Det      Ind   
DQ  46   med 60   b-r 47   em  36   DQ  42   med 53   eW  35   b-r 46
Cr  44   b-r 59   nc  45   eW  36   nc  40   DQ  51   nc  28   eW  39
vz  43   vz  59   Cr  42   vz  33   vz  39   dt  50   tz  28   dt  38
dt  43   nc  57   vz  41   nc  33   Cr  39   tz  49   Cr  27   tz  37
nc  42   em  57   dt  40   dt  33   eW  39   vz  49   dt  26   Cr  35
tz  42   Cr  57   med 40   tz  31   tz  37   em  49   DQ  25   vz  34
em  40   DQ  56   tz  40   Cr  31   dt  36   Cr  49   med 24   nc  34
med 39   dt  56   em  39   DQ  30   med 36   nc  48   vz  22   em  33
eW  39   tz  55   DQ  38   b-r 27   b-r 33   eW  44   em  21   med 33
b-r 39   eW  52   eW  36   med 27   em  32   b-r 44   b-r 18   DQ  32
                                             
Mia       Mil      NYK      Orl      Phl      Tor      Was         
DQ  45   em  56   vz  56   b-r 51   b-r 57   dt  42   nc  34      
tz  44   eW  53   med 51   med 41   DQ  54   Cr  42   DQ  34      
b-r 43   vz  52   nc  48   eW  40   vz  52   DQ  42   eW  33      
nc  43   DQ  51   tz  47   DQ  40   nc  51   vz  41   vz  33      
vz  43   dt  50   dt  47   em  37   med 50   nc  41   Cr  31      
dt  42   b-r 50   Cr  46   tz  37   tz  48   eW  40   tz  28      
Cr  42   med 49   b-r 45   dt  34   em  48   tz  40   dt  28      
eW  42   tz  48   DQ  45   Cr  33   dt  46   em  38   med 28      
em  41   nc  47   em  44   nc  33   Cr  44   med 38   b-r 20      
med 41   Cr  47   eW  42   vz  33   eW  38   b-r 38   em  20      
                                             
Dal       Den      GSW      Hou      LAC      LAL      Mem      Min   
b-r 47   em  60   em  52   b-r 48   b-r 46   DQ  52   dt  47   b-r 54
dt  44   med 54   nc  50   med 39   DQ  45   b-r 45   tz  46   em  53
nc  43   Cr  53   dt  48   eW  36   eW  44   eW  45   Cr  45   med 48
DQ  42   tz  51   vz  47   dt  34   med 44   tz  44   eW  45   Cr  48
tz  41   b-r 50   tz  46   Cr  32   Cr  43   med 43   DQ  45   dt  47
med 41   vz  49   Cr  46   tz  31   vz  43   em  43   em  44   nc  47
eW  40   DQ  48   med 45   vz  31   dt  42   nc  42   med 44   vz  46
vz  40   dt  48   b-r 43   nc  31   tz  41   dt  41   nc  43   tz  46
em  39   nc  48   eW  41   em  29   nc  41   Cr  40   vz  43   DQ  41
Cr  39   eW  45   DQ  37   DQ  25   em  40   vz  39   b-r 26   eW  39
                                             
NOP       OKC      Phx      Por      Sac      SAS      Uta         
DQ  48   b-r 54   dt  49   eW  39   em  49   em  36   DQ  41      
em  48   med 48   DQ  48   nc  33   Cr  47   eW  35   Cr  40      
Cr  46   tz  44   tz  48   dt  32   med 45   nc  30   eW  39      
eW  45   eW  44   med 48   tz  31   b-r 44   dt  28   vz  39      
vz  45   dt  41   vz  47   Cr  29   vz  43   tz  28   nc  38      
tz  45   em  41   Cr  46   DQ  29   nc  41   DQ  27   dt  37      
dt  44   Cr  40   eW  46   vz  28   eW  41   Cr  27   tz  37      
b-r 44   vz  37   nc  45   em  28   tz  41   med 24   med 34      
med 43   DQ  35   em  45   b-r 26   DQ  39   vz  24   em  32      
nc  43   nc  33   b-r 42   med 21   dt  37   b-r 20   b-r 26      

https://www.basketball-reference.com/fr ... _prob.html

[Crow]

Lots of folks not doing so well. Me, Darko, LeBron, others...

[Mike G]

We really don't know. The order changes rapidly. DARKO was the gold standard, until they weren't.
mediocre's huge lead may diminish but will probably hold up due to predicted success of Orl, Hou, Min, OKC; still more to the plunging fortunes of Was, Por, SAS, and Utah.
More hope is available (to the rest of us) should we see even modest improvements by Atl, Cha, Chi, Tor, NOP; or any slumping from Boston.

Code: Select all

.  avg err   rmse   r^2     .   avg err   rmse   r^2
medi   4.83   6.4   .68      Crow   6.77   8.5   .43
ChKl   5.89   7.6   .54      vzro   6.82   8.6   .43
vegas  5.98   7.7   .53      KPel   6.90   8.8   .40
EExp   6.01   7.7   .53      DRKO   7.01   8.2   .47
IanL   6.09   7.7   .55      ncs.   7.17   9.1   .35
avgA   6.36   7.9   .51      23re   7.23   9.3   .35
trzu   6.41   7.9   .51      LEBR   7.47   8.8   .39
emin   6.47   8.3   .48      DQin   7.73   9.6   .31
EBPI   6.69   8.1   .49      NuFi   7.76  10.4   .26
dtka   6.74   8.3   .46      eWin   7.78   9.7   .27

In the last week, we've had 4 different entries in the #2 slot.

[v-zero]

DARKO doing poorly now is heartening frankly, as I happen to know that's an excellent model.

[Mike G]

Good attitude there ^^
Season worst average errors almost daily of late. But top to bottom disparity peaked at 3.16 a couple days ago.

Code: Select all

.  avg err   rmse   r^2     .   avg err   rmse   r^2
medi   5.13   6.4   .71      dtka   6.98   8.5   .49
EExp   6.16   7.9   .55      EBPI   7.01   8.3   .52
vegas  6.26   8.0   .54      KPel   7.19   9.0   .43
ChKl   6.33   8.0   .53      DRKO   7.30   8.4   .51
IanL   6.36   8.0   .55      ncs.   7.48   9.2   .40
avgA   6.52   8.1   .54      23re   7.56   9.5   .39
emin   6.58   8.4   .51      LEBR   7.76   9.0   .44
trzu   6.70   8.1   .54      DQin   8.00   9.7   .34
Crow   6.91   8.6   .47      NuFi   8.09  10.4   .29
vzro   6.92   8.5   .48      eWin   8.10  10.1   .28

Oh and mediocre's lead hit 1.19 on Sat. also

[Crow]

.4 from 6th. That's my immediate goal.

[DarkStar48]

DARKO doing poorly now is heartening frankly, as I happen to know that's an excellent model.

ESPN "Expert Forecast" doing so well now is heartening frankly, as I happen to know that's a horrible model.

[Crow]

67+% of season to go for teams and model match-ups.

Actual results don't necessary perfectly track the concept of "true talent", which models may be based on.

[v-zero]

ESPN "Expert Forecast" doing so well now is heartening frankly, as I happen to know that's a horrible model.

Hah, very good!

[Mike G]

Last year, being .90 from the lead would put you 12th of 20.

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
medi   4.83   6.0   .75      dtka   6.60   8.2   .53
EExp   5.73   7.6   .59      EBPI   6.79   8.0   .57
vegas  5.90   7.6   .59      DRKO   6.84   8.2   .55
ChKl   6.05   7.8   .56      KPel   7.02   8.8   .47
avgA   6.16   7.8   .60      ncs.   7.28   9.0   .43
IanL   6.25   7.8   .57      LEBR   7.39   8.8   .48
trzu   6.33   7.8   .59      23re   7.47   9.3   .44
vzro   6.34   8.1   .53      NuFi   7.78  10.1   .33
Crow   6.42   8.2   .53      DQin   7.82   9.4   .38
emin   6.50   8.1   .54      eWin   7.99   9.9   .32

[Crow]

.18 from 6th.

[Mike G]

Segregated update

Code: Select all

.  avg err   rmse   r^2      .  avg err   rmse   r^2
medi   4.89   6.1   .74      EExp   5.75   7.7   .58
trzu   6.42   8.0   .58     vegas   6.00   7.8   .58
emin   6.46   8.2   .54      ChKl   6.05   7.9   .56
vzro   6.51   8.3   .52      IanL   6.20   7.9   .57
Crow   6.52   8.4   .52      EBPI   6.91   8.2   .56
dtka   6.84   8.4   .52      DRKO   7.02   8.4   .54
ncs.   7.45   9.1   .42      KPel   7.19   9.0   .46
DQin   7.93   9.6   .36      LEBR   7.53   9.0   .47
eWin   8.11  10.1   .31      NuFi   7.96  10.3   .32
avg    6.79   8.5   .50             6.73   8.5   .51

There was some recalibration at b-r.com. Projections changed almost randomly, whether a team won, lost, or didn't play.

The nth root of the sum of the nth power of errors doesn't have to be limited to n = 1 or 2. Here are the leaders at various exponents:

Code: Select all

0.03              0.10             0.30             1.00   
medi   1.20      medi   2.73      medi   3.67      medi   4.89
trzu   1.53      EExp   3.31      EExp   3.88      EExp   5.75
eWin   1.94      trzu   3.47      vegas  4.26      vegas  6.00
EExp   3.12      vegas  3.67      ChKl   4.33      ChKl   6.05
vegas  3.45      ChKl   3.80      Crow   4.66      IanL   6.20
                              
3.00              10.0             30               100   
medi   7.12      medi   11.7      medi   14.6      medi   15.8516
vegas  9.10      EBPI   12.6      IanL   14.7      IanL   15.8521
EExp   9.12      IanL   12.7      EBPI   14.8      EBPI   15.95
IanL   9.16      vegas  13.2      ChKl   15.5      ChKl   16.6
trzu   9.18      ChKl   13.3      Crow   15.7      Crow   16.9

Mediocre's lead shrinks but never is lost at higher exponent, because they have the smallest worst errors.
I don't recall one entry ever leading at every exponent.

[Crow]

OK, .11 from 2nd...in the contest.