2023-24 team win projection contest
Re: 2023-24 team win projection contest
I'll take the beginner's luck. Hope it holds
Re: 2023-24 team win projection contest
Since I am overwriting last year's contest log, I notice dis/similarities with this season's progress. Same date both years:
The 3.20 at bottom shows identical top-to-bottom spread.
1.0 year ago, Pelton's lead was greater than this year's frontrunner (82-75). He would end up losing all that and another .73
Big turnaround for ESPNxperts -- so far.
Code: Select all
12-29-23 .75 12-29-22 .82
medi 4.90 KPel 4.34
EExp 5.65 ncs. 5.16
vegas 5.92 DRKO 5.18
ChKl 5.97 dtka 5.21
IanL 6.09 EBPI 5.37
avgA 6.45 trzu 5.40
emin 6.50 Crow 5.45
trzu 6.50 LEBR 5.46
vzro 6.77 vzro 5.70
EBPI 6.81 emin 5.86
dtka 6.93 BIPM 6.15
Crow 7.03 538R 6.35
KPel 7.17 TmRk 6.44
DRKO 7.27 nuFi 6.49
ncs. 7.58 vegas 6.51
LEBR 7.64 4141 6.57
23re 7.75 538E 6.60
NuFi 7.83 MPra 6.96
DQin 7.95 AnBa 7.08
eWin 8.10 EExp 7.55
3.20 6.84 3.20 5.97
1.0 year ago, Pelton's lead was greater than this year's frontrunner (82-75). He would end up losing all that and another .73
Big turnaround for ESPNxperts -- so far.
Re: 2023-24 team win projection contest
Not much change in the order lately, but it has been a while.
Outsiders are kickin our asses by .20 avg. Without medi, it would be .43
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
medi 5.18 6.3 .74 Crow 7.14 8.6 .51
EExp 5.84 7.6 .61 dtka 7.14 8.5 .52
vegas 5.97 7.7 .60 KPel 7.32 9.1 .46
ChKl 6.08 7.8 .59 DRKO 7.44 8.7 .51
IanL 6.30 7.8 .58 LEBR 7.68 9.2 .46
emin 6.60 8.4 .52 23re 7.69 9.6 .44
avgA 6.64 8.1 .58 ncs. 7.75 9.4 .40
trzu 6.70 8.0 .59 NuFi 7.83 10.3 .32
vzro 6.94 8.6 .50 DQin 7.88 9.6 .38
EBPI 6.96 8.2 .57 eWin 7.97 10.0 .36
Re: 2023-24 team win projection contest
.2 is about 3% difference.
That isn't much difference and much of the currently appearing difference could be fairly random variance.
That isn't much difference and much of the currently appearing difference could be fairly random variance.
Re: 2023-24 team win projection contest
Best- and worst-looking guesses at 40% of the season. Ties (within 0.5 roughly) included.
These are best and worst current errors among APBR entries listed.
Of course the magnitude of the miss is not shown. Some of the 'worst' are not bad, and some of the 'best' are only good relative to others.
Code: Select all
apbr best worst
medi Atl Bos Cha Orl Tor Cle Mia
. Den GSW Hou OKC Sac SAS
emin Brk Det NYK Was Cha Chi Mia
. Min Phx Por Uta Dal GSW LAC SAS
trzu Brk Chi NOP Mil
vzro Chi Mem NOP Por SAS NYK Orl
Crow Chi Den NOP Sac Mil Orl Tor Dal
dtka Dal LAL Tor Mem Phx Sac
ncs. Phx Brk Mil Orl Was
. LAC OKC
DQin Brk Mia NYK Phl NOP Atl Ind Hou LAL Uta
eWin Atl Chi Cle Mil NOP Bos Cha Det Phl
. Den Min PorOf course the magnitude of the miss is not shown. Some of the 'worst' are not bad, and some of the 'best' are only good relative to others.
Re: 2023-24 team win projection contest
Teams ordered by their overachievement relative to the avg of our preseason guesses.
The disappointing teams are at the bottom.
Code: Select all
OKC Hou Min Orl Phl Ind LAC
b-r 59 b-r 46 b-r 57 b-r 47 b-r 58 b-r 45 b-r 49
med 48 med 39 em 53 med 41 DQ 54 eW 39 DQ 45
tz 44 eW 36 med 48 eW 40 vz 52 dt 38 eW 44
eW 44 dt 34 Cr 48 DQ 40 nc 51 tz 37 med 44
dt 41 Cr 32 dt 47 em 37 med 50 Cr 35 Cr 43
em 41 tz 31 nc 47 tz 37 tz 48 vz 34 vz 43
Cr 40 vz 31 vz 46 dt 34 em 48 nc 34 dt 42
vz 37 nc 31 tz 46 Cr 33 dt 46 em 33 tz 41
DQ 35 em 29 DQ 41 nc 33 Cr 44 med 33 nc 41
nc 33 DQ 25 eW 39 vz 33 eW 38 DQ 32 em 40
Bos Den Mil Sac NOP Mia Dal
b-r 62 em 60 em 56 em 49 DQ 48 DQ 45 dt 44
med 60 med 54 b-r 53 Cr 47 em 48 tz 44 nc 43
vz 59 b-r 54 eW 53 b-r 45 b-r 47 b-r 44 b-r 42
nc 57 Cr 53 vz 52 med 45 Cr 46 nc 43 DQ 42
em 57 tz 51 DQ 51 vz 43 eW 45 vz 43 tz 41
Cr 57 vz 49 dt 50 nc 41 vz 45 dt 42 med 41
DQ 56 DQ 48 med 49 eW 41 tz 45 Cr 42 eW 40
dt 56 dt 48 tz 48 tz 41 dt 44 eW 42 vz 40
tz 55 nc 48 nc 47 DQ 39 med 43 em 41 em 39
eW 52 eW 45 Cr 47 dt 37 nc 43 med 41 Cr 39
Chi GSW NYK LAL Brk Tor Cle Atl
DQ 42 em 52 vz 56 DQ 52 nc 45 dt 42 med 53 DQ 46
nc 40 nc 50 med 51 eW 45 Cr 42 Cr 42 DQ 51 Cr 44
vz 39 dt 48 nc 48 tz 44 vz 41 DQ 42 dt 50 vz 43
Cr 39 vz 47 tz 47 med 43 dt 40 vz 41 tz 49 dt 43
eW 39 tz 46 dt 47 em 43 med 40 nc 41 vz 49 nc 42
b-r 38 Cr 46 Cr 46 nc 42 tz 40 eW 40 em 49 tz 42
tz 37 med 45 b-r 45 dt 41 em 39 tz 40 Cr 49 em 40
dt 36 b-r 44 DQ 45 b-r 41 DQ 38 em 38 nc 48 med 39
med 36 eW 41 em 44 Cr 40 b-r 37 med 38 b-r 45 eW 39
em 32 DQ 37 eW 42 vz 39 eW 36 b-r 36 eW 44 b-r 37
Uta Phx Por Was Det Cha SAS Mem
DQ 41 dt 49 eW 39 nc 34 eW 35 em 36 em 36 dt 47
Cr 40 DQ 48 nc 33 DQ 34 nc 28 eW 36 eW 35 tz 46
eW 39 tz 48 dt 32 eW 33 tz 28 vz 33 nc 30 Cr 45
vz 39 med 48 tz 31 vz 33 Cr 27 nc 33 dt 28 eW 45
nc 38 vz 47 Cr 29 Cr 31 dt 26 dt 33 tz 28 DQ 45
dt 37 Cr 46 DQ 29 tz 28 DQ 25 tz 31 DQ 27 em 44
tz 37 eW 46 vz 28 dt 28 med 24 Cr 31 Cr 27 med 44
med 34 nc 45 em 28 med 28 vz 22 DQ 30 med 24 nc 43
b-r 32 em 45 b-r 24 b-r 20 em 21 med 27 vz 24 vz 43
em 32 b-r 41 med 21 em 20 b-r 16 b-r 22 b-r 17 b-r 28
Re: 2023-24 team win projection contest
Noticing how several of the most unexpectedly successful teams are also among the youngest (on-court) in the league, I yet again wonder how much an age adjustment would help in making predictions. Younger players tend to be better than they were last year, while older guys are on the decline. Right?
So, ranked by 'overachievement' relative to our avg predictions; with team age above or below the league avg of 26.5 :
While the top 3 'overs' are at least a year younger than NBA avg, the 6 worst performers are also young.
Correlation between the columns is .18 -- older teams are less likely to be huge overachievers, but even more unlikely to be major flops.
And this may be a relatively injury-free season, thus far.
So, ranked by 'overachievement' relative to our avg predictions; with team age above or below the league avg of 26.5 :
Code: Select all
over tm old over tm old
18.6 OKC -2.4 -17.3 Mem -1.3
14.5 Hou -1.2 -11.4 SAS -3.1
10.4 Ind -1.3 -10.8 Was -1.1
10.2 Phl 2.0 -10.6 Cha -1.1
9.9 Min 0.9 -10.3 Det -2.9
9.7 Orl -2.7 -7.0 Por -2.0
7.4 LAC 4.1 -6.7 Phx 2.4
5.9 Bos 1.9 -4.9 Uta -0.8
3.3 Den 0.4 -4.0 Tor 0.0
2.7 Sac 0.0 -3.6 Atl -0.2
2.7 NOP -0.8 -3.5 Brk 0.0
2.3 Mia 2.0 -3.4 Cle -0.3
2.2 Dal -0.2 -3.2 LAL 1.4
1.8 Mil 3.4 -1.9 GSW 2.5
. -1.1 NYK -0.3
. -0.8 Chi 1.4Correlation between the columns is .18 -- older teams are less likely to be huge overachievers, but even more unlikely to be major flops.
And this may be a relatively injury-free season, thus far.
Re: 2023-24 team win projection contest
Average gains or loss per year for teams for x, y, z years
from their win min or max would be interesting.
Correlation to overall spending or spending trend would be another analysis and could even check correlations for # maxes and progression or with # of rookies scale deals.
# of closest projections has 2 clear leaders then a pack.
Need .2 relative improvement to get to 4th in contest (not counting APBR average).
Come March, the segregated lists will offer clarity on that dimension.
from their win min or max would be interesting.
Correlation to overall spending or spending trend would be another analysis and could even check correlations for # maxes and progression or with # of rookies scale deals.
# of closest projections has 2 clear leaders then a pack.
Need .2 relative improvement to get to 4th in contest (not counting APBR average).
Come March, the segregated lists will offer clarity on that dimension.
Re: 2023-24 team win projection contest
Not much new but for another season-worst avg-of-all, and biggest top to bottom disparity, and mediocre's biggest separation from the pack since mid-Dec.
"We" are now essentially tied with "them" on avg. Without medi we are .30 worse.
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
medi 4.98 6.2 .76 dtka 7.16 8.5 .56
EExp 6.10 7.6 .63 EBPI 7.42 8.5 .56
ChKl 6.17 7.8 .61 DRKO 7.48 8.8 .53
vegas 6.26 7.8 .62 KPel 7.53 9.2 .49
IanL 6.43 7.8 .60 LEBR 7.83 9.4 .46
avgA 6.71 8.1 .61 ncs. 7.86 9.5 .41
trzu 6.76 8.1 .61 DQin 8.01 9.8 .37
vzro 6.84 8.4 .54 23re 8.07 9.9 .43
emin 6.87 8.5 .53 eWin 8.38 10.2 .38
Crow 7.05 8.5 .55 NuFi 8.47 10.5 .32Re: 2023-24 team win projection contest
Without Vegas they'd be worse, right? Excluding one is a perspective but not my preference.
Re: 2023-24 team win projection contest
On avg, the field is .01 worse than yesterday; but with much variation, as those near the top took it badly, while the bottom tier prospered.
Everyone else within .01 of their yesterday's MAE.
UPDATE Jan.10
UPDATE Jan.11
Code: Select all
delta who jan7 jan8
-.18 eWin 8.38 8.21
-.16 DQin 8.01 7.85
-.08 NuFi 8.47 8.39
-.03 LEBR 7.83 7.80
0.03 vzro 6.84 6.87
0.04 emin 6.87 6.92
0.11 medi 4.98 5.09
0.11 ChKl 6.17 6.28
0.11 Crow 7.05 7.16
0.18 EExp 6.10 6.28
0.27 IanL 6.43 6.70UPDATE Jan.10
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
medi 5.06 6.2 .76 dtka 7.17 8.4 .56
EExp 6.24 7.6 .63 EBPI 7.45 8.4 .56
vegas 6.29 7.7 .62 DRKO 7.47 8.8 .52
ChKl 6.29 7.8 .60 KPel 7.54 9.0 .50
avgA 6.72 8.0 .61 LEBR 7.81 9.4 .45
IanL 6.72 7.9 .58 DQin 7.82 9.6 .39
trzu 6.75 8.0 .62 ncs. 7.86 9.5 .41
vzro 6.93 8.3 .54 23re 8.00 9.7 .44
emin 7.07 8.5 .52 eWin 8.27 10.1 .38
Crow 7.10 8.4 .56 NuFi 8.64 10.5 .31Code: Select all
. avg err rmse r^2 . avg err rmse r^2
medi 5.26 6.3 .74 emin 7.20 8.6 .52
EExp 6.22 7.6 .62 EBPI 7.44 8.5 .56
ChKl 6.24 7.9 .59 DRKO 7.46 8.9 .51
vegas 6.37 7.8 .62 KPel 7.56 9.1 .49
avgA 6.70 8.0 .61 DQin 7.72 9.5 .40
IanL 6.74 8.0 .58 ncs. 7.83 9.5 .40
trzu 6.78 8.0 .62 LEBR 7.88 9.4 .45
vzro 6.90 8.4 .54 23re 7.97 9.7 .44
Crow 7.06 8.3 .56 eWin 8.16 10.1 .39
dtka 7.20 8.4 .55 NuFi 8.78 10.5 .31Re: 2023-24 team win projection contest
Based in January 7th, I was 0.3 from second in contest. January 8 list doesn't show tarrazu. But I am still fairly close to 2nd or 3rd.
Re: 2023-24 team win projection contest
Absolute errors to date:
Outlier predictions -- above or below everyone else by 3 or more -- have historically not been good for your averages. This year they are about 50/50, but not exactly random. Mediocre is 4-0 on his wild guesses, and I am 1-5.
Meanwhile, the whole lot are at alltime high avg error. Again.ESPXperts almost a month in the #2 slot.
Code: Select all
tm medi vzro trzu Crow emin dtka DQin ncs eWin
Atl 4 8 7 9 5 8 11- 7 4
Bos 3 4 8 6 6 7 6 6 11-
Brk 4 5 4 6 3 4 2 9 0
Cha 8* 15 12 12 17 14 12 14 17
Chi 2 1 1 1 6 2 4 2 1
Cle 6 3 3 2 2 3 4 2 2
Det 11 8 14 14 8 13 12 14 21-
Ind 15 14 11 13 15 10 16 14 9
Mia 4 2 1 3 4 3 0 2 3
Mil 4 0 5 6 4 3 2 5 1
tm medi vzro trzu Crow emin dtka DQin ncs eWin
NYK 2 7 2 3 5 2 5 1 7
Orl 4 13 9 12 8 11 6 12 5
Phl 4 3 6 10 6 8 0* 4 16-
Tor 3 6 5 7 3 7 7 6 6
Was 8 14 9 12 1* 9 14 15 14
Dal 3 4 3 5 5 0 3 1 4
Den 1 4 3 0 7 5 5 6 8
GSW 5 7 6 6 12- 8 3 9 0*
Hou 4* 12 12 11 14 9 18- 12 7
LAC 8 9 11 9 12 10 7 11 8
tm medi vzro trzu Crow emin dtka DQin ncs eWin
LAL 5 1 5 2 5 3 14- 4 6
Mem 13 12 16 14 13 16 14 13 14
Min 9 11 12 9 4* 10 16 11 18-
NOP 7 5 5 4 2 6 2 7 5
OKC 11* 21 14 18 17 17 23 25 15
Phx 8 7 8 6 5 9 8 5 6
Por 0* 7 10 8 7 11 8 12 18-
Sac 0 1 4 2 4 8 6 4 4
SAS 4 4 8 8 17 9 8 10 16
Uta 5 1 2 2 7 2 3 0 1
Meanwhile, the whole lot are at alltime high avg error. Again.
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
medi 5.51 6.6 .74 EBPI 7.52 8.7 .55
EExp 6.38 7.7 .63 dtka 7.52 8.7 .55
vegas 6.58 8.0 .62 KPel 7.79 9.4 .48
ChKl 6.60 8.1 .58 DRKO 7.85 9.2 .49
IanL 6.82 8.1 .58 DQin 7.90 9.8 .39
vzro 6.96 8.6 .54 ncs. 8.09 9.8 .40
avgA 7.00 8.3 .60 23re 8.25 9.9 .45
trzu 7.14 8.3 .61 LEBR 8.26 9.7 .43
Crow 7.32 8.6 .55 eWin 8.27 10.3 .40
emin 7.42 8.8 .51 NuFi 8.96 10.7 .31
Re: 2023-24 team win projection contest
ESPXperts is an average, I believe of dozens. There is wisdom inside but a lot of the wisdom is the wisdom of the mass average.
The APBR average shows the same dynamic, if not the same level of accuracy.
The APBR average shows the same dynamic, if not the same level of accuracy.