Page 4 of 18
Re: 2024-25 team win projection contest
Posted: Sun Nov 03, 2024 2:46 am
by TeemoTeejay
well all of our 76ers predictions are gonna be off now with embiid being out for punching someone lol
Re: 2024-25 team win projection contest
Posted: Sun Nov 03, 2024 3:29 pm
by Mike G
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
24pr 5.44 6.9 .26 perW 7.41 9.0 .07
4141 5.86 7.6 dtka 7.54 9.0 .26
WShr 6.06 7.2 .30 DQin 8.12 9.7 .26
bpmW 6.25 7.3 .30 emin 8.17 9.8 .26
TmTj 6.99 9.0 .30 Crow 8.24 9.9 .27
avgA 7.04 8.4 .26 medi 8.29 10.2 .25
eWin 7.24 8.8 .10 24py 8.68 10.6 .26
Re: 2024-25 team win projection contest
Posted: Sun Nov 03, 2024 8:03 pm
by Crow
Might have some regrets on 5 projections. But it is early. Too much? We'll see.
Re: 2024-25 team win projection contest
Posted: Sun Nov 03, 2024 9:04 pm
by Crow
As you adjust the measuring stick, let us know what it is currently.
2.8 from lead, 1.3 from 5th at the moment.
Re: 2024-25 team win projection contest
Posted: Mon Nov 04, 2024 12:57 am
by TeemoTeejay
It looks like not predicting a wide range leads to better results, I’ve noticed when I saw LEBRON projections a few years back it was more narrow like that
I wonder how mine would look if the range was more 20-55 wins than what I went with
Re: 2024-25 team win projection contest
Posted: Mon Nov 04, 2024 1:10 am
by Crow
The ranking of 41-41 indicates how favored regressed entries are right now.
24pr is prior season regressed halfway to 41?
Re: 2024-25 team win projection contest
Posted: Mon Nov 04, 2024 2:25 pm
by Mike G
Crow wrote: ↑Mon Nov 04, 2024 1:10 am
The ranking of 41-41 indicates how favored regressed entries are right now.
24pr is prior season regressed halfway to 41?
Yes, it's (24py + 41)/2
Right now the b-r.com projections are totaling 1200 wins or 40.0 avg; so I am adding one win per team.
I'm not de-regressing. They have in the past been abruptly adjusted, so that's a possibility.
https://www.basketball-reference.com/fr ... _prob.html
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
24pr 5.71 7.1 .23 perW 7.22 8.7 .11
WShr 6.02 7.2 .31 dtka 7.48 9.0 .27
bpmW 6.13 7.2 .33 DQin 8.16 9.6 .27
4141 6.18 7.7 medi 8.26 10.3 .24
TmTj 6.92 8.9 .32 Crow 8.38 9.9 .27
eWin 7.00 8.5 .13 emin 8.44 9.8 .26
avgA 7.05 8.3 .28 24py 8.65 10.8 .23
Here are rankings of our average deviations from last year's Pythagorean Wins, and from the regressed version.
Code: Select all
24py avg. dev. 24pr
7.1 perW 6.1 Crow
6.8 eWin 5.9 emin
5.4 bpmW 5.6 medi
5.2 emin 5.1 DQin
5.2 Crow 4.9 TmTj
5.1 TmTj 4.6 dtka
5.0 DQin 4.2 perW
4.6 WShr 3.8 eWin
4.3 dtka 3.0 bpmW
4.3 medi 2.2 WShr
Win Shares is leading by sticking close to both; like in between.
Re: 2024-25 team win projection contest
Posted: Mon Nov 04, 2024 3:11 pm
by Crow
Every point on current SRS appears to change BRef projection by about 1 win generally but only a half win at the very bottom.
I think +1 point on net margin has been estimated in other work to be "worth" +2.7 wins, on average. Correct me if I'm wrong; but based on that (or a little less), there is a lot of regression from current results. Certainly should be some but how much? It will decline with time and the credibility of the projection will gradually improve.
Re: 2024-25 team win projection contest
Posted: Wed Nov 06, 2024 10:49 am
by Mike G
Over- and under-achieving teams relative to our AvgA
Code: Select all
over tm avgA b-r over tm avgA b-r
12.8 Was 22.9 35.7 -18.3 Phl 48.6 30.3
11.1 Por 25.7 36.8 -14.4 NOP 45.4 31.0
10.5 Brk 27.0 37.5 -11.4 Mil 45.9 34.5
9.9 SAS 33.7 43.6 -9.0 Min 51.3 42.3
9.5 Det 29.5 39.0 -7.2 Den 49.3 42.1
8.8 GSW 46.1 54.9 -6.5 Ind 44.4 37.9
7.9 Cle 47.7 55.6 -6.1 Dal 49.8 43.7
7.6 LAC 37.2 44.8 -5.6 Sac 46.8 41.2
6.6 Cha 29.5 36.1 -5.3 Mem 42.8 37.5
6.0 Hou 42.0 48.0 -4.9 Atl 38.8 33.9
3.5 Tor 30.0 33.5 -3.6 Uta 29.7 26.1
3.4 Chi 30.8 34.2 -3.4 Orl 42.8 39.4
2.3 OKC 55.9 58.2 -2.8 Bos 57.7 54.9
1.7 NYK 46.8 48.5 -1.2 Phx 47.4 46.2
0.1 LAL 40.8 40.9 -1.0 Mia 42.8 41.8
Re: 2024-25 team win projection contest
Posted: Wed Nov 06, 2024 12:59 pm
by Crow
The top, with a few exceptions, is not doing as well as expected and the bottom is doing better.
That could change or it doesnt. Is a tank coming? The first chapter or the first part of the first chapter is in the books but a long way to the ending.
Re: 2024-25 team win projection contest
Posted: Thu Nov 07, 2024 6:33 pm
by Mike G
Crow wrote: ↑Wed Nov 06, 2024 12:59 pm
The top, with a few exceptions, is not doing as well as expected and the bottom is doing better.
Yes. The "overs" on the left are mostly weak teams that look 'not that bad'. On the right are almost entirely teams that we guessed would do well. A lot of this is due to regression to the mean at b-r.com.
However, there are a few that have thus far flipped from above to below average, or vise versa. The regression actually makes their situation less extreme than their performance to date would project.
I think +1 point on net margin has been estimated in other work to be "worth" +2.7 wins, on average...
This would be a straight-line projection, while the pythagorean estimator is a sort of S-curve that approached but never reaches zero or 82 wins.
With the current 112.5 pts/G avg in the league, and an exponent of 16 in the Pythagorean formula, the 2.7 wins/Pt estimator compares like this:
Code: Select all
MOV 2.7 Pyth 1+
16 84.2 74.4 1.05
15 81.5 73.3 1.17
14 78.8 72.2 1.30
13 76.1 70.9 1.45
12 73.4 69.4 1.59
11 70.7 67.8 1.75
10 68.0 66.1 1.91
9 65.3 64.2 2.07
8 62.6 62.1 2.22
7 59.9 59.9 2.37
6 57.2 57.5 2.51
5 54.5 55.0 2.64
4 51.8 52.4 2.74
3 49.1 49.6 2.82
2 46.4 46.8 2.88
1 43.7 43.9 2.91
0 41.0 41.0
The final column is the Pyth wins value of an additional point in MOV. The 2.7 constant is reasonable up to about +/- 6 or 7 in MOV. Obviously it's not possible to win 84 games out of 82, no matter how good you are.
Re: 2024-25 team win projection contest
Posted: Thu Nov 07, 2024 11:10 pm
by Crow
That is a good curve.
Still suggests lots of regression in projections, especially in the middle I think. The differential exists immediately and is sizeable and compounds though it tapers.
Re: 2024-25 team win projection contest
Posted: Sat Nov 09, 2024 12:43 pm
by Mike G
The dread "41-41 for everyone" entry has sunk to the 2nd division.
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
bpmW 5.75 7.3 .39 4141 7.09 8.8
24pr 5.89 7.8 .25 dtka 7.23 8.6 .36
WShr 5.92 7.3 .38 DQin 7.48 9.1 .35
avgA 6.46 8.1 .36 medi 7.55 9.9 .31
eWin 6.66 8.5 .21 emin 7.70 9.5 .33
TmTj 6.76 8.8 .37 Crow 7.80 9.7 .33
perW 6.99 8.9 .17 24py 8.92 10.9 .25
13 games last night, and 7 were by 19+ points. All upper half entries were dinged (PER by .35) and all lower halfs were helped (medi by .27). The top-to-bottom spread shrank from 2.41 to 2.05
UPDATE Nov10
Code: Select all
w Atl w Bos w Brk w Cha w Chi w Cle w Det w Ind
44 pW 62 em 41 b-r 39 b-r 35 WS 58 b-r 38 b-r 48 pW
41 WS 61 DQ 34 WS 34 pW 35 bW 52 dt 33 eW 47 mo
41 dt 61 mo 31 eW 33 em 34 eW 50 mo 33 pW 47 Cr
41 bW 60 dt 30 pW 31 Cr 34 pW 50 Cr 32 TT 46 dt
39 DQ 59 Cr 29 bW 31 eW 34 b-r 49 em 32 bW 44 em
39 eW 57 TT 29 DQ 30 DQ 31 TT 48 DQ 32 WS 44 TT
38 mo 57 WS 26 dt 30 WS 30 dt 48 TT 28 em 43 bW
37 TT 56 bW 26 mo 29 dt 29 Cr 46 WS 28 dt 43 WS
35 Cr 54 b-r 24 TT 28 bW 28 DQ 46 bW 27 Cr 42 DQ
34 em 52 pW 23 em 26 TT 27 em 45 eW 26 DQ 41 eW
33 b-r 52 eW 18 Cr 24 mo 25 mo 44 pW 26 mo 36 b-r
w Mia w Mil w NYK w Orl w Phl w Tor w Was
47 em 50 dt 58 em 49 Cr 55 mo 34 b-r 34 eW
45 eW 49 DQ 53 Cr 48 mo 53 eW 32 pW 33 pW
43 bW 48 em 50 TT 46 dt 52 pW 32 WS 32 b-r
43 dt 48 Cr 50 dt 44 WS 51 TT 32 em 28 WS
43 WS 45 pW 48 mo 43 DQ 47 DQ 32 bW 28 bW
42 mo 45 TT 48 b-r 43 TT 47 dt 30 eW 19 em
42 DQ 44 mo 44 WS 42 em 46 WS 30 dt 14 DQ
42 TT 44 bW 43 DQ 41 bW 46 Cr 29 mo 15 TT
41 pW 44 WS 43 bW 41 b-r 46 bW 29 DQ 18 mo
40 Cr 43 eW 41 eW 39 eW 43 em 29 TT 19 Cr
40 b-r 36 b-r 38 pW 34 pW 29 b-r 25 Cr 21 dt
w Dal w Den w GSW w Hou w LAC w LAL w Mem w Min
56 DQ 53 em 53 b-r 50 b-r 45 b-r 45 TT 49 Cr 57 em
53 pW 51 mo 52 Cr 47 eW 45 DQ 44 Cr 48 em 55 DQ
52 eW 51 dt 49 DQ 46 mo 44 WS 44 dt 45 DQ 53 eW
52 TT 51 eW 49 TT 45 bW 40 bW 43 DQ 43 b-r 52 mo
51 Cr 49 pW 48 mo 44 TT 40 dt 42 em 43 dt 51 Cr
50 bW 49 DQ 47 em 44 Cr 36 TT 40 b-r 43 eW 50 WS
47 WS 49 Cr 46 bW 44 WS 35 mo 39 mo 43 mo 50 bW
47 dt 48 b-r 45 dt 43 pW 35 pW 39 bW 42 pW 50 dt
47 mo 47 WS 44 WS 38 dt 33 Cr 38 WS 41 bW 48 TT
45 b-r 47 bW 41 eW 36 DQ 32 em 37 pW 41 TT 47 b-r
43 em 46 TT 40 pW 33 em 31 eW 36 eW 34 WS 46 pW
w NOP w OKC w Phx w Por w Sac w SAS w Uta
49 mo 62 mo 51 TT 31 eW 50 Cr 40 b-r 33 pW
49 em 61 DQ 50 mo 31 b-r 49 bW 38 em 32 eW
48 Cr 59 b-r 49 dt 28 pW 48 pW 37 pW 32 WS
45 DQ 59 TT 48 Cr 27 bW 47 eW 36 eW 31 mo
44 bW 58 em 48 pW 26 DQ 47 DQ 36 bW 29 DQ
44 pW 55 WS 47 bW 26 Cr 47 TT 34 WS 29 dt
44 TT 54 bW 47 WS 26 WS 47 dt 34 TT 29 bW
44 WS 54 Cr 47 b-r 24 TT 46 WS 31 DQ 28 em
44 eW 54 dt 46 em 24 mo 44 em 31 dt 27 TT
42 dt 52 pW 45 eW 23 em 42 mo 31 mo 26 Cr
26 b-r 49 eW 43 DQ 22 dt 42 b-r 29 Cr 25 b-r
b-r is the current projection at b-r.com, from which our errors are measured.
Re: 2024-25 team win projection contest
Posted: Mon Nov 11, 2024 6:37 am
by TeemoTeejay
Well my projection is cooked with the Chet injury
Re: 2024-25 team win projection contest
Posted: Mon Nov 11, 2024 4:35 pm
by Mike G
The co-leaders are more than 1.00 ahead of the other actual entries.
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
WShr 5.34 7.1 .41 DQin 6.79 8.6 .41
bpmW 5.41 7.0 .43 perW 6.89 8.8 .19
24pr 5.66 7.7 .29 emin 6.92 9.2 .37
avgA 6.00 7.7 .41 Crow 7.01 9.0 .40
TmTj 6.43 8.3 .42 4141 7.26 9.0
eWin 6.67 8.4 .24 medi 7.40 9.4 .36
dtka 6.72 8.0 .43 24py 8.34 10.5 .29