APBRmetrics

Posted: **Sun Dec 05, 2021 6:01 pm**

Cali having a good week march up rankings.

Lots of turns at top. Will that continue or will an order become more constant?

Posted: **Tue Dec 07, 2021 11:01 am**

Code: Select all

.   avg err   rmse          avg err   rmse
avgA   5.70   7.60     vegas   6.20   7.82
Crow   5.71   7.70      dtka   6.43   8.81
cali   5.76   7.74      21re   6.45   7.97
shad   5.81   7.67      WShr   6.52   8.44
lisp   5.91   8.05      eWin   6.67   8.77
trzu   5.91   7.87      bpmW   6.67   8.22
5.38   6.08   8.28      2021   6.80   9.02
emin   6.16   7.71      perW   6.94   8.94

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

update Dec 9 -- separation at the top; and a new entry "pewb" is amalgam of PER, eWins, WS, and BPM. Totally cheating, adjusted daily for optimal weight on each component.

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   5.51   7.69      21re   6.09   7.83
avgA   5.62   7.62      5.38   6.17   8.40
cali   5.79   7.80      WShr   6.30   8.33
shad   5.81   7.67      eWin   6.33   8.70
emin   5.86   7.68      bpmW   6.57   8.19
lisp   6.05   8.15      2021   6.67   8.84
vegas  6.07   7.74      perW   6.73   8.93
trzu   6.07   8.00      dtka   6.76   8.99
pewb   6.08   8.21

update Dec 10 -- the ghost of last year continues to haunt:

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   5.71   7.73     vegas   6.27   7.89
avgA   5.82   7.70      5.38   6.30   8.45
cali   5.91   7.87      eWin   6.46   8.76
shad   6.02   7.80      WShr   6.50   8.42
emin   6.07   7.79      2021   6.72   8.94
21re   6.18   7.86      bpmW   6.75   8.31
lisp   6.18   8.20      dtka   6.83   9.01
pewb   6.25   8.29      perW   6.84   8.98
trzu   6.26   8.10

Posted: **Mon Dec 13, 2021 6:24 pm**

been a while since i've updated...

https://docs.google.com/spreadsheets/d/ ... OH861z2U9/

of course some difference from mike g's tracking but it looks like the various live projection models are converging - brooklyn, boston, and cleveland seem to have the largest ranges currently, along with 538 continuing to be relatively pessimistic about gsw

charts: https://imgur.com/a/dVbCj8P

Posted: **Tue Dec 14, 2021 3:02 am**

I am second on mean average error to a party not directly entered in the local contest.

Posted: **Tue Dec 14, 2021 2:58 pm**

The rankings have stabilized of late.

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   5.65   7.80      pewb   6.25   8.42
cali   5.78   7.82      21re   6.27   7.96
avgA   5.87   7.76      WShr   6.34   8.49
shad   5.90   7.73      eWin   6.74   8.99
trzu   5.99   8.10      bpmW   6.75   8.50
lisp   6.05   8.07      dtka   6.75   9.00
emin   6.07   7.82      2021   6.82   8.87
vegas  6.10   7.81      perW   6.97   9.23
5.38   6.11   8.35

Last night, West teams went 5-1 against East teams; this has been a trend of late.
After winning more than half of interconference games most of the season, the East is now projecting to avg 40.8 wins vs 41.2 by West teams.
The Cavs are the best in the East -- by SRS (4.98), by projected wins (50.4), by chance of winning the conference (27.6%) -- and it's seen as 78% likely that the winner of the West will win the Finals.
About 95% likely the West produces Phx, Uta, or GS, and 76% chance one is the champ.
https://www.basketball-reference.com/fr ... _prob.html

update Dec 19: Not much change in the order, but closer at the top. Vegas drops to the 2nd tier, and '21-regressed refuses to go away. Continuing over- and under-performances (Cle! GS! Phx, Mem / NO! Hou! Phl, LAL, NYK, Atl ...) have made errors worse all around.

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   6.08   8.31      21re   6.55   8.19
cali   6.10   8.21      vegas  6.61   8.26
avgA   6.20   8.15      pewb   6.67   8.82
trzu   6.30   8.52      WShr   6.81   9.00
shad   6.31   8.14      dtka   6.82   9.23
lisp   6.40   8.44      eWin   6.99   9.21
emin   6.41   8.28      bpmW   7.13   8.87
5.38   6.47   8.67      2021   7.28   9.42
.                       perW   7.35   9.36

update Dec 21 -- The whole field is getting tighter, as the bottom end improves.

Code: Select all

.   avg err   rmse          avg err   rmse
cali   6.05   8.14      pewb   6.54   8.65
Crow   6.06   8.24     vegas   6.62   8.25
avgA   6.15   8.03      dtka   6.68   9.13
shad   6.29   8.10      eWin   6.75   9.02
emin   6.31   8.14      WShr   6.81   8.89
21re   6.31   8.04      bpmW   6.95   8.69
trzu   6.38   8.38      2021   7.07   9.37
lisp   6.40   8.45      perW   7.15   9.15
5.38   6.46   8.66

Update Dec 27 -- still not much change in the order:

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   6.41   8.49      pewb   6.82   8.96
cali   6.42   8.50      trzu   6.83   8.75
avgA   6.56   8.37     vegas   7.03   8.61
21re   6.64   8.25      eWin   7.12   9.35
shad   6.66   8.50      bpmW   7.18   9.00
emin   6.66   8.44      WShr   7.21   9.11
5.38   6.69   8.95      dtka   7.26   9.45
lisp   6.77   8.79      2021   7.31   9.50
.                       perW   7.46   9.52

update Dec 29 -- much improvement for all:

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   6.18   8.15     vegas   6.85   8.27
cali   6.20   8.09      eWin   6.95   9.06
emin   6.33   8.07      bpmW   6.96   8.68
avgA   6.39   8.02      2021   6.97   9.18
5.38   6.46   8.58      WShr   6.98   8.74
shad   6.47   8.14      dtka   7.11   9.13
21re   6.53   8.02      perW   7.24   9.25
trzu   6.63   8.45            
pewb   6.64   8.61            
lisp   6.64   8.48

Last night, 8 games were played; relative to the avg of our predictions (and our presumed desire to have smaller errors), every game went as well as possible. Teams that we (mostly) 'wish' would win or lose did so when possible.

Code: Select all

goodgood   goodbad
NOP>Cle    Mil>Orl
NYK>Min    Mia>Was
Phl>Tor    LAL>Hou
Den>GSW
Sac>OKC

Had any of the games on the left gone the other way, they'd be "bad-bad".
Under "goodbad", we see games where "it's a shame someone had to lose (or win)".
Overnight, eminence improved by .35 and gained a spot; eWins by only .17 lost a spot.

update Jan. 2

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   6.14   7.94      pewb   6.66   8.42
cali   6.27   7.92     vegas   6.89   8.03
emin   6.37   8.00      dtka   6.96   8.95
avgA   6.42   7.83      bpmW   6.99   8.54
5.38   6.47   8.40      eWin   7.00   8.85
21re   6.49   7.83      WShr   7.01   8.63
shad   6.50   7.91      2021   7.05   9.05
trzu   6.58   8.23      perW   7.30   9.13
lisp   6.62   8.25

Crow and cali have held the top 2 spots for a month now. This is about the biggest separation between 1 and 2 in that time.
I've stopped "de-regressing" the b-r.com projections; so these errors are relative to :
https://www.basketball-reference.com/fr ... _prob.html

Posted: **Mon Jan 03, 2022 1:03 am**

Thanks for the update.

Posted: **Sun Jan 09, 2022 2:56 pm**

A little shakeup:

Code: Select all

.   avg err   rmse          avg err   rmse
Crow   6.36   8.20      lisp   6.88   8.38
emin   6.38   8.16      pewb   6.90   8.70
cali   6.51   7.98      dtka   7.02   8.99
trzu   6.61   8.23     vegas   7.16   8.19
5.38   6.68   8.55      eWin   7.18   9.06
avgA   6.69   8.00      bpmW   7.20   8.78
shad   6.77   8.00      WShr   7.29   8.94
21re   6.84   8.27      2021   7.38   9.49
.                       perW   7.66   9.36

Update Jan. 13

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.54   8.36      pewb   7.02   8.86
Crow   6.54   8.43      21re   7.03   8.35
cali   6.59   8.18     vegas   7.08   8.28
5.38   6.64   8.64      dtka   7.17   9.16
avgA   6.66   8.19      eWin   7.32   9.27
emin   6.67   8.42      bpmW   7.36   8.94
shad   6.70   8.11      WShr   7.48   9.10
lisp   6.84   8.44      2021   7.72   9.56
.                       perW   7.89   9.54

Posted: **Thu Jan 13, 2022 10:31 pm**

seems like your best update of the season

Posted: **Fri Jan 14, 2022 2:54 pm**

yeah well you can pick your favorite exponent. One is a solid bet, and Two is also popular.
In the ultra-trivial arena, trzu leads in an unprecedented 3 distinct exponents:

Code: Select all

e=0.20           e=0.44           e=1.40   
trzu   4.64      5.38   5.42      cali   7.20
pewb   4.72      trzu   5.43      shad   7.24
5.38   4.84      pewb   5.48      avgA   7.25
                     
e=0.32           e=1.00           e=2.00   
pewb   5.107     trzu   6.55      shad   8.08
trzu   5.111     Crow   6.57      cali   8.11
5.38   5.130     emin   6.61      avgA   8.12
                     
e=0.35           e=1.15           e=3.00   
trzu   5.200     Crow   6.82      21re   9.30
pewb   5.201     trzu   6.83      avgA   9.57
5.38   5.202     cali   6.83      shad   9.60

That's a lot of different "leaders."

Update Jan. 20

Code: Select all

.   avg err   rmse          avg err   rmse
cali   6.42   7.99      pewb   6.84   8.66
trzu   6.48   8.20      21re   6.87   8.14
avgA   6.48   8.04     vegas   6.91   8.16
emin   6.50   8.39      dtka   6.98   9.02
shad   6.54   8.02      bpmW   7.10   8.75
Crow   6.54   8.41      eWin   7.26   9.09
5.38   6.68   8.48      WShr   7.33   8.88
lisp   6.76   8.37      2021   7.68   9.43
.                       perW   7.72   9.38

Update Jan. 24

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.12   7.86      pewb   6.64   8.40
cali   6.14   7.74      dtka   6.66   8.66
shad   6.17   7.68      21re   6.75   7.96
avgA   6.20   7.74      bpmW   6.85   8.50
Crow   6.27   8.14      eWin   7.04   8.89
emin   6.29   8.10      WShr   7.05   8.62
5.38   6.35   8.14      2021   7.41   9.20
lisp   6.43   8.05      perW   7.45   9.13
vegas  6.51   7.85

special update Jan. 26 -- b-r.com page has not updated (I wrote them about it) so I've entered SRS and current W-L, applying SRS to future G, to find no real change in the order.

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   5.98   7.66      pewb   6.52   8.26
cali   6.01   7.63      dtka   6.54   8.53
shad   6.04   7.51      21re   6.68   7.90
avgA   6.07   7.58      bpmW   6.72   8.31
Crow   6.11   8.00      WShr   6.88   8.47
emin   6.14   7.81      eWin   6.90   8.85
5.38   6.21   8.00      2021   7.25   8.97
lisp   6.32   7.88      perW   7.32   9.05
vegas  6.38   7.70

Overall errors are the best since mid Dec.

And then last night (1-27) and none of the 10 games turned out favorably for most of us. Everyone's error jumped by .32 to .57; almost no change in the order.

Code: Select all

bad/bad     good/bad
Cle>Mil      LAC>Orl
Cha>Ind      Atl>Sac
Mia>NYK      Den>Brk
Mem>SAS      Chi>Tor
Phx>Uta      Dal>Por

You could say the average entry was 5-15 (W-L) on the night, though you may have done better or even worse.

Update Feb 2 (first of many?)

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.47   8.20      dtka   6.97   9.03
cali   6.57   8.30      pewb   7.01   9.03
avgA   6.61   8.31      lisp   7.02   8.59
shad   6.62   8.23      bpmW   7.12   9.07
Crow   6.64   8.67      21re   7.12   8.49
5.38   6.78   8.56      eWin   7.56   9.46
emin   6.79   8.67      WShr   7.61   9.21
vegas  6.90   8.39      2021   7.92   9.66
.                       perW   8.05   9.72

trzu has consolidated the lead, from exponents 0.73 to 2.15. Below that it's Crow, and higher is shad.

Update Feb 3

Code: Select all

.   avg err   rmse          avg err   rmse
cali   6.60   8.28      pewb   7.01   8.97
trzu   6.60   8.21      21re   7.02   8.34
shad   6.66   8.21      bpmW   7.05   9.03
avgA   6.66   8.26      dtka   7.08   9.01
Crow   6.80   8.61      eWin   7.47   9.33
5.38   6.84   8.56      WShr   7.63   9.20
vegas  6.94   8.34      perW   7.96   9.60
emin   6.98   8.67      2021   8.01   9.65
lisp   6.99   8.57

Shakeup due to wins by Orl, Hou, Bos, OKC, and Utah. Losses by Dal, Den, Phl also in the mix.

Posted: **Sat Feb 12, 2022 4:59 pm**

Got any of them updates? I'd like one on a fresh piece of post.

Posted: **Sat Feb 12, 2022 10:02 pm**

The order hardly changes at all, though we all get worse by the day.

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.93   8.64     vegas   7.32   8.97
cali   6.98   8.75      pewb   7.53   9.68
shad   6.98   8.74      bpmW   7.71   9.76
Crow   7.03   8.99      21re   7.84   9.07
avgA   7.06   8.80      WShr   8.05   9.69
emin   7.07   9.07      eWin   8.18  10.02
5.38   7.11   8.96      2021   8.32  10.20
dtka   7.23   9.31      perW   8.61  10.27
lisp   7.27   9.04

On Dec. 9 the worst error was 6.76. This was with <10% regression to the mean.
With few exceptions, over-performing teams keep overing; some unders are just disintegrating.

Code: Select all

over   tm     under   tm
23.3  Cle     -13.0   Por
15.2  GSW     -10.7   Ind
14.0  Phx     -10.6   LAL
13.7  Mem     -10.1   Brk
9.4   Tor      -9.9   Atl
7.2   Mia      -9.2   Det
7.1   Min      -8.2   NOP
5.3   Cha      -7.0   NYK
4.8   Chi      -6.6   Sac
2.3   OKC      -5.1   Mil
1.8   Bos      -4.0   Hou
1.6   Was      -3.1   LAC
0.2   Dal      -2.9   Phl
.              -2.5   Orl
.              -1.2   Uta
.              -1.0   SAS
.              -0.8   Den

These are differences between projected wins and the average of our guesses.

Posted: **Mon Feb 14, 2022 4:01 am**

Thanks.

Cut my margin from the lead in half and moved up a spot. Not much but something.

Posted: **Fri Feb 25, 2022 12:35 pm**

At the allstar break:

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.76   8.32      lisp   7.20   8.76
cali   6.78   8.46      pewb   7.35   9.46
shad   6.83   8.46      21re   7.56   8.84
Crow   6.89   8.75      bpmW   7.60   9.53
emin   6.91   8.76      WShr   7.87   9.48
5.38   6.94   8.65      eWin   8.09   9.88
avgA   6.95   8.54      2021   8.10   9.85
dtka   7.10   9.03      perW   8.63  10.12
vegas  7.14   8.74

UPDATE Mar. 1

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.56   8.18      vegas  7.01   8.64
cali   6.58   8.34      lisp   7.07   8.66
shad   6.66   8.38      pewb   7.16   9.40
5.38   6.70   8.44      21re   7.58   8.72
emin   6.72   8.70      bpmW   7.59   9.51
avgA   6.79   8.44      WShr   7.65   9.31
Crow   6.88   8.65      2021   7.86   9.74
dtka   6.93   8.86      eWin   8.11   9.79
.                       perW   8.70  10.07

Not much change in the ordering, but a steady increase in the top-to-bottom difference. The best avg error is now 2.14 better than the worst. That differential has grown by 50% in the last 24 days. On Jan. 14, the diff. was just 1.20
The bottom has dropped out for some of us!

Posted: **Tue Mar 01, 2022 6:02 pm**

A couple had a strong week. Still enough weeks left to catch maybe, but it will be harder.

Posted: **Thu Mar 03, 2022 1:36 pm**

trzu pulls away, maybe

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.61   8.17      lisp   7.08   8.64
cali   6.72   8.40     vegas   7.10   8.69
shad   6.74   8.40      pewb   7.24   9.46
5.38   6.80   8.40      bpmW   7.63   9.56
avgA   6.87   8.47      21re   7.66   8.77
emin   6.88   8.78      WShr   7.77   9.37
dtka   6.96   8.80      2021   7.97   9.78
Crow   6.98   8.68      eWin   8.20   9.83
.                       perW   8.79  10.13

This is the largest separation between 1 and 2 that we've had since Jan. 8 (Crow > cali)

Update Mar. 4

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.46   8.07     vegas   6.98   8.61
shad   6.63   8.31      pewb   7.22   9.36
cali   6.64   8.31      21re   7.59   8.67
5.38   6.70   8.29      bpmW   7.60   9.49
avgA   6.84   8.38      WShr   7.70   9.26
emin   6.85   8.73      2021   7.89   9.69
dtka   6.87   8.70      eWin   8.17   9.75
Crow   6.92   8.59      perW   8.77  10.05
lisp   6.93   8.53

Update Mar. 13:

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.36   7.92      lisp   6.99   8.42
cali   6.64   8.16      pewb   7.24   9.23
shad   6.64   8.22      bpmW   7.43   9.38
5.38   6.69   8.24      21re   7.68   8.64
avgA   6.76   8.29      WShr   7.71   9.25
vegas  6.76   8.48      2021   7.95   9.78
dtka   6.82   8.60      eWin   8.16   9.63
emin   6.94   8.80      perW   8.65   9.93
Crow   6.94   8.59

Lead by trzu is largest since Nov. when 21re was dominating.

Update Mar. 19:

Code: Select all

.   avg err   rmse          avg err   rmse
trzu   6.44   8.02      Crow   7.03   8.71
shad   6.65   8.34      lisp   7.08   8.54
cali   6.68   8.29      pewb   7.38   9.42
vegas  6.78   8.64      bpmW   7.63   9.60
5.38   6.81   8.34      WShr   7.81   9.38
emin   6.85   8.86      21re   8.04   8.98
avgA   6.92   8.45      2021   8.08   9.92
dtka   6.97   8.69      eWin   8.45   9.89
.                       perW   8.91  10.19

Top-to-bottom difference is more than twice what it was in mid-Jan.

APBRmetrics

2021-22 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest

Re: 2022 team win prediction contest