Page 8 of 8

Re: 2020-21 team win projection contest

Posted: Tue Apr 27, 2021 3:49 pm
by Crow
Slid backwards last night.

Re: 2020-21 team win projection contest

Posted: Tue Apr 27, 2021 7:41 pm
by Mike G
As of today, the battleground teams in Crow v Vegas are (Crow gains by this outcome): Det (L), GSW (L), Mem (W), and Por (L)
Det won and Mem lost, so Crow was 0-2 vs Vegas; and no other games affected their separation.

Teams within 2 projected wins or losses of entering the battlefield: Bos, Brk, Chi, Cle, Ind, Phl, Den, OKC, Sac; and any could happen in a week or less.

These are from the liminal sheet, where you are .28 off the lead; by BB-Ref you're off just .20

Re: 2020-21 team win projection contest

Posted: Tue Apr 27, 2021 10:24 pm
by Crow
Thanks for the further analysis. I feel pretty good about the future of those battles.

Could go either way, though more likely I stay a bit short. The accuracy or tilt of the measuring stick may be the final deciding factor.

Re: 2020-21 team win projection contest

Posted: Fri Apr 30, 2021 10:35 am
by Mike G
Abrupt improvements all around -- APBR Avg shows .21 lower than last report.

Code: Select all

.   avg err   rmse       .  avg err   rmse
vegas  5.19   6.76      dtka   5.69   7.52
EExp   5.28   6.50      trzu   5.78   7.63
cali   5.33   6.86      KPel   5.81   7.77
shad   5.39   7.06      DSM1   5.95   7.77
bbst   5.43   7.05      lisp   6.15   8.14
Crow   5.47   7.35      538R   6.28   8.21
avgA   5.55   7.26      emin   6.68   8.27
TeRa   5.57   7.09      538E   7.60   9.41
.                       EBPI   8.84  10.89
The liminal sheet has a couple of missing values right now, so I fudged them a bit.

Re: 2020-21 team win projection contest

Posted: Fri Apr 30, 2021 11:54 am
by DSMok1
When I did my projections, I did not attempt to predict player movement (i.e. James Harden leaving Houston). Did some of the other entrants? My Houston projection was ridiculously off. I notice that Caliban's Houston projection was particularly low.

Re: 2020-21 team win projection contest

Posted: Fri Apr 30, 2021 3:43 pm
by Crow
I don't recall my thoughts on Rockets. Trade Harden and hurt some or resist and hurt some. Probably should have thought more and expected worse.

Re: 2020-21 team win projection contest

Posted: Sat May 01, 2021 9:58 am
by Mike G
Player movements, injuries, and now Covid/ protocols, are perhaps equal in importance to 'team talent' as a predictor. If you follow the rumors and the injury reports, and you can effectively quantify their impacts, you may have the best 'data'.

Teams that are doing much better than expected -- NYK, Phx, Utah, Cha -- have been relatively healthy and/or deep. You can project everyone will be unavailable for some 20% of games; but when a team stays abnormally healthy, they over-achieve.

The liminal sheet seems to be broke, so here are current errors based only on b-r.com

Code: Select all

.   avg err   rmse       .  avg err   rmse
vegas  5.29   6.98      dtka   5.70   7.74
EExp   5.33   6.74      trzu   5.78   7.82
cali   5.39   7.10      KPel   5.86   7.90
shad   5.43   7.29      DSM1   5.91   8.00
bbst   5.44   7.25      lisp   6.17   8.33
avgA   5.51   7.48      538R   6.32   8.38
Crow   5.55   7.60      emin   6.63   8.48
TeRa   5.58   7.33      538E   7.65   9.46
.                       EBPI   8.84  11.05
Everyone on the left is within a couple good days of vying for the lead.

UPDATE May 2 (still b-r.com projections)

Code: Select all

.   avg err   rmse       .  avg err   rmse
vegas  5.23   7.00      dtka   5.67   7.79
EExp   5.25   6.74      trzu   5.73   7.88
shad   5.31   7.32      DSM1   5.88   8.05
cali   5.36   7.17      KPel   5.93   7.95
bbst   5.40   7.28      lisp   6.09   8.39
avgA   5.49   7.53      538R   6.27   8.41
TeRa   5.49   7.35      emin   6.60   8.51
Crow   5.55   7.62      538E   7.71   9.54
.                       EBPI   8.92  11.11
On avg,, contestants got .03 better last night; but the range was from .11 improved (shad) to .07 worse (KPel and 538e)

May 3:

Code: Select all

.   avg err   rmse       .  avg err   rmse
vegas  5.25   7.01      trzu   5.78   7.90
EExp   5.27   6.76      DSM1   5.88   8.07
shad   5.35   7.34      KPel   5.92   7.96
cali   5.38   7.18      lisp   6.13   8.41
bbst   5.44   7.31      538R   6.26   8.44
avgA   5.51   7.54      emin   6.62   8.50
TeRa   5.51   7.37      538E   7.71   9.54
dtka   5.63   7.82      EBPI   8.82  11.09
Crow   5.63   7.66      .      
May 4 -- abrupt change at the top.

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.26   6.79      trzu   5.78   7.94
vegas  5.34   7.05      DSM1   5.84   8.10
shad   5.35   7.36      KPel   5.96   8.01
bbst   5.45   7.33      lisp   6.12   8.45
cali   5.45   7.25      538R   6.30   8.46
TeRa   5.51   7.39      emin   6.62   8.53
avgA   5.54   7.58      538E   7.80   9.59
Crow   5.66   7.69      EBPI   8.77  11.12
dtka   5.70   7.85      .      
May 6

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.31   6.86      trzu   5.87   8.03
vegas  5.38   7.13      DSM1   5.89   8.15
shad   5.40   7.43      KPel   6.11   8.09
cali   5.50   7.31      lisp   6.20   8.53
bbst   5.50   7.41      538R   6.36   8.53
TeRa   5.59   7.45      emin   6.77   8.60
avgA   5.60   7.65      538E   7.94   9.68
dtka   5.69   7.89      EBPI   8.77  11.16
Crow   5.71   7.75            
May 7:

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.23   6.83      Crow   5.80   7.79
vegas  5.31   7.10      DSM1   5.83   8.16
shad   5.35   7.41      trzu   5.84   8.01
cali   5.43   7.27      KPel   6.02   8.08
TeRa   5.50   7.43      lisp   6.24   8.50
bbst   5.51   7.40      538R   6.37   8.55
avgA   5.57   7.63      emin   6.68   8.55
dtka   5.66   7.86      538E   7.98   9.71
.                       EBPI   8.75  11.17
Update May 8
Yesterday, the espn experts had their best numbers of the season.
Today Crow, Pelton, and 538E have their worst.

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.28   6.90      Crow   5.90   7.89
vegas  5.32   7.17      DSM1   5.93   8.23
shad   5.44   7.49      trzu   5.94   8.09
cali   5.50   7.35      KPel   6.15   8.16
TeRa   5.56   7.51      lisp   6.32   8.57
bbst   5.61   7.49      538R   6.44   8.64
avgA   5.66   7.72      emin   6.76   8.62
dtka   5.77   7.93      538E   8.10   9.81
.                       EBPI   8.88  11.24
I believe caliban has the best chance of taking this, among our gang. Shadow is closer, but all his guesses are too close to EExp.
The Bucks are projecting to win 46.5 , but we know they won't; they will win 46 or 47 or another integral number of games. If we toggle their projection to 46.0, this moves cali .03 closer to the lead.
If we similarly downgrade Cle, Was, Por -- and expect a strong finish from Det, Dal, OKC, Sac, and SAS -- then cali wins this thing by a hair.

It's unlikely all 9 of these 'battleground state' teams will cooperate. But with tanking, you never know.
(No, I haven't checked the schedules to see if all these are really possible.)

May 9
Nobody cooperated

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.31   6.98      DSM1   6.00   8.29
vegas  5.41   7.24      trzu   6.05   8.16
shad   5.55   7.57      KPel   6.23   8.25
cali   5.64   7.44      lisp   6.40   8.65
TeRa   5.65   7.58      538R   6.57   8.74
bbst   5.72   7.58      emin   6.80   8.66
avgA   5.77   7.79      538E   8.21   9.92
dtka   5.89   8.00      EBPI   8.99  11.31
Crow   5.93   7.95            
A bad night for most, on avg .10 worse than yesterday. Caliban suffered by .14
Crow, bbst, KPel, both 538's with worst numbers of the season.

May 10
EExperts running away with it.

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.26   6.97      DSM1   5.95   8.23
vegas  5.40   7.24      trzu   6.05   8.14
shad   5.53   7.56      KPel   6.21   8.22
cali   5.60   7.40      lisp   6.34   8.61
TeRa   5.63   7.57      538R   6.51   8.74
bbst   5.67   7.57      emin   6.80   8.65
avgA   5.75   7.76      538E   8.19   9.96
dtka   5.85   7.97      EBPI   8.96  11.29
Crow   5.88   7.90            
May 12
Teams mostly have 3 games left.

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.35   6.92      DSM1   6.05   8.20
vegas  5.48   7.20      trzu   6.14   8.06
cali   5.58   7.37      KPel   6.24   8.19
shad   5.63   7.52      lisp   6.29   8.51
TeRa   5.65   7.50      538R   6.46   8.71
bbst   5.73   7.52      emin   6.79   8.59
Crow   5.83   7.82      538E   8.12   9.98
avgA   5.85   7.70      EBPI   8.93  11.30
dtka   5.89   7.91            
May 13

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.39   6.94      DSM1   6.12   8.22
vegas  5.47   7.21      trzu   6.14   8.08
cali   5.64   7.40      KPel   6.29   8.22
shad   5.65   7.54      lisp   6.30   8.53
TeRa   5.66   7.51      538R   6.53   8.73
bbst   5.73   7.53      emin   6.78   8.59
avgA   5.87   7.72      538E   8.15  10.03
Crow   5.90   7.86      EBPI   9.01  11.37
dtka   5.94   7.94            
538E making a late push for the cellar! Tanking?

May 14

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.41   7.01      DSM1   6.13   8.28
vegas  5.48   7.27      trzu   6.16   8.15
shad   5.66   7.60      lisp   6.34   8.60
TeRa   5.68   7.58      KPel   6.35   8.31
cali   5.73   7.48      538R   6.59   8.81
bbst   5.75   7.60      emin   6.82   8.64
avgA   5.88   7.79      538E   8.21  10.10
Crow   5.90   7.89      EBPI   9.00  11.42
dtka   5.98   8.00            
The top 2 have really separated from the rest of the field.
Worst numbers of the year for cali, trzu, KPel, and the 538 bro's.
From May 3 to May 12, the avgA went from .25 off the lead to .50 out.

May 16
One day left in the regular season.

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.41   7.04      DSM1   6.16   8.32
vegas  5.53   7.30      trzu   6.19   8.18
shad   5.69   7.63      lisp   6.43   8.64
TeRa   5.70   7.61      KPel   6.44   8.38
bbst   5.80   7.63      538R   6.67   8.82
cali   5.82   7.57      emin   6.84   8.65
avgA   5.91   7.83      538E   8.31  10.16
Crow   5.92   7.87      EBPI   9.11  11.43
dtka   6.03   8.03            
Worst average numbers of the season. Several are in free fall.

Re: 2020-21 team win projection contest

Posted: Mon May 17, 2021 6:34 am
by Mike G
Final standings.

Code: Select all

.   avg err   rmse       .  avg err   rmse
EExp   5.47   7.13      dtka   6.13   8.12
vegas  5.55   7.37      DSM1   6.17   8.38
shad   5.73   7.72      trzu   6.23   8.24
TeRa   5.74   7.68      KPel   6.40   8.44
bbst   5.82   7.70      lisp   6.50   8.69
cali   5.88   7.68      538R   6.60   8.89
Crow   5.93   7.97      emin   6.73   8.67
avgA   5.95   7.90      538E   8.33  10.24
.                       EBPI   9.13  11.52
EExp won at exponent from 0.52 to 5.2. Above that it was cali; from .28 to .51 TeRa, and lower was 538R.
(Very low exponent favors those with integer predictions; very high is won by whomever had the least worst guess.)

Every entry had at least one best/worst guess at a team. Including ties:

Code: Select all

.         best      /   worst
EExp   Cle Hou Por  / 
vegas  Chi Cle Phl Dal OKC / Ind
shad   Cle LAC OKC  / 
TeRa   Cle Dal GSW LAC / Mil
bbst   Det Dal      / 
cali  Chi Mil OKC Uta / Det Ind Phx
Crow   Brk Det Dal Den Min /
.               Chi Mia LAL NOP Por
dtka   Mia SAS      /  Cle Dal Den
DSM1   Bos Cle Orl Was Uta / Phx
trzu                /  Ind
KPel  Bos Cha Chi Mia LAL SAS /
.           Ind Phl Was Den LAC Min
lisp   Det          /  GSW Mem
538R   Cle Ind Mil Phl Dal Sac /
.                           Was Hou 
emin   Atl Orl Was Dal /
.                   Cha NYK Mem SAS
538E   NYK / 
Atl Bos Brk Orl Phl Tor Dal Den GSW OKC Sac Uta
These are best and worst without consideration of the doormat EBPI entry.
Success is better correlated with avoiding bad guesses, than with getting a lot of them right.

Re: 2020-21 team win projection contest

Posted: Mon May 17, 2021 7:37 pm
by Crow
Congrats to Shadow for the MAE contest win.


And Caliban, if using RMSE.