2019-20 team win projection contest

[Mike G]

A day makes a difference.

Code: Select all

.  avg err   rmse
lisp   5.75   7.9
bbst   5.76   8.0
gold   5.80   7.8
cali   5.84   7.8
avg.   5.91   8.1
Crow   5.93   8.0
RyRi   5.95   8.3
trzu   6.17   7.9
shad   6.24   8.7
sndi   6.30   8.6
emin   6.43   9.9
.538   6.47   9.1
vegas  6.67   8.4
dtka   6.85   8.8
ncs.   6.88   8.8
KPel   6.90   9.1
Rd11   7.99   9.8
Mike   8.74  10.9

NOP rises as Mem sinks

Code: Select all

LAL   62      Mil   67
LAC   54      Tor   58
Den   53      Bos   56
Hou   53      Mia   51
Uta   51      Phl   51
Dal   50      Ind   46
OKC   49      Brk   38
NOP   39      Orl   37

Por   38      Was   29
Mem   38      Chi   29
SAS   37      Det   27
Sac   35      Cha   27
Phx   35      Atl   26
Min   26      NYK   26
GSW   19      Cle   23

UPDATE Feb. 27
New leader

Code: Select all

.  avg err   rmse
gold   5.87   7.9
lisp   5.89   8.0
bbst   5.92   8.1
avg.   6.01   8.3
cali   6.05   7.9
Crow   6.05   8.1
RyRi   6.16   8.4
trzu   6.29   8.0
sndi   6.36   8.7
shad   6.38   8.8
emin   6.39  10.0
.538   6.62   9.3
vegas  6.81   8.6
KPel   6.91   9.2
ncs.   6.94   8.9
dtka   7.02   8.9
Rd11   8.07   9.9
Mike   8.92  11.0

Yesterday gold did not lead at exponent 1 or 2, but from 1.16 to 1.95

Code: Select all

exponents     lead
0.01 - 0.10   ncs.
0.11 - 0.38   emin
0.39 - 0.80   bbst
0.81 - 0.89   lisp
0.90 - 2.19   gold
2.20 - 2.75   cali
2.76 - 7.2    trzu
7.3 - 10.6    cali
10.7 - -      Mike

UPDATE Feb 28

Code: Select all

.  avg err   rmse
lisp   5.93   8.1
gold   5.94   8.0
bbst   5.95   8.2
avg.   6.07   8.3
Crow   6.09   8.2
cali   6.11   8.0
RyRi   6.21   8.5
trzu   6.35   8.1
sndi   6.43   8.8
shad   6.44   8.9
emin   6.45  10.1
.538   6.65   9.4
vegas  6.87   8.7
KPel   6.98   9.3
ncs.   7.01   9.0
dtka   7.06   9.0
Rd11   8.15   9.9
Mike   8.98  11.0

UPDATE Feb. 29 -- gold surges.

Code: Select all

.  avg err   rmse
gold   5.78   7.9
lisp   5.93   8.0
avg.   5.95   8.3
bbst   5.95   8.1
cali   5.99   7.9
Crow   6.00   8.1
RyRi   6.17   8.4
trzu   6.18   8.0
sndi   6.27   8.7
emin   6.31  10.0
shad   6.35   8.8
.538   6.65   9.3
vegas  6.80   8.6
ncs.   6.93   9.0
KPel   6.95   9.2
dtka   7.00   8.9
Rd11   8.04   9.8
Mike   8.94  10.9

[Mike G]

Most improve overnight but also fall further from the lead.

Code: Select all

.  avg err   rmse
gold   5.69   7.8
avg.   5.93   8.1
lisp   5.94   7.9
cali   5.96   7.8
bbst   5.96   8.1
Crow   5.96   7.9
RyRi   6.07   8.3
trzu   6.11   7.9
sndi   6.22   8.6
emin   6.28   9.9
shad   6.29   8.6
.538   6.66   9.2
vegas  6.74   8.4
ncs.   6.90   8.9
KPel   6.94   9.1
dtka   6.95   8.8
Rd11   8.08   9.7
Mike   8.93  10.8

If I remove Crow's predictions from the contest, avg. falls to 7th place (mae 6.08)
Removing my own numbers, avg. nearly takes the lead (5.76)

UPDATE March 2

Code: Select all

.  avg err   rmse
gold   5.78   7.9
Crow   5.98   8.1
avg.   6.05   8.3
lisp   6.06   8.0
bbst   6.07   8.2
cali   6.07   8.0
RyRi   6.19   8.4
trzu   6.21   8.0
sndi   6.29   8.7
emin   6.32  10.0
shad   6.42   8.8
.538   6.77   9.3
vegas  6.80   8.6
KPel   6.95   9.3
ncs.   6.99   9.0
dtka   7.06   8.9
Rd11   8.15   9.8
Mike   9.06  11.0

UPDATE Mar. 3
Everyone but Crow had an off night.

Code: Select all

.  avg err   rmse
Crow   5.81   8.0
gold   5.83   7.9
cali   6.09   8.0
avg.   6.10   8.3
lisp   6.17   8.1
bbst   6.18   8.2
RyRi   6.25   8.4
trzu   6.27   8.1
sndi   6.28   8.7
emin   6.41  10.0
shad   6.49   8.8
vegas  6.86   8.6
.538   6.87   9.3
KPel   7.02   9.2
ncs.   7.03   8.9
dtka   7.14   9.0
Rd11   8.13   9.8
Mike   9.09  11.0

Normally we all get better or worse, some more than others. Last night had a number of results that favored Crow vs most of us:
Wins by NYK, Mia, Ind
Loss by Atl, Cle, Hou
Crow gains .16 to .29 on everyone.

Code: Select all

e=.50   
gold   4.62
Crow   4.80
emin   4.83
avg.   4.93
cali   5.06
   
e=1.1   
Crow   6.02
gold   6.05
cali   6.28
avg.   6.32
lisp   6.36
   
e=1.85   
gold   7.61
cali   7.69
Crow   7.70
lisp   7.78
trzu   7.83
   
e=2.5   
cali   8.81
trzu   8.91
gold   8.93
lisp   9.00
bbst   9.16
   
e=10   
Mike   16.6
cali   16.8
trzu   16.8
Rd11   17.2
lisp   17.8

[Crow]

Woo.

[Mike G]

Dead heat.

Code: Select all

.  avg err   rmse
gold   5.85   7.8
Crow   5.85   8.0
cali   6.16   7.9
avg.   6.18   8.2
trzu   6.18   8.0
lisp   6.23   8.0
bbst   6.27   8.1
RyRi   6.28   8.4
sndi   6.31   8.6
emin   6.34   9.8
shad   6.45   8.7
vegas  6.79   8.5
.538   6.91   9.2
ncs.   7.11   8.9
KPel   7.14   9.2
dtka   7.16   8.9
Rd11   8.20   9.8
Mike   9.09  10.9

A .0003 lead for gold, which is nothing in the world of rounding.
Avg error of 6.69 is worst since Jan. 25. Separation from #2 to 3 is unprecedented.
Kings move into the race for #8

Code: Select all

LAL   62      Mil   68
LAC   56      Tor   56
Hou   53      Bos   56
Den   53      Mia   52
Uta   51      Phl   49
Dal   50      Ind   48
OKC   49      Orl   37
Mem   40      Brk   37

Por   38      Was   29
NOP   37      Chi   28
Sac   37      Cha   28
SAS   37      Det   27
Phx   32      Atl   26
Min   27      NYK   26
GSW   20      Cle   23

Chance of securing that last western slot:
Mem 51%
Por 15%
NOP 13%
Sac 11%
SAS 9%

[Crow]

MIA, IND and TOR doing very well are important to my competitive chances.

Other stuff with smaller differences too, of course.

[Mike G]

Code: Select all

.  avg err   rmse
Crow   5.88   8.1
gold   5.91   7.9
cali   6.20   8.0
avg.   6.21   8.3
lisp   6.24   8.1
trzu   6.24   8.0
bbst   6.27   8.2
RyRi   6.28   8.4
sndi   6.39   8.6
emin   6.41   9.9
shad   6.50   8.8
vegas  6.84   8.5
.538   6.93   9.3
dtka   7.17   8.9
ncs.   7.19   9.0
KPel   7.24   9.3
Rd11   8.25   9.8
Mike   9.14  11.0

Avg error of 6.74 is now worst since opening day of the b-r.com forecast, Nov. 14
Ten of y'all have at one time or another had a projected error better than the current leading number.

UPDATE Mar. 9
Crow pulls ahead of an ever worsening field.

Code: Select all

.  avg err   rmse
Crow   6.01   8.3
gold   6.21   8.1
cali   6.31   8.1
avg.   6.32   8.4
lisp   6.41   8.3
bbst   6.50   8.3
trzu   6.51   8.3
emin   6.58   9.9
sndi   6.62   8.7
RyRi   6.62   8.6
shad   6.65   8.9
.538   7.13   9.4
vegas  7.20   8.8
ncs.   7.26   9.0
KPel   7.26   9.3
dtka   7.35   9.0
Rd11   8.09   9.8
Mike   9.22  11.0

Worst absolute error of the season for bbst, dtka, 538, Vegas, and the whole field (6.90)
(I was worse yesterday.)

[Crow]

Good weekend. Tonight was probably alright.

Still could be volatile. Rest might have some impact. Tanking might in a few cases.

[Mike G]

This may or may not be the final regular-season "projection" from b-r.com
For some reason, their projections add up to just 80 games (W+L) for most teams, but 81 or 79 for a few.
Extrapolating all to 82 G, we get this:

Code: Select all

62.8   LAL      65.8   Mil
56.3   LAC      57.5   Tor
52.8   Den      54.9   Bos
51.4   Uta      50.6   Mia
51.0   Hou      49.3   Phl
50.1   OKC      48.6   Ind
49.1   Dal      38.4   Orl
39.2   Mem      38.4   Brk

38.6   NOP      29.8   Was
36.7   Por      28.6   Cha
36.3   Sac      27.8   Chi
36.1   SAS      26.5   NYK
33.4   Phx      25.9   Det
25.9   Min      24.9   Atl
19.8   GSW      23.6   Cle

And relative to these:

Code: Select all

.  avg err   rmse
Crow   5.84   8.1
gold   6.08   8.0
cali   6.11   8.0
avg.   6.14   8.3
lisp   6.24   8.1
bbst   6.28   8.2
RyRi   6.41   8.5
emin   6.46   9.9
sndi   6.47   8.6
trzu   6.50   8.2
shad   6.53   8.8
.538   6.99   9.3
vegas  7.03   8.7
ncs.   7.11   8.9
KPel   7.12   9.2
dtka   7.20   8.9
Rd11   7.98   9.7
Mike   9.13  10.9

This would be Crow's largest margin since Feb. 14
Crow leads at exponent up to 1.56, Gold from 1.57 to 1.96, and cali above 1.96


And IF there are no more regular season games, the 63 to 67 games' Win% can all be extrapolated to 82, with these results:

Code: Select all

63.8   LAL      66.9   Mil
56.4   LAC      58.9   Tor
54.2   Den      55.1   Bos
52.5   Uta      51.7   Mia
51.3   Hou      49.2   Ind
51.3   OKC      49.2   Phl
49.0   Dal      38.4   Brk
40.4   Mem      37.8   Orl

36.0   Por      30.8   Was
35.9   NOP      29.0   Cha
35.9   Sac      27.8   Chi
35.1   SAS      26.1   NYK
32.8   Phx      24.8   Det
24.3   Min      24.5   Atl
18.9   GSW      24.0   Cle

There's actually a clearer distinction between who is in or out of the playoffs in the West.
Not sure of tiebreaking procedure for the 5 and 6 seeds in either conference.

IF these are indeed the final standings, then --

Code: Select all

.   avg err   rmse
Crow   6.15   8.64
emin   6.75  10.32
gold   6.75   8.64
cali   6.78   8.63
avg.   6.80   8.90
lisp   6.87   8.78
bbst   6.89   8.79
RyRi   7.01   9.09
trzu   7.06   8.80
sndi   7.10   9.21
shad   7.12   9.43
vegas  7.45   9.26
.538   7.62   9.92
KPel   7.63   9.81
ncs.   7.73   9.50
dtka   7.75   9.59
Rd11   8.61  10.29
Mike   9.84  11.61

Crow in a rout: leads at exponent below 1.998, caliban above that.

[Mike G]

Well, nobody reminded me to get back on this project, and I had pretty much written off the rest of the season. But here we are again, the bubble hasn't burst as yet, and the re-season is winding down.
Basketball-Reference.com has gone commercial, so many search features are unavailable now. Yet they still have the Playoff Probabilities Report.

Lottery teams "project" to the (64 to 67 G) records they had as of mid March. "Bubble" teams project to anywhere between 71 and 75 games. I've extrapolated their projected (or current) Win% onto 82 games for this exercise.
While it looks as if the sum of all wins and losses should end up at 1059 of each, the avg Win% is a bit under .500, and the 82G projection averages 40.58 wins.

In our contest, everyone has improved with the bubble games, and Crow's lead is cut in half.

Code: Select all

.   avg err   rmse
Crow   5.99   8.10
cali   6.31   8.03
lisp   6.36   8.35
bbst   6.42   8.40
gold   6.45   8.29
avg.   6.47   8.45
RyRi   6.57   8.68
emin   6.66  10.07
shad   6.78   8.94
sndi   6.85   8.95
trzu   6.88   8.39
KPel   6.93   9.27
vegas  7.06   8.72
.538   7.09   9.59
ncs.   7.19   9.07
dtka   7.51   9.29
Rd11   8.05   9.83
Mike   9.15  11.02

https://www.basketball-reference.com/fr ... _prob.html
I've not looked into what 'play-in' means, or whether b-r.com is figuring that into their projections. Assume they will be counted as 'regular season' games rather than playoff??

Right now 8 east teams and 7 west teams have a 100% chance of making the playoffs. Then we have:
.546 Por
.179 Mem
.153 Phx
.122 SA
Not sure why Grizz are given the edge over the Suns. Same 33-39 record, Phx with 1.4 better SRS

[Crow]

I considered it a win at the interruption.

The score after the bubble is a bit affected by all the resting, but let's see what it shows.

Thanks for the reporting.

[Mike G]

b-r.com has done a strange thing and projected an additional 7 game results onto all bubble teams but the Lakers and the 4 also-rans. It shouldn't matter, since I'm already scaling everyone up to 82.

Code: Select all

.   avg err   rmse
Crow   6.07   8.11
cali   6.27   8.01
lisp   6.29   8.38
gold   6.37   8.28
avg.   6.37   8.39
bbst   6.44   8.36
emin   6.63  10.04
RyRi   6.64   8.71
sndi   6.75   8.87
trzu   6.85   8.42
shad   6.85   8.92
KPel   6.87   9.16
ncs.   7.01   8.91
.538   7.04   9.56
vegas  7.19   8.76
dtka   7.43   9.30
Rd11   7.78   9.59
Mike   9.12  10.91

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

Errors are relative to these projections :

Code: Select all

LAL   60.1      Mil   62.9
LAC   54.8      Tor   60.1
Den   50.7      Bos   54.2
Hou   49.4      Ind   49.5
Uta   49.3      Mia   49.0
OKC   49.1      Phl   47.0
Dal   46.2      Brk   38.6
Por   38.8      Orl   35.5
Mem   38.2      Cha   29.0
Phx   38.2      Was   28.5
SAS   37.0      Chi   27.8
Sac   35.3      NYK   26.1
NOP   34.2      Det   24.8
Min   24.3      Atl   24.5
GSW   18.9      Cle   24.0

This is still a bit screwy, since 30 teams now 'average' just 40.2 extrapolated Wins.