Page 5 of 13
Re: 2022-23 team win projection contest
Posted: Sat Nov 26, 2022 11:50 am
by Mike G
There's a negative-0.41 correlation between the age of a team and how they're over-achieving relative to predictions.
Code: Select all
tm age over tm age over
LAC 29.7 -10.0 NOP 25.9 5.5
Mil 29.5 -0.5 Tor 25.7 -3.4
Brk 28.7 -5.3 Cle 25.6 8.6
LAL 28.4 -8.2 Min 25.5 -7.3
Phl 28.3 -5.8 Por 25.4 9.4
Dal 28.2 -0.5 Sac 25.4 8.3
Mia 28.1 -9.0 Atl 25.1 -2.6
Chi 28.0 2.5 NYK 25.1 -3.0
Phx 27.7 3.3 Mem 24.6 -3.5
GSW 27.5 -10.1 SAS 24.4 -7.3
Bos 27.4 3.7 Ind 24.2 14.0
Uta 27.1 12.6 Det 24.0 -1.1
Den 26.8 -6.9 Orl 23.1 2.4
Was 26.6 4.5 OKC 22.9 12.6
Cha 26.0 -6.3 Hou 22.4 4.0
Relative to the avg of predictions here (not including dummy entries), the 10 oldest teams (by age on the court) are projecting to under-achieve by 43 wins.
The 10 youngest teams appear headed to win 24 more games, and the middle 10 total 20 more than expected.
The contrast is even more dramatic if the old guys were getting their expected minutes: Lakers without LeBron, for example, register as younger than they'd be with him on the floor.
Re: 2022-23 team win projection contest
Posted: Sat Nov 26, 2022 5:54 pm
by Crow
(I don't see enough reason to edit past summaries. As soon as a new summary is posted, anything earlier fades away imo. In the end, it is about the last one.)
Re: 2022-23 team win projection contest
Posted: Sat Nov 26, 2022 9:54 pm
by Crow
But if the past is remembered and valued, trend lines or smoothed trend lines or trend lines with weighted decay on past results could be explored.
At one point DARKO had a team rating (I thought) but it isn't on site now.
There probably is better than BRef. Could use average of 4 current best entries or 4 best from last season or a blend of whatever you want. I know other things have been tried in past. Or keep it simple as is / good enough.
Re: 2022-23 team win projection contest
Posted: Sun Nov 27, 2022 4:43 pm
by Mike G
I am de-regressing the b-r.com forecast a bit more each day. Today, they say Bos is headed to 56 wins, but I'm comparing your guesses to 59 for Bos.
Worst team Spurs are exacerbated from 22.4 down to 19.1
Seeing how closely teams are trending toward
last year's record and
41-41, I've ditched the Pythagorean '22 and replaced it with ('22 + 41)/2 -- called
22Re.
Also added
41-41 for every team as a 'submission', and they are riding in the middle of the pack.
Code: Select all
. avg err rmse avg err rmse
LEBR 5.11 6.45 emin 6.82 8.11
22Re 5.45 7.24 vegas 6.86 8.75
ncs. 5.70 7.01 TmRk 6.88 8.59
DRKO 5.91 7.10 538E 6.93 9.04
vzro 5.94 7.20 538R 7.08 7.84
EBPI 5.97 6.99 AnBa 7.33 8.86
dtka 6.15 6.91 nuFi 7.46 8.98
trzu 6.27 7.47 EExp 7.52 9.00
Crow 6.40 7.75 MPra 7.86 9.47
4141 6.65 8.74 2022 8.03 9.62
BIPM 6.66 8.29
UPDATE Nov.28 -- big improvement for all but dummies
Code: Select all
. avg err rmse avg err rmse
LEBR 5.01 6.33 emin 6.54 7.86
22Re 5.45 7.27 TmRk 6.64 8.24
ncs. 5.53 6.77 vegas 6.64 8.39
EBPI 5.69 6.71 538E 6.77 8.84
vzro 5.72 6.96 4141 6.88 9.15
dtka 5.82 6.67 538R 6.88 7.65
DRKO 5.83 6.89 AnBa 7.01 8.43
trzu 6.11 7.14 nuFi 7.21 8.61
BIPM 6.24 7.84 EExp 7.33 8.64
Crow 6.24 7.39 2022 7.68 9.30
. MPra 7.72 9.10
Re: 2022-23 team win projection contest
Posted: Mon Nov 28, 2022 6:06 pm
by knarsu3
Either works for me. BIPM might make more sense
Mike G wrote: ↑Fri Nov 25, 2022 6:16 am
Great, that makes all kind of sense. I'll try to work it in for the next summary.
Meantime I suppose I can edit all the previous ones, replacing LEBR with BBIP -- or BIPM, if you prefer -- I just like 4-character column headers.
Re: 2022-23 team win projection contest
Posted: Mon Nov 28, 2022 7:15 pm
by bbstats
@mike - Can we add Pelton's in?
https://www.espn.com/nba/insider/inside ... s-slipping
Also if you could share / show how your target is generated, or just post to google sheets that would help analyze a bit.
Re: 2022-23 team win projection contest
Posted: Mon Nov 28, 2022 7:23 pm
by Mike G
Without objection, yes.
This link was offered up earlier, but:
1) I'm not gonna pay money for it; and
2) He didn't offer up his
player minutes, nor did he actively submit his entry. He not only is aware of our annual contest, he started this site.
If you (or anyone) posts his predictions here, I can add them to the field.
Re: 2022-23 team win projection contest
Posted: Mon Nov 28, 2022 7:40 pm
by knarsu3
@mike Can you add in R^2 as well?
Re: 2022-23 team win projection contest
Posted: Mon Nov 28, 2022 7:48 pm
by knarsu3
Here are Pelton's:
Code: Select all
Pelton Wins
PHO 49.0
DEN 47.9
NO 47.6
MEM 46.1
MIN 45.7
LAC 43.6
DAL 43.1
GSW 41.9
LAL 36.8
POR 36.6
SAC 36.5
UTA 34.9
SAS 31.2
OKC 26.9
HOU 26.7
BOS 53.9
MIL 49.5
PHI 48.3
TOR 47.4
ATL 46.2
MIA 45.9
BKN 45.6
CLE 42.9
NYK 41.5
CHA 40.8
WAS 40.0
CHI 38.1
IND 38.0
ORL 30.6
DET 25.7
Mike G wrote: ↑Mon Nov 28, 2022 7:23 pm
Without objection, yes.
This link was offered up earlier, but:
1) I'm not gonna pay money for it; and
2) He didn't offer up his
player minutes, nor did he actively submit his entry. He not only is aware of our annual contest, he started this site.
If you (or anyone) posts his predictions here, I can add them to the field.
Re: 2022-23 team win projection contest
Posted: Tue Nov 29, 2022 2:00 pm
by Mike G
The 'target' projections are from b-r.com, which has in the last couple of days toned down their regressing-to-.500.
I was de-regressing it to more closely resemble teams' actual performance to date; and now I am scaling back that effort, so that in a week or so it will be just what they say here:
https://www.basketball-reference.com/fr ... _prob.html
Adding Kevin Pelton and
squared mean root error column.
Code: Select all
. avg err rmse smre avg err rmse smre
KPel 4.85 6.01 4.10 TmRk 6.60 8.24 5.60
LEBR 4.92 6.24 4.01 vegas 6.60 8.35 5.64
ncs. 5.47 6.66 4.79 emin 6.66 7.95 5.83
22Re 5.56 7.32 4.51 538R 6.83 7.64 6.28
EBPI 5.78 6.85 5.13 538E 6.88 8.87 5.63
vzro 5.80 6.96 5.01 AnBa 6.91 8.43 6.01
dtka 5.82 6.67 5.23 4141 7.03 9.22 5.65
DRKO 5.91 6.94 5.32 EExp 7.21 8.63 6.38
trzu 6.10 7.16 5.57 nuFi 7.28 8.59 6.40
Crow 6.23 7.37 5.55 2022 7.62 9.32 6.44
BIPM 6.23 7.70 5.36 MPra 7.64 9.08 6.69
Kevin's predictions looked kind of wishy-washy at first glance, with all teams between 26 and 53 (Det/Bos).
But he has the highest guess on Ind (by far), OKC, and Was (tie); lowest on Phl, Dal, GSW, Mem -- all of which look very
good right now.
TeamRanking is so close to Vegas, it's hard to consider it a legitimately original entry. I already tossed out 'vegas2' for being a shadow entity.
4141 recently took a dive toward the bottom; but they are the leader at exponent<0.07 -- currently with direct hits on Ind, Was, Mia, and GSW; also best guess on Min, OKC, Sac, and Por.
Re: 2022-23 team win projection contest
Posted: Tue Nov 29, 2022 4:41 pm
by bbstats
Very helpful Mike! Could you add 4141 back in to this post?
Re: 2022-23 team win projection contest
Posted: Thu Dec 01, 2022 10:41 am
by Mike G
Code: Select all
. avg err rmse avg err rmse
KPel 4.59 5.76 TmRk 6.38 8.01
LEBR 4.65 5.89 vegas 6.39 8.14
ncs. 5.23 6.36 emin 6.39 7.70
22Re 5.24 7.02 4141 6.64 8.95
vzro 5.40 6.53 538E 6.68 8.51
EBPI 5.47 6.48 538R 6.73 7.50
dtka 5.53 6.34 AnBa 6.83 8.23
DRKO 5.65 6.69 nuFi 6.95 8.33
trzu 5.83 6.96 EExp 7.05 8.37
BIPM 5.94 7.44 MPra 7.53 8.92
Crow 5.97 7.13 2022 7.59 9.11
UPDATE Dec.3
Code: Select all
. avg err rmse avg err rmse
KPel 4.26 5.53 emin 6.01 7.50
LEBR 4.83 5.88 TmRk 6.20 7.80
22Re 4.85 6.55 vegas 6.21 7.97
ncs. 4.93 6.21 4141 6.23 8.32
EBPI 5.18 6.23 538E 6.31 8.19
vzro 5.20 6.34 538R 6.60 7.39
DRKO 5.26 6.47 AnBa 6.69 8.09
dtka 5.39 6.23 nuFi 6.88 8.14
trzu 5.58 6.86 EExp 7.02 8.25
Crow 5.81 6.98 MPra 7.45 8.79
BIPM 5.96 7.67 2022 7.71 9.01
Re: 2022-23 team win projection contest
Posted: Sat Dec 03, 2022 6:57 pm
by v-zero
I described the Pelton predictions after they were initially posted on page one as 'fascinatingly conservative'. It certainly seems that 'presciently conservative' would have been a better phrase.
Re: 2022-23 team win projection contest
Posted: Sun Dec 04, 2022 2:11 am
by tarrazu
v-zero wrote: ↑Sat Dec 03, 2022 6:57 pm
I described the Pelton predictions after they were initially posted on page one as 'fascinatingly conservative'. It certainly seems that 'presciently conservative' would have been a better phrase.
That's probably been a fair criticism most seasons and will also look better early in the season.
Re: 2022-23 team win projection contest
Posted: Tue Dec 06, 2022 2:36 pm
by Mike G
"Conservative" could mean all guesses within a narrow range; or close to last year's record; or close to what everyone else thinks. Kevin had at least as many
highest and lowest guesses as any of our self submitted entries.
Imagine where the ESPN Experts' consensus would be without KP's input.
Code: Select all
. avg err rmse avg err rmse
KPel 4.99 6.26 emin 6.56 8.10
22Re 5.24 7.29 TmRk 6.59 8.20
LEBR 5.34 6.45 vegas 6.60 8.33
ncs. 5.35 6.86 4141 6.72 8.90
EBPI 5.52 6.89 538E 6.75 8.76
DRKO 5.59 6.96 538R 6.90 7.88
vzro 5.62 7.05 AnBa 6.94 8.44
dtka 5.69 6.73 nuFi 7.08 8.56
avg. 5.76 6.94 EExp 7.30 8.63
trzu 5.94 7.33 MPra 7.51 9.11
BIPM 6.10 7.92 2022 8.03 9.57
Crow 6.25 7.39