2024-25 team win projection contest

[knarsu3]

Yeah both were predictive versions of the metrics and used the previous 3 seasons + an aging curve and some other data. They weren't the prior seasons individual metrics which probably would not do as well.

Pretty sure Darko would be winning and LEBRON is about where Vegas is this year or a tad better

I think the LEBRON or EPM over-unders cooked, altho I guess it’s like a predictive version of both more than just the metrics themselves iirc right?

[Crow]

Thanks for the links.

Posts of these a day late initially doesn't bother me. They just weren't found. Could have looked harder perhaps. I think I looked beforehand but not after season start. Don't obviously know forgot to post from don't care or choose not to.

I'd say include them and see what they show.

[Mike G]

Pretty sure Darko would be winning and LEBRON is about where Vegas is this year or a tad better

Last year, of 18 entries, DARKO finished 12th and LEBRON 17th.
They were 2.85 and 3.18, respectively, from the lead at the end.

Why should we think they do much better this time around?

[Crow]

Their performance is pretty important to me. Do not really have a guess whether it will be a strong year or not for each. Probably more likely strong, but want to
see. Not that one more season settles it. But how do they look after say 5 included seasons? Worth knowing.

If I had found them before the deadline, they would have factored to a degree into my initial baselines along with others.

[TeemoTeejay]

this time around?

I checked it awhile ago for a LinkedIn post but it was awhile ago, Darko is on his Twitter and I’m pretty sure it was in the lead most of the year if not still so. My predictions were pretty in line with darkos except not being regressed to the mean whatsoever + the aging curve probably helping Darko more since going lm(metric - wins) probably didn’t make sense lol

LEBRON and EPM weren’t doing as well as that but still doing well, so and we’re beating out Vegas better on an over under basis

No clue how it’s looking now

[dtkavana]

Pretty sure Darko would be winning and LEBRON is about where Vegas is this year or a tad better

Last year, of 18 entries, DARKO finished 12th and LEBRON 17th.
They were 2.85 and 3.18, respectively, from the lead at the end.

Why should we think they do much better this time around?

Two years ago I won the contest, last year I finished 10th, and this year I am currently in the lead. What I am saying is that it is certainly possible that LeBron and Darko are doing much better this year. I would love to see them included.

[Mike G]

dtka and medi are Not experiencing their biggest errors of the season.
Everyone else is, and daily.

Code: Select all

.  avg err   rmse   r^2      .   avg err   rmse   r^2
dtka   7.10   9.1   .49      WShr   7.89    9.8   .40
TmTj   7.12   9.3   .48     vegas   7.92   10.0   .42
avgA   7.22   9.4   .45      emin   7.95   10.2   .41
medi   7.42  10.3   .41      24pr   8.07   10.5   .32
bpmW   7.51   9.4   .45      24py   8.53   11.5   .32
DQin   7.56   9.7   .45      eWin   8.80   11.0   .25
KPel   7.69   9.2   .57      bmgm   9.19   12.0   .32
eExp   7.83  10.0   .48      perW   9.22   11.3   .22
Crow   7.86   9.9   .44               

[TeemoTeejay]

Yeah both were predictive versions of the metrics and used the previous 3 seasons + an aging curve and some other data. They weren't the prior seasons individual metrics which probably would not do as well.

Pretty sure Darko would be winning and LEBRON is about where Vegas is this year or a tad better

I think the LEBRON or EPM over-unders cooked, altho I guess it’s like a predictive version of both more than just the metrics themselves iirc right?

So was it like, for the predictive versions

was it more so a time decay or less weight on early season type of deal or a regression using the year results seperately?

[Crow]

At minimum, LeBron and Darko can be scored and evaluated at the end of season.

[Crow]

Is the projection target simply current BRef projection or is there any de-regressing still going on?

If there is no de-regression, then it appears LeBron would currently be 4th at 7.39 avg error.

Baring a data entry on my part, it appears Darko would lead at 6.68 average error.

That would make the top 4 individual entries all include RAPM in some form / degree. Top 5 if you count the APBR average which includes sets with RAPM.

Every metric and entry will have error. Error caused by data noise & unforseen choices & events. Including RAPM of course but certainly not different / alone in that.

Different approaches.

Strong performance of APBR average lends some support to the concept / philosophy of using blends.

[Mike G]

easy come, easy go

Code: Select all

.  avg err   rmse   r^2       .  avg err   rmse   r^2
TmTj   7.09   9.2   .48      Crow   7.89    9.9   .44
dtka   7.12   9.1   .49      WShr   7.93    9.7   .40
avgA   7.23   9.4   .45      vegas  7.94   10.0   .42
medi   7.40  10.3   .41      emin   8.01   10.2   .41
bpmW   7.49   9.3   .45      24pr   8.12   10.4   .32
DQin   7.55   9.7   .44      24py   8.62   11.5   .32
KPel   7.67   9.1   .57      eWin   8.76   11.0   .24
eExp   7.82  10.0   .47      perW   9.07   11.2   .22
.                            bmgm   9.23   12.0   .32

Teemo now leads at e=.75 to 1.41, kavana from 1.42 to 2.09
KPel owns the higher exponents, while medi runs the low end.

[Mike G]

Is the projection target simply current BRef projection or is there any de-regressing still going on?

Strong performance of APBR average lends some support to the concept / philosophy of using blends.

1 - Yes, all season just b-r.com, but for some adjustment for their 80-game projections prior to that silly in-season 'tournament'.
2 - 'Using blends' of course defeats the purpose of trying one system against others. It's all good though; sport is fun.

[TeemoTeejay]

How would my results be if they were regressed to the to have the same standard deviation as dtka’s, just curious

[Mike G]

You'd be .10 ahead instead of just .01
If we'd kept your original slate of guesses, you'd be .17 off the lead.

And mediocre is stealthily working back into contention. Since end of Jan.:

Code: Select all

date   TmTj   medi   diff
3-21   7.06   7.35   .29
3-14   6.95   7.34   .39
3-07   6.76   7.21   .45
2-28   6.44   6.96   .52
2-21   6.32   6.83   .51
2-14   6.38   6.92   .54
2-07   6.19   6.84   .65
1-30   5.77   6.73   .96

UPDATE Mar. 23

Code: Select all

.  avg err   rmse   r^2       .  avg err   rmse   r^2
dtka   7.07   9.1   .48      Crow   7.90    9.9   .44
TmTj   7.12   9.2   .48      vegas  7.91   10.0   .41
avgA   7.27   9.3   .44      WShr   7.93    9.6   .40
medi   7.45  10.3   .40      emin   8.04   10.2   .40
bpmW   7.51   9.2   .45      24pr   8.10   10.3   .32
DQin   7.52   9.7   .44      24py   8.56   11.5   .32
KPel   7.75   9.1   .57      eWin   8.85   11.0   .24
eExp   7.79   9.9   .46      perW   9.09   11.1   .22
.                            bmgm   9.10   11.9   .32

TmTj held the lead continuously (other than a few days behind AvgA) for 2 months starting Jan. 15
Then dtka for 5 days, tmtj for 3, now dtka.
KPel has pulled away in the r^2 column and leads at e>2.15

[Mike G]

LA Clippers have won 8 of their last 9. With Kawhi this season they have 3.0 fewer TO/100 .
https://www.basketball-reference.com/te ... 25/on-off/
Player rates from the first 61 games, and the last 9:

Code: Select all

LAC   mpg   TS%  TRb%   Ast%   Stl%  Blk%   TO%   Usg%   ORt  DRt   BPM  O-D
.      33  .559   9.3   37.1   2.2   1.9   19.0   29.6   109  110   3.0   -1
Harden 38  .655   9.3   35.1   1.8   2.2   14.5   31.6   130  112   8.1   18

.      31  .634  21.9   12.3   1.2   3.7   11.9   18.9   126  108   2.6   18
Zubac  36  .630  20.8   12.4    .3   3.1    8.7   22.7   127  111   1.3   16
                                    
.       9  .550   9.7   14.8   2.1   1.6   11.6   28.3   106  112   -.1   -6
Kawhi  28  .617  10.9   16.7   3.2   1.1    8.0   27.2   120  107   4.9   13
                                    
LAC   mpg   TS%  TRb%   Ast%   Stl%  Blk%   TO%   Usg%   ORt  DRt   BPM  O-D
.      13  .505   4.4   18.1   1.4    .4   11.0   20.0    99  121  -3.9  -22
Bogie  27  .758  10.3   18.6    .8    .8   16.2   20.0   132  114   3.6   18
                                    
.      16  .560   9.1    8.9   1.8   2.4   11.6    8.5   115  111   -.3    4
Batum  24  .763   9.1    6.3   2.1   4.3   13.8   10.6   135  110   4.3   25
                                    
.      25  .620   5.3    6.5   1.0    .5    8.0   15.1   120  115  -1.7    5
Coffey 23  .573   4.5    4.9   1.4    .0    1.4   15.1   120  117  -3.1    3
                                    
LAC   mpg   TS%  TRb%   Ast%   Stl%  Blk%   TO%   Usg%   ORt  DRt   BPM  O-D
.      21  .509   8.4   17.6   3.7   1.4   15.0   13.1   106  108    .5   -2
Dunn   22  .501   6.1   14.5   3.3   2.7   12.5   11.0   106  111   -.5   -5
                                    
.      23  .597   8.0    5.2   2.3   1.3    9.7   16.6   115  111   -.6    4
Jones  20  .714   7.2    1.4    .9   3.1    6.2   15.9   133  113    .9   20
                                    
.      25  .631   5.9   11.1   1.9    .5    9.7   27.2   117  113   2.9    4
Powell  8  .510   4.2    7.5   2.2    .0    8.7   22.5   110  111  -6.2   -1

mpg are min/61 and min/9 but for Bogdan Bogdanovic; first line is just in LA.
The only 2 players (Bogie and Batum) with increased TO% also have the greatest (ORtg-DRtg) upswings. Stupendously better shooting.
Harden looks better in longer/easier minutes w Kawhi. Zu hasn't suffered too much with more minutes. No backup center?

This is looking like their playoff nine. If Powell gets close to previous form after a bunch of missed games, they look like a serious contender.