2015-16 Team win projections

[Yooper]

Haven't basically posted here since I last tried this in 2013-14. See if I can do better.

Miami 53 Golden State 60
Cleveland 51 Memphis 55
Chicago 50 Oklahoma 55
Atlanta 50 San Antonio 53
Washington 44 Houston 52
Charlotte 44 LAC 52
Boston 43 New Orleans 44
Toronto 42 Dallas 38
Indiana 37 Utah 38
Milwaukee 36 Sacramento 38
Orlando 32 Minnesota 34
Detroit 30 Portland 31
New York 30 Phoenix 30
Brooklyn 28 Denver 28
Philadelphia 26 LAL 26

East 596
West 634

Sorry about the formatting. Let me know if it needs fixed.

[sndesai1]

sorry to those of you without a widescreen monitor...we have a lot of entries this year


google doc: https://docs.google.com/spreadsheets/d/ ... edit?pli=1

rsmth and yooper, i'll add yours in later as well as the espn fall forecast. if anybody knows of any other predictions, feel free to post a link here and i'll add them in too

[ampersand5]

I came across an error last night and finally finished fixing it. Unfortunately, I came across another error in the process of fixing the first error- oh well, here is the best I can do for today:

Updated projections
----

Code: Select all

+----------------+-------------+
| Daniel Frank's | Projections |
+----------------+-------------+
| GSW            |          63 |
| Cle            |          60 |
| SAS            |          58 |
| OKC            |          56 |
| LAC            |          54 |
| Hou            |          53 |
| Mem            |          50 |
| Bos            |          48 |
| Atl            |          48 |
| Chi            |          47 |
| NOP            |          46 |
| Tor            |          46 |
| Was            |          42 |
| Uta            |          41 |
| Ind            |          40 |
| Mia            |          40 |
| Dal            |          39 |
| Cha            |          38 |
| PHX            |          37 |
| SAC            |          37 |
| Det            |          37 |
| Mil            |          36 |
| POR            |          34 |
| Orl            |          31 |
| Den            |          27 |
| Min            |          27 |
| Nyk            |          26 |
| BKN            |          24 |
| LAL            |          23 |
| Phl            |          22 |
+----------------+-------------+

[ampersand5]

I just saw that Evan Zamir posted his projections on Twitter, so you should probably add those too:

Code: Select all

+------+------------+------+
|   Evan Zamir      | Wins |
+------+------------+------+
| GSW  | W          |   61 |
| HOU  | W          |   57 |
| SAS  | W          |   55 |
| CLE  | E          |   54 |
| ATL  | E          |   53 |
| OKC  | W          |   53 |
| CHI  | E          |   51 |
| LAC  | W          |   51 |
| MEM  | W          |   49 |
| MIL  | E          |   47 |
| WAS  | E          |   47 |
| NOP  | W          |   47 |
| BOS  | E          |   45 |
| TOR  | E          |   44 |
| IND  | E          |   43 |
| DAL  | W          |   43 |
| UTA  | W          |   43 |
| MIA  | E          |   41 |
| POR  | W          |   37 |
| DEN  | W          |   35 |
| CHA  | E          |   34 |
| PHO  | W          |   34 |
| BKN  | E          |   31 |
| LAL  | W          |   29 |
| SAC  | W          |   29 |
| ORL  | E          |   27 |
| DET  | E          |   27 |
| NYK  | E          |   24 |
| MIN  | W          |   22 |
| PHI  | E          |   16 |
+------+------------+------+

[Statman]

Sorry, bunch of stuff came up - & since I don't get paid for this, I'm still working on the projections. All the theory stuff is done - too much minutiae to finish tonight. I should have fully fleshed out stuff tomorrow - I'll post here & at my site a day late & a dollar short.

The cool thing about using Pelton's minutes projections is that I'll be able to rank every player by projected 2016 WAR & WAR/48. The not cool thing is how some players are named slightly different between my league files & Pelton's list (Pelton, NBA, D League, Preseason, NCAA). Cross checking is a pain & has taken more time than expected.

If you could post your individual WS predictions, it would be cool to compare our systems does better after the season (and anyone else who has individual projections)

I will post projected WAR for every guy Kevin has listed as playing at least 1 minute this season.

[Crow]

This is a friendly contest. With only 3 games played last night, I'd be willing to let Statman or others in if submitted before 7pm eastern tonight. But if multiple people object, I won't press for it.

[tarrazu]

This is a friendly contest. With only 3 games played last night, I'd be willing to let Statman or others in if submitted before 7pm eastern tonight. But if multiple people object, I won't press for it.

If you will have me late, with my wildly off-consensus CHI projection (considering they have a win in hand!):

Code: Select all

team	proj_wins
GSW	59
OKC	58
HOU	57
CLE	56
LAC	53
SAS	51
BOS	48
TOR	47
ATL	45
MIA	44
MEM	43
NOP	43
UTA	43
WAS	42
DET	41
IND	40
DAL	40
CHI	39
ORL	39
MIL	37
CHA	37
POR	36
SAC	33
MIN	30
LAL	30
PHX	29
NYK	29
BKN	27
DEN	27
PHI	27

[bbstats]

I think I still have a good chance of winning with a pool this size BECAUSE MY PICKS WERE CRAZY.

[bbstats]

Also - SUPER-love the chart. But can we remove 15 Pyth + Regression from the "Max" and "Min" calculation? It covers the plurality of it.

[Mike G]

Agree w bbstats, both on the great chart and the need to remove dummy entries from the high-low line.

On some teams, we are largely in agreement, while for other teams there's a very wide range of predictions.
Do we suppose, then, that some teams are more 'predictable' than others? And perhaps that those teams with least divergence in our guesses are more likely to end up closer to their average prediction?

Here are last year's team prediction averages; ranked by the standard deviation in the range of predictions; and the 'error' between that avg and their actual wins.

Code: Select all

tm     avg    SD    W   avEr
Atl   43.7   1.4   60   16.3
Tor   47.4   1.5   49    1.6
Hou   48.6   1.9   56    7.4
Phl   16.5   2.0   18    1.5
Okl   54.5   2.0   45    9.5
LAC   57.5   2.2   56    1.5
Por   48.2   2.3   51    2.8
Cle   58.9   2.4   53    5.9
NOP   38.5   2.4   45    6.5
SAS   56.7   2.5   55    1.7
Was   44.8   2.7   46    1.2
Sac   29.4   2.7   29     .4
GSW   54.0   2.7   67   13.0
Mil   26.5   2.9   41   14.5
Ind   37.9   2.9   38     .1
Phx   45.7   3.0   39    6.7
Uta   27.7   3.0   38   10.3
NYK   34.7   3.0   17   17.7
Mem   49.4   3.1   55    5.6
Dal   51.3   3.2   50    1.3
LAL   25.6   3.3   21    4.6
Orl   26.7   3.3   25    1.7
Brk   36.7   3.4   38    1.3
Cha   40.2   3.6   33    7.2
Bos   29.5   3.6   40   10.5
Det   34.2   3.8   32    2.2
Chi   49.9   3.8   50     .1
Mia   44.4   4.0   37    7.4
Min   33.1   4.0   16   17.1
Den   37.7   4.1   30    7.7

We all figured Atl for 41 to 46 wins. Not one person (of 15) went outside that range.
Then they won 60; so we were all 14 to 19 wins low.

If you don't see a pattern between range of predictions and avg error, it's because there is none. Correlation is .04

[sndesai1]

glad you guys liked the chart
i just updated it up there, completely removing that entry, adding in the newest entries, plus a few sites i don't think we've tracked before like numberFire and AccuScore
the google doc has also been updated with all 32(!) entries

i'll probably post the first tracking update at the beginning of next week. it'll be almost pointless, but still fun to see


in regards to the standard deviation of last year's entries, the only thing it seems to have any correlation with is how poorly a team actually did, i.e. the teams with the biggest std.dev ended up having the fewest wins.

[Statman]

My very late entry - I got sidetracked with lineup optimization stuff & just didn't get around to just plain finishing the work. Obviously this ignores any games already played - I will be posting ALL projected player WARs (using Kevin's projected minutes) at my site later today, I'm just too tired to format anything right now.

One thing I did was make sure I didn't have any teams with the same record - just because. I rounded up or down on tight middle bunching of teams (when summing all the player projections) to get them all at least that game separation. All predictions still fall within my standard deviations for each team.

Rank Team Wins
1 okc 69
2 hou 61
3 cle 59
4 lac 58
5 sas 56
6 mia 55
7 gsw 53
8 tor 49
9 atl 48
10 cha 47
11 ind 46
12 uta 45
13 mem 44
14 por 43
15 sac 42
16 bos 41
17 nop 40
18 mil 39
19 orl 38
20 chi 37
21 phx 36
22 det 33
23 dal 32
24 was 30
26 min 25
25 nyk 24
27 den 23
28 lal 22
29 phi 18
30 bkn 17

[Mike G]

Everyone but I seem to think the Lakers suck. They look to me like a nearly-average team in a tough conference.
Available boxscore stats seem to mostly concur:

Code: Select all

LA   Lakers       Min   mpg   eW   e484    PER   WS/48   BPM   WS
Jordan Clarkson  2496   30   5.4   1.02   16.9   .078    -.3   4.1
D'Angelo Russell 2184   27   4.7   1.05   10.    .04    -2.    1.8
Lou Williams     2016   25   5.7   1.36   19.9   .134     .9   5.6
Brandon Bass     1950   24   4.3   1.07   16.3   .120    -.5   4.9
Roy Hibbert      1950   24   4.8   1.20   15.4   .105   -1.0   4.3

Julius Randle    1700   21   3.2    .93   10.    .04    -2.    1.4
Kobe Bryant      1696   21   4.4   1.25   17.6   .053     .3   1.9
Nick Young       1420   17   2.5    .83   14.2   .058   -3.5   1.7
Tarik Black      1170   14   2.3    .92   16.3   .115    -.5   2.8
Anthony Brown     922   11    .5    .25   10.    .04    -2.     .8

Marcel. Huertas   780   10    .3    .20   10.    .04    -2.     .7
Ryan Kelly        616    8    .3    .17    8.6   .027   -2.8    .3
Robert Sacre      390    5    .3    .35   10.9   .035   -2.8    .3
Larry Nance       390    5    .3    .43   10.    .04    -2.     .3

.  totals       19680  240  39.1    .95   14.4   .075   -1.1  30.8

The minutes are K Pelton's suggestions, and mpg are over 82 games.
Rookies are all given PER = 10, WS/48 = .040, and BPM = -2
I gave them varying eWin rates depending on draft position.

Even with these harsh assessments, PER, WS, and BPM predict more than 30 wins.
If I toggle rookie WS/48 down to zero, they still get 26 WS. Our group avg is <25 wins predicted.

Randle* is considered a rookie (drafted #7); Russell is #2 from this year.
Kobe's eW/484 is just pulled from thin air. It's less than last year and more than the year before. His WS/48 is from b-r.com's Projection line.
I also used these WS Projections for Bass, Williams, Black, and Hibbert. They were actually better last year.

This is a pretty young team, projecting to 25.2 -- Does that count against them? or does it mean these values are understated?

*EDIT: Randle is projected at .067 ws/48 at b-r.com. This raises team WS to 32
If Russell is good for .100 ws/48, the team totals 34.5
And that's with a terrible final season from Kobe.

[Statman]

Everyone but I seem to think the Lakers suck. They look to me like a nearly-average team in a tough conference.
Available boxscore stats seem to mostly concur

Couple things - the way Kevin projected minutes, my work had the average team rating over 2.5% "better" than what will be league average. So on average, every team "looked" better than they actually will be when the real minutes are doled out.

The Lakers easily have the WORST "best" player of any team in the league (Lou Williams).

Kevin has over 20% of available team minutes going to slightly above replacement level to downright bad players (Young, Nance, Sacre, Kelly, Brown).

[Mike G]

... the way Kevin projected minutes, my work had the average team rating over 2.5% "better" than what will be league average. So on average, every team "looked" better than they actually will be when the real minutes are doled out..

That's right. On page 1 of this thread, I said team wins are 2-4 per team too many.

So you are guessing 22 wins for LAL, scaling down from 24 or 25 with Kevin's minutes? That's still a long way from 30-something that these other stats suggest.
They won 21 last year -- 23 PythW -- and you don't think they should be any better this year?
They've added Hibbert, Lou W, Bass; the #2 draft pick, and last year's #7.

They may not have a superstar, but they've got 9 players who are decent, or should be.
Is Nick Young a 'bad' player? He's good for some things.