APBRmetrics Forum Index APBRmetrics
The statistical revolution will not be televised.
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Preseason Predictions
Goto page Previous  1, 2
 
Post new topic   Reply to topic    APBRmetrics Forum Index -> General discussion
View previous topic :: View next topic  
Author Message
THWilson



Joined: 19 Jul 2005
Posts: 144
Location: phoenix

PostPosted: Mon Oct 15, 2007 2:11 pm    Post subject: Reply with quote

Tim Lehrbach wrote:

No matter when Miles returns or what he's got left, he has to be better than Martell Webster and Travis Outlaw have been for the past two seasons, right? And what of Webster and Outlaw, don't they get a "1.1 multiplier" too?


I don't think Darius Miles is better than Travis Outlaw. Check the numbers.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Mon Nov 19, 2007 12:39 pm    Post subject: Reply with quote

Mike G wrote:
...
Last year, the West won 68 games more than the East. Both DW and I have the East cutting that deficit by more than half; John sees it almost disappearing ...

Last year, the East was 192-260 (.425) vs the West. This year, they're 20-38 (.338) after an 0-5 night. At this rate, the East will wind up 155-295 -- 70 games below .500. On average, then, JohnH's guesses would average 4.7 wins too many, per Eastern team; about 2.3 wins too high for DW's and mine.

According to Sagarin, only Bos and Orl, from the East, are among the top 10 teams. Only Cle, Mil, and Atl have had a top-10-toughest schedule so far.
_________________
40% of all statistics are wrong.
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Fri Nov 30, 2007 6:47 am    Post subject: Reply with quote

Applying pythagorean expectations to the remainder of the season, teams are on these winning paces (first column; 2nd col is avg of 3 predictions):
Code:
Pyth   Avg.   WEST   JohnH  DW21    MikeG
53.4   61.2    Dal    58*   62.2    63.5
63.4   61.0    SA     59    65.2*   58.8
49.7   57.5    Hou    61    55.2*   56.3
55.8   56.5    Phx    54    56.9*   58.6
59.8   49.8    Uta    50    45.3    54.1*

53.4   46.7    Den    53*   44.4    42.6
51.6   40.5    LAL    43*   40.2    38.2
50.7   40.2    NO     39    40.3    41.4*
39.8   36.7    GS     36    34.4    39.6*
32.2   35.6    Mem    34*   35.2    37.5

17.2   32.3    Sea    35    34.0    27.9*
28.6   31.7    Sac    27*   36.0    32.0
30.1   29.1    LAC    20    33.7    33.6*
21.0   25.0    Min    22*   26.2    26.9
27.2   22.1    Por    26*   21.8    18.6

633.7     total      617     631*    630

The * indicate best guesser.

Pyth   Avg.   EAST  JohnH   DW21    MikeG
18.6   55.3    Chi    55    57.7    53.2*
35.8   54.9    Cle    54    57.1    53.5*
68.6   51.7    Bos    51    50.2    54.0*
55.9   48.1    Det    48    48.7*   47.6
61.9   45.3    Orl    49*   45.8    41.0

50.9   41.8    Tor    40    40.0    45.5*
30.0   38.8    Mia    43    37.6    35.9*
25.2   38.7    NJ     39    36.9*   40.1
39.2   37.5    Was    33    39.9    39.5*
24.2   36.0    NY     38    32.7*   37.3

35.6   35.8    Atl    42    33.4*   31.9
30.5   33.0    Cha    34    29.7*   35.3
35.0   30.7    Mil    35*   27.4    29.7
37.3   30.0    Ind    28    32.0*   30.1
30.6   25.5    Phi    21    29.7*   25.7

579.0      total      610    599*    600

The Pyth sum to only 40.4 wins per team. Perhaps an artifact of uneven schedules at this time.
John takes the West, DW the East. I'm 'most consistent'.
Cavs are 9-7, but have been outscored by 3 ppg.
No SOS, to date or in future, have been considered.
_________________
40% of all statistics are wrong.
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Wed Dec 12, 2007 8:31 am    Post subject: Reply with quote

Ranked by the # of wins that the 3 of us (Avg.) appear to 'off', by conference. The East has the 3 biggest 'gainers' on our prediction, as well as the 4 most disappointing so far.
Code:
WEST   Avg.  Pyth   gain      EAST   Avg.  Pyth   gain
LAL   40.5   52.1   11.6      Bos   51.7   72.4   20.7
NO    40.2   49.9    9.7      Orl   45.3   58.6   13.3
GS    36.7   46.3    9.6      Det   48.1   61.3   13.2
Por   22.1   29.0    6.9      Phi   25.5   35.3    9.8
Den   46.7   52.9    6.3      Ind   30.0   38.8    8.8

Uta   49.8   55.6    5.8      Tor   41.8   47.3    5.4
SA    61.0   65.2    4.2      Was   37.5   42.3    4.9
Phx   56.5   57.2    0.7      Atl   35.8   37.5    1.8
Sac   31.7   31.5   -0.2      Mil   30.7   29.0   -1.7
LAC   29.1   28.2   -0.9      Cha   33.0   25.5   -7.5

Mem   35.6   33.0   -2.6      NJ    38.7   26.6  -12.0
Min   25.0   18.3   -6.7      Mia   38.8   25.7  -13.2
Sea   32.3   22.1  -10.2      NY    36.0   20.2  -15.8
Dal   61.2   50.4  -10.8      Cle   54.9   29.2  -25.7
Hou   57.5   44.9  -12.6      Chi   55.3   26.5  -28.8
The biggest bust in the predictions has been DW's call for 58 wins from the Bulls. DW also has the 2 closest calls: SA (looking like 65) and Phx (57).
All 3 of us are averaging about a 10-win error thus far.
_________________
40% of all statistics are wrong.
Back to top
View user's profile Send private message Send e-mail
NickS



Joined: 30 Dec 2004
Posts: 287

PostPosted: Wed Dec 12, 2007 1:28 pm    Post subject: Reply with quote

Mike G, thanks for keeping this thread updated. Both of the last tables are fun, and I hope you update them again sometime in January.

Interesting, but not surprising, to see that the best guesser is usally the person on the edge of the range. It looks like there are only 8 of 30 teams where the Pyth estimate is within the range established by the 3 estimates.
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 374

PostPosted: Wed Dec 12, 2007 5:24 pm    Post subject: Reply with quote

I posted predictions but it was not system based and it was after the set of methods. I certainly did look at the other methods but I didn't heavily game to be on the edge though I might have reasoned that way to some extent on a few. I gave it brief thought, not exhaustive research & judgment. I didnt use fractions. It is not the same, not claiming it is, but I thought it might be useful to some extent to have such a non-method reference point.

How does it look so far for my predictions (considered in that context)?

My average win error is 8.6. In the west if I counted right I am closest on 5 and tied for closest on 3 more. In the east which I admit to knowing far less about I am closest on just 2.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Thu Dec 13, 2007 8:48 am    Post subject: Reply with quote

Mtn, thanks for reminding that you had submitted an entry. The others I had already tabulated and worked from there.
Here, and still a few days out of date, are the Gang of 4 predictors, showing how far off the mark we've been -- if the season goes on as it has -- which is a major caveat, since now Cle has Lebron and Varejao, etc.
Code:
Pyth+  EAST  JH   DW21   MikeG  Mtn    John  DW21  Mike   Mtn
72.4   Bos   51   50.2   54.0   51     -21   -22   -18   -21
58.6   Orl   49   45.8   41.0   40     -10   -13   -18   -19
61.3   Det   48   48.7   47.6   53     -13   -13   -14    -8
38.8   Ind   28   32.0   30.1   28     -11    -7    -9   -11
35.3   Phi   21   29.7   25.7   28     -14    -6   -10    -7

47.3   Tor   40   40.0   45.5   42      -7    -7    -2    -5
42.3   Was   33   39.9   39.5   37      -9    -2    -3    -5
37.5   Atl   42   33.4   31.9   35       4    -4    -6    -3
29.0   Mil   35   27.4   29.7   32       6    -2     1     3
25.5   Cha   34   29.7   35.3   30       8     4    10     4

26.6   NJ    39   36.9   40.1   42      12    10    13    15
25.7   Mia   43   37.6   35.9   40      17    12    10    14
20.2   NY    38   32.7   37.3   42      18    12    17    22
29.2   Cle   54   57.1   53.5   49      25    28    24    20
26.5   Chi   55   57.7   53.2   54      28    31    27    27

   EAST              average error:    13.7  11.6  12.0  12.4

                           
Pyth+ WEST   JH   DW21   MikeG  Mtn    John  DW21  Mike   Mtn
52.1   LAL   43   40.2   38.2   40      -9   -12   -14   -12
46.3   GS    36   34.4   39.6   39     -10   -12    -7    -7
49.9   NO    39   40.3   41.4   43     -11   -10    -8    -7
29.0   Por   26   21.8   18.6   26      -3    -7   -10    -3
55.6   Uta   50   45.3   54.1   50      -6   -10    -1    -6

52.9   Den   53   44.4   42.6   53       0    -9   -10     0
65.2   SAS   59   65.2   58.8   59      -6     0    -6    -6
57.2   Phx   54   56.9   58.6   57      -3     0     1     0
28.2   LAC   20   33.7   33.6   28      -8     5     5     0
31.5   Sac   27   36.0   32.0   35      -4     5     1     4

33.0   Mem   34   35.2   37.5   34       1     2     4     1
18.3   Min   22   26.2   26.9   23       4     8     9     5
22.1   Sea   35   34.0   27.9   29      13    12     6     7
50.4   Dal   58   62.2   63.5   58       8    12    13     8
44.9   Hou   61   55.2   56.3   53      16    10    11     8

   WEST                                 6.8   7.6   7.2   4.9
                                       - - - - - - - - - - - -

      NBA             average error:   10.3   9.6   9.6   8.6


I guess in the previous post I 'ranked' the teams relative to what we (averaged) thought, and in this one we are being ranked. Mtn has the lead, so he's the enemy.
I didn't see this improvement coming for Orlando. Dwight's off the chart, Nelson has jumped his game up. Lewis isn't the reason.
I'm also worst with Det, and several of their guys are doing great off the bench.
And for Atl, it seems Horford is having a bigger impact than all of us ('cept John?) expected.

Out west, I never saw this Bynum event (LAL); and Fish has been great, and Farmar. Por is much less dismal than I expected. In Den, KMart is making a difference, and AI is not dropping off.
I didn't think Minny would be this bad, and what's up in Dallas? I actually ignored indicators that they'd be even better than what I called (63.5).
_________________
40% of all statistics are wrong.


Last edited by Mike G on Tue Jan 01, 2008 7:14 am; edited 1 time in total
Back to top
View user's profile Send private message Send e-mail
davis21wylie2121



Joined: 13 Oct 2005
Posts: 463
Location: Atlanta, GA

PostPosted: Thu Dec 13, 2007 10:30 am    Post subject: Reply with quote

Thanks for putting this together, Mike. This is interesting: I forced the spread of pW% in the league to a weighted average of the previous 3 years' standard deviations in pW%, which in effect compresses the league. In the East, this turned out to be my biggest strength, in the West it's my biggest weakness, suggesting that the distribution of talent out West is much wider than it is in the East. Two direct improvements for next year would be: 1) Force the fit to the standard deviation of winning %, not pW% (since win% is what we're after, I can't figure out why I didn't do this anyway); 2) Maybe break up the deviations by conference (this seems arbitrary to me, though, since theoretically the distribution in the past really shouldn't predict the current year's beyond the individual players themselves, which are already accounted for). Anyway, thanks for the updates!
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 374

PostPosted: Thu Dec 13, 2007 12:57 pm    Post subject: Reply with quote

Thanks Mike for the kind inclusion of my list for tracking, despite its different basis & timing.

The method race will be the main story but a set drawing on those different methods and more might add something to the test.

I expected at least half of the teams with big variances from the average of the predictions now will move back a good ways toward it but will study the situation further and perhaps comment more later.
Back to top
View user's profile Send private message
John Hollinger



Joined: 14 Feb 2005
Posts: 96

PostPosted: Thu Dec 13, 2007 5:01 pm    Post subject: Reply with quote

Nice to see I'm winning the "average error" category. High numbers are good .... um, right?
Back to top
View user's profile Send private message AIM Address
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Tue Jan 01, 2008 7:11 am    Post subject: Reply with quote

Certainly. But I called for Portland to win 18 games, and I've nailed it.
_________________
40% of all statistics are wrong.
Back to top
View user's profile Send private message Send e-mail
Tim Lehrbach



Joined: 03 Dec 2006
Posts: 34

PostPosted: Tue Jan 01, 2008 7:32 pm    Post subject: Reply with quote

Tim Lehrbach wrote:
Unless something goes terribly wrong, there is little chance Martell Webster is third on this team in minutes played at the end of the year.


Whaddya know, Martell Webster is second on the team in minutes played, and the Blazers are 18-13.
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Wed Jan 09, 2008 3:45 am    Post subject: Reply with quote

Blazers are now the 'most surprising' team, relative to what our experts picked. The Bulls still lead as underachievers.
Code:
                 Predicted Wins      Difference (Absolute values)
Pyth+  tm  JohnH  DW21   MikeG  Mtn   JohnH   DW21   MikeG   Mtn
72.3   Bos   51   50.2   54.0   51     21.3   22.1   18.3   21.3
65.0   Det   48   48.7   47.6   53     17.0   16.3   17.4   12.0
46.5   Was   33   39.9   39.5   37     13.5    6.6    7.0    9.5
34.4   Phi   21   29.7   25.7   28     13.4    4.7    8.7    6.4
37.9   Ind   28   32.0   30.1   28      9.9    5.9    7.8    9.9

50.7   Orl   49   45.8   41.0   40      1.7    4.9    9.7   10.7
38.0   Atl   42   33.4   31.9   35      4.0    4.6    6.1    3.0
42.9   Tor   40   40.0   45.5   42      2.9    2.9    2.6     .9
27.6   Mil   35   27.4   29.7   32      7.4     .2    2.1    4.4
34.6   NJ    39   36.9   40.1   42      4.4    2.3    5.5    7.4

27.1   Cha   34   29.7   35.3   30      6.9    2.6    8.2    2.9
23.3   Mia   43   37.6   35.9   40     19.7   14.3   12.6   16.7
20.1   NY    38   32.7   37.3   42     17.9   12.6   17.2   21.9
35.9   Cle   54   57.1   53.5   49     18.1   21.2   17.6   13.1
32.1   Chi   55   57.7   53.2   54     22.9   25.6   21.1   21.9

39.2   east  40.7 39.9   40.0   40.2   12.1    9.8   10.8   10.8

                             
Pyth+  tm  JohnH  DW21   MikeG  Mtn   JohnH   DW21   MikeG   Mtn
46.4   Por   26   21.8   18.6   26     20.4   24.6   27.8   20.4
54.9   LAL   43   40.2   38.2   40     11.9   14.7   16.7   14.9
54.0   NO    39   40.3   41.4   43     15.0   13.7   12.6   11.0
45.8   GS    36   34.4   39.6   39      9.8   11.4    6.2    6.8
49.7   Den   53   44.4   42.6   53      3.3    5.3    7.1    3.3

56.5   Phx   54   56.9   58.6   57      2.5     .4    2.1     .5
31.7   Sac   27   36.0   32.0   35      4.7    4.3     .3    3.3
47.1   Uta   50   45.3   54.1   50      2.9    1.8    7.0    2.9
26.3   LAC   20   33.7   33.6   28      6.3    7.4    7.3    1.7
57.4   SAS   59   65.2   58.8   59      1.6    7.8    1.4    1.6

54.2   Dal   58   62.2   63.5   58      3.8    8.0    9.3    3.8
28.3   Mem   34   35.2   37.5   34      5.7    6.9    9.2    5.7
23.2   Sea   35   34.0   27.9   29     11.8   10.8    4.7    5.8
14.6   Min   22   26.2   26.9   23      7.4   11.6   12.3    8.4
42.8   Hou   61   55.2   56.3   53     18.2   12.4   13.5   10.2

42.2   west  41.1 42.1   42.0   41.8    8.3    9.4    9.2    6.7
                             
                                       10.2    9.6   10.0    8.7

DW leads in the east; Mtn rules the west.
_________________
40% of all statistics are wrong.
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 1712
Location: Delphi, Indiana

PostPosted: Tue Jan 15, 2008 12:14 pm    Post subject: Reply with quote

Update: I am now leading in both 'best' and 'worst' predictions: closest on 9, worst on 11.
Mtn and John gave the same prediction for 6 teams. Giving them each 1/2 credit for best/worst picks, the standings are:

Code:
progno: John  DW21  Mike   Mtn
Best -   5.5   7.0   9.0   8.5
Worst - 10.0   6.5  11.0   2.5
Avg err  9.8   9.5   9.5   8.3


These errors are relative to pythagorean extrapolations of remaining games, on top of current W-L.
Everyone's worst picks are underestimating Por and overestimating Chi. Plus JH's picking Mia to win 43.
_________________
40% of all statistics are wrong.
Back to top
View user's profile Send private message Send e-mail
Mountain



Joined: 13 Mar 2007
Posts: 374

PostPosted: Tue Jan 15, 2008 2:39 pm    Post subject: Reply with quote

If I had noticed the ties with Hollinger more I would have tried to avoid them. Split credit for ties makes sense.
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    APBRmetrics Forum Index -> General discussion All times are GMT - 5 Hours
Goto page Previous  1, 2
Page 2 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group