APBRmetrics

Posted: **Tue Jan 10, 2012 4:35 pm**

Mike G wrote:
Still, it's a bit dismaying how many regularly bad teams continue to be bad -- or even worse -- even those you'd think could hardly get any worse. Right now, Cha is headed for 13-53, Det the same, GS 19-47, NJ 12-54, Sac 17-49, Was 8-58 !

Competitive imbalance.

Posted: **Tue Jan 10, 2012 9:56 pm**

Just a day later, significant improvements all around:

Code: Select all

7.23   JohnH     8.02   KP
7.52   J.E.      8.18   MG
7.59   ezPM      8.22   Vegas
7.70   Crow      8.33   DSM
7.71   ez2       8.38   zRA
7.74   dis       8.90   11py

Season at 13.2%

Posted: **Wed Jan 11, 2012 4:37 pm**

Now the avg error is 7.26, or twice the minimum possible (3.65), given the spread of predictions.

Code: Select all

6.69   JohnH     7.47   Vegas
6.93   ez2       7.47   KP
7.01   JE        7.55   zRA
7.02   ezPM      7.59   MG
7.06   Crow      7.90   DSM
7.12   dis       7.92  '11py

Season is at .143, or 1/7

Posted: **Wed Jan 11, 2012 4:50 pm**

Seems like it's shrinking so fast.

Posted: **Thu Jan 12, 2012 9:10 pm**

Season at 15.5%

Code: Select all

6.42   JohnH    7.15   Vegas
6.65   JE       7.18   MG
6.80   ez2      7.31   KP
6.81   Crow     7.39   zRA
6.89   dis      7.70   DSM
6.90   ezPM     7.77  '11py

Posted: **Fri Jan 13, 2012 2:29 am**

Thanks for the updates.

Posted: **Sun Jan 15, 2012 4:38 pm**

ah, this is more like it -

Code: Select all

6.37   JohnH      7.11    ezPM
6.61    J.E.      7.18   Vegas
6.84   MikeG      7.58    zRA
6.90   Crow       7.61   KevinP
7.04    dis       7.70   11Pyth
7.06    ez2       7.75   DSMok1

The median hasn't changed, but the order has.
.182 of season

Posted: **Mon Jan 16, 2012 1:11 am**

At the quarterpole it might be nice to see more detail, i.e. the actual / pyth. projected win totals for the teams alongside the contestant projections and / or team by team variances between these two win figures for each contestant.

Posted: **Mon Jan 16, 2012 9:55 pm**

Pyth+ is team's current wins + (Pythagorean W%) *(remaining games)
For 17 of 30 teams, we have all guessed too high or too low, and this is shown. As games are won and lost, these lines affect the magnitude, but not the ranking, of the predictions.
For the 13 other teams, where pyth+ is within the range of predictions, every game won or lost changes the average errors' ranking in the bottom line.

Code: Select all

tm   pyth+  all?  DSM  ezPM  MG  ez2 Vegas  JH   JE  zRA  dis  Crow  KP
Atl    47    lo    15   10   16   11   13   14   14   12   13   11   16
Bos    27    hi     9   15   14   16   12   16   15   18   13   14    8
Cha    13           8   11    9   11    4    0    5   11    9    5   15
Chi    54    lo    13    9    4   11    7    6    6   12    6    8    8
Cle    33    lo    12   14   12   11   17   15   15    7   14   15   11

Dal    43           5    1    1    0    1    3    0    0    1    2    6
Den    43           9    8   12    7    5    0    1    6    9    6    8
Det    14    hi    13    7    9    9    8    8   13   11    9    9   12
GSW    23    hi     9    6    6    6    4    3    6    5   10    4    3
Hou    28    hi     4    4    5    3    4    5    7    2    3    7    5

Ind    44    lo    10   12    6   11    7    8   13   11    9    7   11
LAC    40           3    3    2    1    3    1    0    6    2    1    0
LAL    42           2    1    3    0    7    2    0    1    6    1    8
Mem    30    hi     7    5    7    6    9    7    7    8    7   10    3
Mia    46           0    8    1    5    5    6    3    2    7    4    2

Mil    26    hi     7    5    4    6    6    8    6    7    5    7    6
Min    32           7   10    0    9   10    3   10    8    5    8    1
NJN    15    hi    12    9    9   10   12   13    7   11    7    9   14
NOH    21           4    4    0    5    3    1    3    6    3    6   10
NYK    32    hi     3    1    6    4   10    3    6    6    6    4    4

Orl    44           4    1    4    1    6    4    4    4    3    8    4
Phi    57    lo    22   21   23   21   21   20   18   21   23   20   21
Phx    28           1    7    5    4    2    4    1    0    1    0    0
Por    39           2    4    1    3    1    1    1    2    2    0    1
Sac    14    hi    15   13   16   12    7   11    6   11   11    9   12

SAS    45    lo     2    6    2    7    3    3    1    8    7    5    6
Okl    45           4    1    6    2    2    0    2    4    1    0    5
Tor    20           5    1    1    2    4    1    3    3    1    1    8
Uta    35    lo     7   10    5    9    9    8   10    8    6    9    5
Was     9    hi    15    7   13   10   11   13   12   14   11   13   10

err               DSM  ezPM  MG  ez2 Vegas  JH   JE  zRA  dis  Crow  KP
avg     7.00      7.7  7.1  6.7  7.1  7.1  6.3  6.5  7.5  7.0  6.8  7.5

John H is right on regarding Cha, while the rest of us were 4 to 15 wins too high.

Posted: **Wed Jan 18, 2012 3:35 am**

Thanks for the extra detail.

Posted: **Wed Jan 18, 2012 1:48 pm**

thx for updates. i'm surprised that i'm actually doing pretty well.

Posted: **Wed Jan 18, 2012 1:49 pm**

So what you're telling us is that you're stoked? Are you from Northern California by any chance?

Posted: **Wed Jan 18, 2012 9:37 pm**

dudeimstoked wrote:thx for updates. i'm surprised that i'm actually doing pretty well.

Dude, you're right at the average.
But just like every year, it's amazing how little we can improve on just repeating the previous year's pythagorean W%. That 'prediction' would be next-to-last among our entries, and none of us are more than 1.3 wins closer to the mark than that.

While most of us have successfully predicted a few major advances and a few major declines, these are mostly offset by teams which should have gotten better but didn't; and others which should decline but just refuse to cooperate.

Predictions based strictly on a 'system' don't seem to fare any better than those based on gut feelings; but I suspect there's great advantage in having an inside track on developments like: who is showing up out of shape, who has been able to perfect some moves, etc.

In other words, there's no shame in losing to Hollinger here.

Update: errors up for almost everyone; now ranging from 6.33 to 8.11

Posted: **Thu Jan 19, 2012 2:31 pm**

After a night of most teams doing what they're supposed to do, everyone looks better again.

Code: Select all

avg err          avg err
5.80   JohnH      6.91    ezPM
6.39    J.E.      6.95    ez2
6.48   Crow       7.14   KevinP
6.52   MikeG      7.23    zRA
6.69   Vegas      7.41   11Pyth
6.81    dis       7.52   DSMok1

Posted: **Thu Jan 19, 2012 2:53 pm**

This is one of the reasons whenever someone criticizes JH, I take issue with it. Dude knows what he's doing when it comes to this kind of stuff. Whether it's pure smarts or having the resources or whatever. Respect.

APBRmetrics

2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions

Re: 2012 predictions