2012 predictions

[Mike G]

Season 92% over.

Code: Select all

current W + pyth% remaining           pythagorean% X 66 games
3.38   Crow     4.04   MikeG        3.05   Crow     3.90   DSMok1
3.70   JohnH    4.14   DSMok1       3.35   J.E.     3.90   MikeG
3.85   J.E.     4.15   ezPM         3.36   JohnH    4.12   KevinP
3.88   Vegas    4.60   zRA          3.48   dis      4.27   ezPM
3.97   dis      4.72   KevinP       3.83   Vegas    4.33   zRA
4.01   ez2      5.08   11Pyth       3.87   ez2      5.23   11Pyth

[Crow]

Thanks for the updates thru the season.

[Bobbofitos]

nice work crow

[Mike G]

Nice is an understatement. I don't think we've seen such a runaway in several years of tracking predictions.
Of course, it could be luck on top of insight. Some teams may be "tanking" of late. But still, the objective is to predict final records, and that is immutable.

[Crow]

Thanks Rob, though the finish line is still a week away.

It could be luck on top of insight but my method has done about as well on average as any over my last 4-5 entries.

I spent little time on the predictions. Just computed a custom-weighted average of a handful of other predictions (that have good track records and that I respect) and then tweaked the win projections up or down (mostly modestly) based on my general NBA knowledge.

[RoyceWebb]

Currently the ESPN Winter Forecast is at 3.63 relative to what the Hollinger Playoff Odds project as each team's final record.

Because of the lockout, we did a Winter Forecast (with 115 voters) in lieu of the usual Summer Forecast.

[Crow]

Turns out only 1 team won less than 20 games.

My low Orlando prediction came out 1 win low.

[RoyceWebb]

Tanking and such created havoc in the final week. In any case, the ESPN Winter Forecast (115 voters) ended up at 3.87.

[Mike G]

Maybe it's time for the final results.

Code: Select all

Avg error from actual 2012 wins
3.47   Crow        4.47   ez2
4.00   J.E.        4.49   DSMok1
4.07   JohnH       4.87   zRA
4.20   Vegas       4.90   ezPM
4.20   dis         5.12   KevinP
4.31   MikeG       5.37   11Pyth

In the following summary a bold listing means "within one game" and bold italicized is a miss of more than 10.
Crow had the best predictions for Den, Hou, LAL, Mem, NYK, Orl, Sac and Okl. Highest error on Por.
J.E. was closest on Hou, LAC, and SAS. Worst on Ind, Uta, Phl, and Dal.

Ties are listed as best or worst, except for the Wizards: 4 are tied for best and 3 tied for worst.

John Hollinger had the best guess for Okl and Cha; worst for Mil (off by 3).
dudeimstoked was closest on Bos, Mil, NJ; big misses on SAS, GSW
Mike G closest on Chi, Cle, Mia, NO, GS, Uta, and SAS. Worst on Den, Okl, Min, and Phx.

[Crow]

I was closest on 8, 6 in the west. That makes sense as I know the west fairly well and the eastern conference a lot less well.

Mike G closest on 7, 4 west, 3 east.

Pretty good outing for dude i'm stoked.

[Crow]

Not really surprised that 11Pyth did poorly. Another entry finished near the bottom as it usually has. Somewhat surprised and disappointed that ezpm, rapm and the blend were weak. Wonder if there were somewhat tweaked ezpm weights that retrospectively would have done better this season... and into the future. Wonder if there was a better ezpm / rapm weight blend or if a third metric would have helped. Probably are such opportunities to strengthen them.

[wilq]

I guess it's better late than never...
Great stuff Crow! But you were "only" 2nd on the internet
http://weaksideawareness.wordpress.com/ ... -for-wins/

[Crow]

Thanks. And thanks for letting me know your results.