2008-9 Predictions

Home for all your discussion of basketball statistical analysis.
Post Reply
Crow
Posts: 5567
Joined: Thu Apr 14, 2011 11:10 pm

2008-9 Predictions

Post by Crow » Fri Apr 15, 2011 7:49 pm

final page

Author Message
Jacob



Joined: 29 Nov 2007
Posts: 19


PostPosted: Wed Feb 11, 2009 4:49 pm Post subject: Reply with quote
Bill Simmons is an underrated analyst, in his own way. He has a great feel for team & personality dynamics, stuff that is hard to predict with numbers... He's at his best with calls like "when will the Pistons get tired of each other and force Joe Dumars to make a trade" or "how will the subtraction of Jamal Crawford affect the Knicks' chemistry."
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3528
Location: Hendersonville, NC

PostPosted: Thu Feb 12, 2009 7:55 am Post subject: Reply with quote
Simmons called for the Cavs to win 58 games, while the rest of us picked 44-49. And for Philly to win only 43 (we all thought 48-54). Those are his only outstanding 'best' guesses.

He's dead last on Cha (guessed 21 : headed for 33), NO (62 : 49), Sac (37 : 17). But I like that he placed a wide range on his guesses: From 16 (Mem) to 66 (LAL). Hollinger went only from 20 (Okl) to 60 (Bos). DW narrowest at 24 to 61.6 .

I took the wildest swings, from 14 to 67, and I'm last.
_________________
`
36% 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: 3528
Location: Hendersonville, NC

PostPosted: Thu Feb 26, 2009 9:44 am Post subject: Reply with quote
Presently, all our guesses are better than they've been at any time this season. However, teams have just tended to follow last year's records more closely; and a couple of us are again worse than if we'd just penciled in 2007-08 wins.

Here are the 'standings', including Bill Simmons'; and for reference, last year's W and Pythagorean W (difference avg).

2 columns represent current wins + pyth-ex for remaining games; and straight pythagorean over all 82 games. Note that we were all better at guessing pyth, some much more-so than others (luck factor).
Code:
progno. W+pyth pyth
(Bill S) 7.01 6.96
John H 7.84 6.94
D21W21 8.05 7.33
Mountain 8.18 7.95
Mike G 8.42 8.07
Kevin P 8.63 8.07

'08 W-L 8.27 8.51
'08 pyth 8.88 8.89

The persistent superiority of '08 WL over '08 Pyth as a predictor seems to mean that this has been a persistently odd year. The same teams keep on winning/losing the close games?
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Thu Apr 16, 2009 2:42 pm Post subject: Reply with quote
Mike can do the official tally here but in the smaller original set of projections at APBR I think Hollinger edges DavisWylie out for the projection win in just the last few games.

I was way off on Milwaukee, Indy and Miami and that was my downfall. I doubted Skiles (too early), didn't count on Dunleavy being out and doubted Spoelestra and Wade too much. I knew I could predict better if I played the edges again but I went for broke trying to predict these and some other as surprises. I got some right. Throw those 3 out I'd be in photo-finish with late inclusion Bill Simmons who beat the initial group. But Hollinger and DavisWylie can say the same thing with their 3 with the most departure to the worse compared to Simmons.

Looking at a wider set of projections Vegas wins overall by a tiny bit, followed by Simmons and then Accuscore.
Projection of East- Simmons was best. West- Accuscore.
I was better in the West than I was in the East again but not that good in the west this season. Projection accuracy on top finishing teams- Simmons best. Middle- Mountain. Bottom- Accuscore. I see lots of folks who missed by more than 10 games on 10+ including Vegas. At least some folks got within 5 on 10 or more teams. Probably will be that way most seasons.

Other subsets that might be interesting to check:

Top, middle, bottom by actual defensive efficiency- which methods matched up well with this often under-tracked and accounted for half of the game?

And maybe the top , middle, bottom by actual 3 point performance by 3 pt FG% or makes or the product would be better. I wonder how important this factor is right now.

And look at the teams with coaching changes or major injuries. This affects the results in ways that were less predictable than the rest of the considerations. Could also look at how folds did predicting the teams with coaching changes or major injuries last season though.

You could do even more with coaching, looking at teams by coaching career win % or recent performance vs Pythagorean or difference between these and last season's team performance. How much does coaching matter? Hard to isolate but important stuff.

And look at predictive performance for the teams sorted by a net efficiency consistency measure. I guess pace would be worth checking too since it may bias some of the projections.

The predictive performance for the teams sorted by top, middle, bottom by team performance change from 07-08 to 08-09 would be very important. Who nailed the significant movers?

What weighting combination of publicly explained system-based projections fit the actual results best? Would that meta-metric perform well next season? You could play and learn a lot with this, if you wanted to, had the time and a stake.

Last edited by Mountain on Sat Apr 18, 2009 1:06 pm; edited 1 time in total
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Fri Apr 17, 2009 12:08 pm Post subject: Reply with quote
Actually. correcting for one data input error I see Simmons beats Vegas by 2 games. I don't see any other notable changes caused.

Last edited by Mountain on Mon Apr 27, 2009 2:53 am; edited 1 time in total
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Fri Apr 17, 2009 3:16 pm Post subject: Reply with quote
Recalling I also revised my predictions Nov. 13, I see those revised predictions (with an advantage obviously, but a scoffed at advantage at the time- "There is not much data in 7 games") "beat" Simmons.

Last edited by Mountain on Mon Apr 27, 2009 2:53 am; edited 1 time in total
Back to top
View user's profile Send private message
back2newbelf



Joined: 21 Jun 2005
Posts: 250


PostPosted: Fri Apr 17, 2009 6:22 pm Post subject: Reply with quote
cleveland detroit 4-1
boston chicago 4-1
orlando philly 4-1
atlanta miami 4-3

la utah 4-1
denver hornets 4-3
blazers rockets 4-3
spurs mavericks 4-3
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 3528
Location: Hendersonville, NC

PostPosted: Sat Apr 18, 2009 8:14 am Post subject: Reply with quote
And we have a winner:
Code:
Preseason Predictions Actual Absolute Error
Mtn Mike Kev John DW2 Won tm Mtn Mike Kev John DW2
48 44.2 48.0 47 49.0 66 Cle 18 22 18 19 17
42 43.5 32.2 41 42.7 54 Den 12 11 22 13 11
37 34.2 34.4 31 32.7 47 Atl 10 13 13 16 14
50 44.7 46.8 46 44.5 59 Orl 9 14 12 13 15
45 44.3 50.6 42 39.5 54 Por 9 10 3 12 15

27 25.8 47.4 39 34.9 43 Mia 16 17 4 4 8
31 19.4 24.5 28 20.9 32 NYK 1 13 8 4 11
21 30.6 23.0 32 27.4 34 Mil 13 3 11 2 7
50 48.0 41.8 47 53.6 54 SAS 4 6 12 7 0
27 29.1 33.5 27 24.5 34 NJN 7 5 1 7 10

63 65.8 62.3 57 60.2 65 LAL 2 1 3 8 5
19 14.5 21.4 26 24.5 24 Mem 5 10 3 2 1
23 33.7 36.6 40 31.4 35 Cha 12 1 2 5 4
43 40.7 31.7 39 41.5 41 Chi 2 0 9 2 1
18 21.1 23.5 20 24.0 23 Okl 5 2 1 3 1

63 66.9 52.6 60 61.6 62 Bos 1 5 9 2 0
52 50.8 47.3 45 52.1 50 Dal 2 1 3 5 2
55 48.0 48.2 50 49.3 49 NOH 6 1 1 1 0
51 52.6 43.6 45 49.1 46 Phx 5 7 2 1 3
56 51.2 54.8 56 59.6 53 Hou 3 2 2 3 7

46 45.8 37.9 41 35.9 36 Ind 10 10 2 5 0
29 31.3 38.4 31 32.5 24 Min 5 7 14 7 9
48 49.1 53.6 50 48.0 41 Phl 7 8 13 9 7
57 60.9 49.3 58 58.6 48 Uta 9 13 1 10 11
33 38.9 47.2 40 35.2 29 GSW 4 10 18 11 6

44 45.5 43.0 47 41.1 33 Tor 11 13 10 14 8
34 27.3 30.5 23 30.2 17 Sac 17 10 14 6 13
32 28.5 34.2 30 34.7 19 LAC 13 10 15 11 16
55 54.6 53.0 54 54.2 39 Det 16 16 14 15 15
31 39.0 39.1 36 36.5 19 Was 12 20 20 17 18

average errors: 8.20 8.60 8.62 7.80 7.74
Mountain was closest on Atl, GS, Min, NY, Orl, Phl, and Was; worst on Cha, Ind, Mil, NO, Sac, Okl
I got best with Cha, Chi, Dal, Den, LAC, LAL; worst on Cle, Mem, Mia, NY, Phx, Uta.
Kevin got Det, NJ, Por, Okl; furthest on Bos, Chi, Den, GS, Min, Phl, SA, Was.
John won Mia, Mil, Phx, Sac; last with Atl, Dal, LAL, Tor
DWNeil took the enchilada, ending up with Bos, Cle, Ind, Mem, NO, Phl, SA, Tor; losing Hou, LAC, NJ, Orl, Por.

If I replace the 'Won' column with Pythagorean (exp = 13.5), we all are better (avg err). John wins and I lose, both by wide margins.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Neil Paine



Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA

PostPosted: Sat Apr 18, 2009 12:01 pm Post subject: Reply with quote
Thank you, thank you, it's such an amazing honor to win this very prestigious competition! There are so many people to thank! The academy, of course, and Microsoft Excel, without whom none of this would have been possible! And finally, to all the little people out there... You like me, you really like me! I couldn't have done it without you!

Thank you America, and good night!
Back to top
View user's profile Send private message Visit poster's website
Mountain



Joined: 13 Mar 2007
Posts: 1527


PostPosted: Sat Apr 18, 2009 1:03 pm Post subject: Reply with quote
I rounded them all off to nearest integer. The next digit made the difference.
Back to top
View user's profile Send private message
John Hollinger



Joined: 14 Feb 2005
Posts: 174


PostPosted: Thu Apr 23, 2009 11:21 am Post subject: Reply with quote
If Byron Scott had fouled at the end of the New Orleans-San Antonio game instead of letting them shoot a game-tying 3 I would have won. So not only did he cost Houston the division and Portland a No. 3 seed, but he affected me personally. Damn you, Byron Scott.

Post Reply