| |
NHL Forum
Thanks
Posted By: MarkH In Response To: Poisson Modeling won't work (boopotts)
Date: 29 Oct 03, 9:44 pm
I completely take on board what you are saying. As I have said previous posts, I am completely skeptical about my own methodology.
But I think its worth trying a bit more. I am using average goals over a whole season, so the small aberrations you are talking about should not affect how true the average is. This is a low scoring game where the number of points scored in each individual game will not be a very true measure of relative strength (cf high scoring games like basketball) so a small sample will be pretty meaningless as an indicator of strength.
Using a long-run average approach as the basis for my probability estimate, if what I am trying to do has some validity, then it will not emerge for a very long time, maybe a whole season. So I will keep plugging away, even if the immediate results do not look promising so far! Hopefully this way we can say yes it works or no it does not.
Thanks
MarkH
- Wk of 10/28, 7-8-4, -8.3u (nt) -- MarkH -- 28 Oct 03, 6:09 am
- 10/29 -- MarkH -- 28 Oct 03, 6:21 am
- Poisson Modeling won't work -- boopotts -- 28 Oct 03, 10:07 pm
- Thanks -- MarkH -- 29 Oct 03, 9:44 pm
- Heavy Chalk -- Matt Baker -- 30 Oct 03, 6:00 am
- Heavy Chalk -- Matt Baker -- 30 Oct 03, 6:00 am
- Res: 3-2-0, +1.15u (nt) -- MarkH -- 29 Oct 03, 9:45 pm
- Thanks -- MarkH -- 29 Oct 03, 9:44 pm
- 10/30 -- MarkH -- 29 Oct 03, 9:57 pm
- Res: 5-0-0, +5.65u (nt) -- MarkH -- 31 Oct 03, 5:58 am
- 10/31 -- MarkH -- 31 Oct 03, 6:34 am
- Res: 0-1-0, -1.35u (nt) -- MarkH -- 1 Nov 03, 4:05 am
- 11/1 -- MarkH -- 1 Nov 03, 4:40 am
- Res: 3-0-2, +3.1u -- MarkH -- 2 Nov 03, 3:19 am
- 11/2 -- MarkH -- 2 Nov 03, 6:51 am
- Res: 1-0-0, +1u (nt) -- MarkH -- 2 Nov 03, 9:37 pm
- Poisson Modeling won't work -- boopotts -- 28 Oct 03, 10:07 pm
- 10/29 -- MarkH -- 28 Oct 03, 6:21 am
| |
NHL Forum is maintained by Pi Yee Press