Poisson Distribution - Win % between two teams (Excel Spreadsheet) (February 20, 2004)
A long time ago, I think I surmised that the goal scoring in the NHL follows a Poisson Distribution. If this is the case, then the above spreadsheet is useful for those who have the goals scored and goals allowed.
(Now that I think about it, someone's going to ask what to do if they had the goals scored and allowed for 2 teams, and they want the matchup. I'll try to fix that next time.)
Note that this does not apply in baseball. Run scoring in baseball follows the tentatively-named Tango Distribution. A program for that has already been supplied at my site.
--posted by TangoTiger at 03:17 PM EDT
Posted 3:21 p.m.,
February 20, 2004
(#1) -
Tango Distribution (right-click and _save target as_)
(homepage)
.
Posted 4:49 p.m.,
February 20, 2004
(#2) -
tangotiger
I updated the poisson file to allow you to put in "matchup" style data.
Technically, I should have a regression component to convert the performance of the team into a true talent. So, just be careful.
Posted 10:12 p.m.,
February 20, 2004
(#3) -
arbitrage
Thanks tango, I have only looked at it briefly, but it looks great.
[an error occurred while processing this directive]
Posted 9:49 p.m.,
February 21, 2004
(#5) -
tangotiger
Soccer I would guess yes. If you email me the game scores of a soccer league, I can tell you right away.
Football has weird scoring rules (7 pts or 3 pts). It's like having 1 goal and half a goal.
Posted 12:24 p.m.,
February 22, 2004
(#6) -
VoiceOfUnreason
So what's the right answer for sudden death?
My guess is that you reduce lambda by a factor of three, run out the goals scored and for those cases where both teams score, the probability is the ratio of the goals (ie, in the case where poisson suggests 3-2 in the overtime period, the team scoring three scores the first goal 60% of the time, and wins).
Posted 8:49 p.m.,
February 23, 2004
(#7) -
The Waner Brothers
Is this Big Poisson or Little Poisson...?