from Amazon
Monday, February 22, 2010
Olympic Hockey data
From the never-rested Hawerchuk, I agree that the historical perspective that is being given out is ridiculous. Canada v USA has for more history than what you see in the Olympics. And, I found the good v bad data interesting. The top-7 teams score an average of 4.8 goals and allow 1.3 goals. Since hockey scoring follows the Poisson distribution pretty closely, we can figure out the chance that a good team should beat a bad team. I get 89.6% wins, 6.1% ties, and 4.3% losses. I figure that the better team will win 79% of the ties (4.8 divided by 4.8+1.3), so a final win% of 94.4%. (The tie-breaker only works if you play sudden death. If it’s a 5 minute OT, followed by shootout, then the chance of winning the ties will go down somewhat.)
So, the Vegas odds should be around 17:1 that a bad team beats a good team. What are they actually showing?
That’s funny that you come up with 95% or so Tom. I was talking to a friend about this last night and said I figured we had to be at least 90% to take out Germany, to which he responded that he would put $10K on that at those odds. Canada’s stronger than the typical good team and Germany’s probably weaker so he’s probably bang on.
This is frustrating hockey to watch. It just kills me that we’ve evolved from a system of “Pick the best goalie who plays for the coach or GM of Team Canada” to one in which a goalie holds the spot until he is decisively dethroned. Hopefully that’s the end of Brodeur in the Canadian nets - Luongo dropping butterfly after butterfly would have looked great last night, particularly on the asinine fourth American goal.