NHL Totals ( Math Problem )

[FooFghtrs33]

Hey guys, i have been trying to figure out this NHL totals problem for a bit now and am stumped. I use Poisson which works excellent for odd number totals, like 5.5 goals. For example, I expect 5.3 goals scored total, poisson makes the over +130 under -130 for 5.5. Now this same problem with goal total of 5 makes poisson over AND tie (ie. the total is 5 or more) -155 and under (ie. total is 5 or less) -130. This math all makes sense to me, my question is how the books do it. Since their lines are dime lines, do they account for ties (or it landing on 5 total goals)

If you calculate it with out ties, this same example using sbr poisson calc would be Over +120 and under +156.

So if anyone knows how the lines are created, or how they factor ties into the totals, let me know. Any insight is appreciated!

[newbee]

Obviously they factor in the pushes. That is reflected in the difference in ML you got from running Poisson (i.e. 75c difference you got is quite significant, dont you think?). 5 is the most frequent score in NHL, so not accounting for it would be just plain dumb.

[Tackleberry]

How you wrote your post has me scratching my head slightly. You cannot come up with the prices for a total of 5 or 6 without taking into account the pushes.

For a total of 5 you can use poisson to determine how often you expect there to be 4 or less goals, how often you expect there to be exactly 5 goals and how often you expect there to be 6 or more goals. Then remove the amount of times the wager pushes and adjust the probability of the 4 or less and 6 or more accordingly. You now have your line for a o/u total of 5.

[RickySteve]

[ ] NHL totals are Poisson distributed

[Stansfield]

Tack beat me to it. You have to account for the push then figure for each scenario and remove the push.

[mrmarket]

You might want to check this out:

http://www.hockeyanalytics.com/Research_files/Poisson_Toolbox.pdf