Poisson Distribution and the Draw

[Godders84]

Hi all,

Im trying to get my head round this Poisson distribution for predicting scores theory, and think I'm most of the way there, however...

I understand the Poisson undervalues the draw due to the way it is calculated. As an e.g., two teams play each other, both with a calculated mean no. of goals of 1. According to the Poisson distribution, there is a 34.58% chance of either the home or away team winning, but only a 30.86% chance of the draw. Surely there is a greater chance of a draw than this shows?

Is this really correct, or is there something else I could do to see the "true" win/draw chances of the game? If so, could someone explain this to me in laymans terms (my statistical knowledge is not as good as some people's in here!)

PS. I primarily bet the English Premier league, where I beleive the Home team currently wins around 53% of the time, the away team around 20%, and the draw occurs around 27% of the time. Don't know if this info is useful...

[voros]

Here's the draw percentages from the top European leagues:

http://numeridicalcio.wordpress.com/2006/08/20/percentage-of-draws-in-european-leagues/

Obviously the issue is that draws are more likely the more even the matchup is.

I'm not sure the best way to test this since, with home field, it's often hard to determine when a matchup is "even."

In a separate study I did, if you calculate result% as:

result% = (wins + (draws*0.5)) / games

I came up with a formula of:

draw% = (1.31 * result%) - (1.31 * (result%^2) )

For a result% of 0.5 (an even matchup), that gives a draw% of 0.3275. That's a touch more than poisson above.

But I would by no means consider that study definitive.