taking averages of previous games to determine outcome

crapula · Apr 1, 2004

i do this quite often and win, but i was wondering if it is actually a decent way to go about determining over/ unders?

EX: tonight I have over 26 in grand salami, i added all the points scored in all the teams that are playing tonight past games against eachother, divided by number of games they've played against eachother and multiplied by number of games tonight (5) = 27.36 (so i took over 26, if the o/u was 27 i wouldn't have bet). any ideas on my system.

crapula · Apr 1, 2004

here's the math i used:

points in all past games (115)
number of past games (21)
number of games tonight(5)

115/21 = 5.4762

5.4762 * 5 = 27.38 total points tonight

barman · Apr 1, 2004

It's a crappy system.

Oh wait, I cheated and just looked at all the final scores which only Totaled 24....

Seriously, my only thought is that given a large enough sample, most statistics will gravitate towards Center.

Thus your 21 previous games averaged 5.47 while the league average is just a bit below that. So in order to gravitate towards Center, it would suggest an Under tonite.

But of course that's only one example using only 5 matchups and 21 previous games. A larger sample would provide more insight.

You could test the idea by looking at other nites and doing the math, though it would be awkward in hindsight.

Stick around. You'll see more good ideas and will likely contribute some of your own.

Welcome to RX

Phaedrus · Apr 2, 2004

Covers has a nice page on matchups for each sport, which compares similar past games with upcoming ones (i.e. same home/away team, usually same fav/dog situation and similar totals.) I have not followed it in all sports, but the "prediction" (which is really just looking at the most similar recent matchup between a given two teams) is suprisingly accurate in the NBA.

Not exactly the same thing you're talking about, but a similar concept.

Phaedrus