prediction model for totals

[Sartre]

Got the advice to start a new topic on creating a predictive model for totals.

How do I do it?

If you read my other thread on how to get started with modeling you know by now I`m a complete idiot when it comes to this, but I`m interested in learning!

/S

[Crow]

Check around the internet and maybe in Mathletics by Wayne Winston.

Here are a few miscellaneous links I saw on a quick web check that might interest you:

http://blog.smellthedata.com/2011/02/th ... tball.html

http://sportsgambling.about.com/od/nbab ... factor.htm

http://espn.go.com/blog/playbook/dollar ... ba-gambler

http://www.nbastuffer.com/tag/NBA_Score ... tions.html

http://www.teamrankings.com/blog/ncaa-b ... everywhere

[Mike G]

Suppose you want to predict Knicks-Nets.
NBA avg pts/G is 100
Knicks avg 95.3, which is -4.7 from avg; and Nets allow 100.7, which is 0.7 above avg. Therefore, Knicks can be predicted to score 100-4.7+0.7 = 96.0

Nets avg 96.3, which is -3.7 on offense; Knicks allow 98.0, which is -2.0
Nets should score around 100 - 3.7 - 2.0, or 94.3

Unless one team has more home court advantage than the other, the expected total score should be = 190.3

You can fine tune this with strength of schedule (SOS) adjustments. Two teams in the same conference will not be much different in SOS at this time of the year.