Question about creating a program

**Flying Dutchman** · 10-21-10, 07:46 PM

It's called multifactor regression...

...lots of different ways of doing it, lots of different ways of going wrong too...

RESEARCH THE WEB, Grasshopper...

...

**MadTiger** · 10-22-10, 08:22 AM

What you want to research, among other things, is multiple regression using the Excel program, with Data Analysis plug-in (add in?).

What you are describing is actually a common setup.

**Formulawiz** · 10-22-10, 09:43 AM

Originally posted by jscol

Hi guys, I haven't posted in this section before, but have read everyone and think there are great ideas. I have been handicapping for about 10 years and have had some good times and some not so good times. I have a very good feel for certain things and at certain times. But I want to be able to bring in more statistical system to combine with my knowledge.

So I am open to many different ideas and sports. I did great baseball, but I'd prefer to concentrate on football or basketaball.

So here is my idea, tell me if it makes any sense.

I want it so I can pick out say 10-15 important statistics. Then give the scores of all the games, and kind of set up an equation. We have what it equals in the final score or margin, so can we say

X + Y + Z + A = final score. ... Then ask it to find the variables out of all their options to make the score difference come up within a range.

Any thoughts?

I think 15 statistics is way too much data to input. Give us an idea as to what statistic you feel is most important and why. I am sure some of us can help you.

**jscol** · 10-22-10, 11:03 AM

So if less statistics are better, I would want to take larger encompassing stats. Such as Team efficiency, pass/run success rate, Yards per play. Probably some things of that nature.

**jscol** · 10-26-10, 01:16 PM

is nbastuffer.com a good site? It seems pretty interesting with their predictive models, but does anyone have experience with it?

I would like to try to emulate their system and then be able to tinker with it myself?

**Maverick22** · 10-26-10, 01:20 PM

How do you hope to emulate it, without knowing how it works?

**jscol** · 10-26-10, 01:27 PM

Well, I was hoping someone had some familiarity with it. And I am more interested in emulating the idea of a predictive model rather than the exact thing.