Statistic Guys: Get in Here!

Site Notices

TNVC Partners With Envision Technology - MARS Rangefinder Family, MARS-Lc!

[ARCHIVED THREAD] - Statistic Guys: Get in Here!

Posted: 9/4/2014 3:09:26 AM EDT

I posted this on a stats forum in the applied section and didn't get a response. Hopefully, I will have better luck here. Please let me know if I need to clarify my goal.

I would like to apply an algorithm, so I can compare high school football teams from different years. I would like to rank the teams and then, if possible, project matchups, if these teams were to play against eachother (the website Calpreps.com, already does this, but with its own ratings system).

I already have the ratings from Jeff Sagarin with his expressed, written consent to redistribute in nonprofit uses.

- 2013 ratings

- 2012 ratings

So, I understand that it's possible to analyze teams with common opponents or opponents of opponents; however, these 2012 and 2013 data sets are completely isolated from another. I assume there is some way to normalize everything, so we could compare them ... which is why I am here.

How would you do it?

Posted: 9/9/2014 8:13:00 PM EDT

[#1]

Quoted:
I posted this on a stats forum in the applied section and didn't get a response. Hopefully, I will have better luck here. Please let me know if I need to clarify my goal.

I would like to apply an algorithm, so I can compare high school football teams from different years. I would like to rank the teams and then, if possible, project matchups, if these teams were to play against eachother (the website Calpreps.com, already does this, but with its own ratings system).

I already have the ratings from Jeff Sagarin with his expressed, written consent to redistribute in nonprofit uses.
- 2013 ratings
- 2012 ratings

So, I understand that it's possible to analyze teams with common opponents or opponents of opponents; however, these 2012 and 2013 data sets are completely isolated from another. I assume there is some way to normalize everything, so we could compare them ... which is why I am here.

How would you do it?

View Quote

I would not waste my time even trying.

You have far to many degrees of freedom and not enough games to have meaningful results with any statistical value.
Multivariate regression takes HUGE sample sizes.
Teams change their members far to often.

[ARCHIVED THREAD] - Statistic Guys: Get in Here!

General » Math And Science

Win a FREE Membership!

Sign up for the ARFCOM weekly newsletter and be entered to win a free ARFCOM membership. One new winner* is announced every week!

You will receive an email every Friday morning featuring the latest chatter from the hottest topics, breaking news surrounding legislation, as well as exclusive deals only available to ARFCOM email subscribers.

Warning

Confirm Action

About AR15.COM

Stay Connected

Newsletter

Contact Us