3

I have asked around my local network, but no one seems to be able to point me in the right direction. I was a High School football coach for about 10 years, but now I am self employed and am doing some analytical work for some local football teams.

Anyway, in football when breaking down an opposing teams offense, you have a bunch of data. What I think is happening is that the opposing coach is calling plays to the right and left in almost 50/50 split, you can see it in the data.

I want to be able to compare the other teams decision with a truly random data set.

So, if the coach calls three Right plays in a row, how does this compare to his decisions in the past AND how likely is he to call either right or left on the next play.

I was told it had to do with some Bayesian statistics, but I am not sure.

Here is some example data of a recent breakdown I did last season:

Example Data Set

Pere
  • 5,875
  • 1
  • 13
  • 29
Michael Bartley
  • 31
  • 1
  • Can you give a little bit more detailed description for the benefit of us not knowing american football? – kjetil b halvorsen Feb 13 '17 at 17:27
  • Are you trying to predict if next call will be to the left or to the right based just on whether how previous calls have been left or right? Or are you trying to use more information to predict it? – Pere Feb 13 '17 at 17:32
  • 1
    Thanks for the comment. In American Football, if I know that a coach is going to call a play to the right or left, it can be of great value to me. So, I am trying to analyze his past decisions in the game and in the previous games to make a good guess as to which one he will do. My theory is that he is heavily regressing to the mean. He will call two plays to the right, then he will almost definitely call a play to the left. That's my theory, anyways, but I wanted a better way to analyze it than to guess. – Michael Bartley Feb 13 '17 at 17:36
  • @MichaelBartley Then you just want to analyze a series like LLRLRRLLL... and decide if you can predict next outcome from previous ones, don't you? How long is the series? – Pere Feb 13 '17 at 18:36
  • Check http://stats.stackexchange.com/q/229881/35989 and the provided references. – Tim Feb 13 '17 at 20:35
  • @Pere Yes, that's about right. The series will be in two parts. A historical one that is about 100 lines long and a current one that I will try to predict the next L or R on. – Michael Bartley Feb 13 '17 at 22:05
  • 2
    This closed question is not too broad. The OP is just asking if there are suitable tools to find patterns and forecast a binary longitudinal series. It should be reopened. – Pere Feb 14 '17 at 11:15

0 Answers0