combining factors

[muppet77]

does any one have any ideas on combining factors into one model? eg i have a model that looks at number of corners, and another that looks at number of woodworks, (stupid i know but i made these up as an example). how could i combine the two to get HDA % predictions for matches? the fink tank has a view on this (section called 'how it works' in 'launch the predictor' at fink , but wondered if anyone would like to have some input?

[spreadman]

Re: combining factors Can't help on the models, but I can see where you are coming from, or trying to head to, re: corners/ woodwork, rebounds etc. I've not done any conclusive research into corners related data, but I am willing to share what little I do know,(which i've written on the back of this postage stamp :\ ). I have read that the times at which a specific corner occurs is random. The total number of corners in one half is not predictive of how many will occur in the second half. It is true that some teams concede or win more corners than others, on a regular basis, outside an expected 'variation'. It has not been proven that there is a direct relationship between shots on target/ off target and the total number of corners in a match. However, to me, common sense would indicate that the more shots directed goalwards, the more chance for a corner to occur. Corners are the result of many causes: deflected shots, defender error, inaccurate shooting, intentional defensive tactic. I would surmise that most 'models' extrapolate,(?), their predictive assessments from more commonly and reliable data such as goals for/against, league points/position, current form, opposition, ground conditions, time of the season, out on the p*ss the night before a match. Well, maybe not the last one. You didn't seriously expect me to make a completely sensible post, did you? :loon Hope this helps, bounce a few ideas off me, if you want.