AndrewKent Posted March 10, 2011 Share Posted March 10, 2011 Hi guys, Long time no post! I had a lot of good feedback in my previous threads about the bookies being all over the traditional ways to predict match outcomes, and that a unique avenue would need to be found before any profit could start being seen. So, rather than quit, I've been working hard to try and find that solution. Not there yet but will be one day :lol I had a general question (maybe it Punters experience here or maybe it would be mathematics) about weighting systems against each other. Lets set up a scenario. Okay, I have two systems, A and B. Both systems produce a number and from correlating these against the match results, the higher the figure above 0, the more chance of a home win, the lower the figure, the more chance of an away win (a higher draw % is somewhere around 0). Both systems measure different parameters and data ranges but the correlation that I've just mentioned is roughly the same. So with this in mind, I want to weight these two systems against each other to produce one rating for the match that gives me an indication of whether I should bet of not. This is the part I am not sure about. I have tried looking at the Strike Rates of each system individually and then producing a weighted average however I'm unsure if this is the right way to go. How would one rate 2 or more systems against each other to produce one match rating? As for what systems I am using? Well, that's still being worked on. Tweaked versions of the Paul Steele systems, other systems I am build ect. Regardless of the final systems, weighting them against each other is something I would love to find out how to do. Can anyone offer any advice? Kind regards, Andy Quote Link to comment Share on other sites More sharing options...
clay747 Posted March 10, 2011 Share Posted March 10, 2011 Re: Weighting Ratings If i was convinced these input signals(systems A & B) were good, but, i could not find the right balance through trial and error, i might be tempted to try a neural network(I never have). With good signals, trial and error normally works. If its not working its probably better to go back to the drawing board with your input signals than to spend time trying to make them work. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.