Graham Kendall's Page

Gifts Received

Graham Kendall has not received any gifts yet

Give Graham Kendall a Gift

Graham Kendall's Blog

Predicting Football Matches

Hi all

(New member)

I have just started a blog that really talks about my various research interests (see http://research-reflections.blogspot.com).

My latest post (http://research-reflections.blogspot.com/2009/07/predicting-results-of-football-matches.html) has started to discuss how to predict the outcome of football matches. I have some ideas already (e.g. use the shortest odds, select the home team, look at previous encounters etc.), but I wonder if Sports… Continue

Posted on July 21, 2009 at 19:35 — 11 Comments

Comment Wall (4 comments)

You need to be a member of The Sports Exchange to add comments!

Join The Sports Exchange

At 11:34 on September 6, 2009, Adonis First said…: Interesting Graham,
I've never been big on footbal (I'm a horseracing guy!) so it may be that a bit of cooperation could keep things applicable more broadly....
Drop me an email on adonis@integermachines.com and let's see if there's complementary areas.
I'm out for the rest of the day, soon, so don't expect a reply for 24 hours or so!

At 18:04 on September 5, 2009, Graham Kendall said…: Hi Adonis

Yes I do have an interest in genetic algorithms and simulated annealing (and also things like tabu search, great deluge algorithms etc.). And, it goes without saying that NP-Hard/Complete problems feature heavily in the type of problems I look at.

I am also not a mathematician but, as you say, there are some real clever people out there.

WRT the football prediction, I want to use (amongst other things) a neural net (probably back prorogation, but also evolve them). If this is successful I'd like to "benefit" from it before publishing :-).

But, if it does work, I'd like to try on different domains and things like stock market forecasting (both shares and markets) are attractive. The beauty of these domains is (at least) two-fold. 1) Data is easily available. 2) You don't need much of an edge to make a profit.
Also, I have published (a little) in this area.

Of course, that makes it all sound easy and, we both know, that that is far from the case.

Happy to discuss further.

Cheers

At 10:14 on September 4, 2009, Adonis First said…: Hi Graham, I see you have interest in genetic algorithms and simulated annealing. I suspect that means you have (or have had) some interest in NP-complete problems too. In particular the Old Nuts: The Travelling salesman's Problem and Factorisation of Prime Products (my own long-time favourites).
I used to be European Coordinator for a large (the largest) Microprocessor company on all activities involving Encryption, Neural Networks and high-throughput maths routines. Some was "spook" and I can't discuss, but lots of it was "industrial" and some even "fun"! Met some AWFULLY clever people along the way! (I'm not a star mathematician, but can appreciate real talent when I see it.....)
For a long time now, I've thought about (and not done anything) about feeding live Exchange matching data into a Net. I feel quite sure that it would provide some form of honest edge, and even indicate where and when fraud was being practiced too!
I did a lot of work using nets for credit card application/use screening. It made a huge impact on closing down fraud....
Let me know your particular interests, we might have some in common?
Adonis.

At 19:27 on July 21, 2009, john Tuohy said…: Very Welcome Graham.

Nice smile as well.

No comments yet!