Introduction
Correct Score Neural Net
Betting recommendations provided by this site
About me

Introduction

Betting advice posted on this site is based upon the output from a neural network developed in my spare time over a period of approx. 3 years (plus a gestation period of some several years). I've called this neural net "Correct Score Neural Net".

This is a personal project that combines my interests in neural network computer modelling and football betting, both of which I've studied over a period of many years.

Top

Correct Score Neural Net

It's been an interesting and challenging project which has been a great learning experience.

At some point I shall probably document (on this website) some of the more interesting problems I had to overcome.

The network has been trained on 10 years of data for all the English divisions and takes into account factors such as:

• recent form (last 8 overall for both teams, last 6 homes for home team and last 6 away games for away team)


• season form


• match factors including: next game importance, this game importance (derby game?), new manager(s) plus several others, though not too many others because I think it's important to avoid extraneous factors that don't have a proven influence.

Recent form was an interesting problem, because recent results are entirely dependent upon match conditions, the quality of the opposition and their playing style (attacking or defensive etc.), so this had to be factored-in. My solution was to adjust recent form input vectors using the network's output vectors for those matches - a simple but powerful solution.

Training the net has been the most challenging, because the network is not presented with an equal number of each pattern, or even an equal number of patterns with the same output, so mechanisms were required to prevent similar well represented patterns from dominating, whilst allowing the output to be shaped by the natural distribution.

Top

Betting recommendations provided by this site

This site is aimed at punters who use a value based betting strategy.

Football is an extremely unpredictable game. I won't elaborate too much on that statement, other than to say anything can happen in those 90 minutes or so and often does, which means it's basically impossible to predict the outcome of a football match with any degree of certainty (not withstanding divine insight :-)).

What we can do however, is calculate reasonably accurate probabilities and neural networks are the best way to do this).

If we can calculate accurate probabilities, then we can identify bets where the price offered by the bookie(s) represents good value and in the long run we'll win more than we lose.

One reason we can often find good value prices for correct score bets, is that prices for correct scores are usually based upon the win, draw and loss prices, using look-up tables.

For example if a match involves two teams who freely score and concede goals and another match involves two teams who have drawn their last 8 games 0-0, if the draw price is the same for both matches, then the price for scorelines of 0-0, 1-1 and 2-2 will be exactly the same for both matches!

Tips posted on this site will cover all weekend matches with value bets being highlighted. I don't offer any guarantees for recommendations posted on this site, however if you bet on the value selections I believe you will make a profit over the course of a season.

Note: you can usually find a local bookie that pays out triple the odds for just 1 winner, which can make a big difference. Often you will find that a shop on the edge of town will pay out double the odds for 1 winner, whilst a branch of the same bookmaker in the centre of town (surrounded by other bookmaker shops) will pay out triple odds for 1 winner.

Top

About me

I am 39 years old, I have a Bsc (hons) in computing and I currently work for a large software company.

Personally I don't believe in luck, or rather I believe luck is completely random - as far as I'm concerned there's absolutely nothing you can do to change your luck. I think the idea of being 'hot', 'on a roll' or 'having a lucky streak' is largely coincidental.

Having said that, the mind is a very powerful computing machine which can be just as effective as the most sophisticated neural network, when used correctly i.e. when you get into the right frame of mind - when you train your mind correctly. The 'gut feeling' often referred to by gambling folk may often be a form of fuzzy reasoning that takes into account every factor, exactly weighted against each other (sub-consiously). I would suggest that the ability to conceptualise at high levels of fuzzy complexity can only be achieved without thinking and that some people are too intelligent and logical to have such a macroscopic way of not-thinking.

It's amazing what feats of human excellence can be achieved when the grey matter is allowed to map itself in the right way - just think of what Ronnie O'Sullivan can do with a snooker cue. And yet there are many people who play snooker all their lives and rarely make a break past 30.

In my experience, most people don't want to learn, they want to be taught.

In bookies around the country you'll find people who have been betting most of their adult lives, using the same old simple betting systems handed down to them, never evolving, frequently losing and never even learning from that lesson.

My advice is: if you want to win, you need discipline and patience. You're not going to win every time you walk into the bookies and you can't force your luck, but if you bet with intelligence (and always try to get 'value'), you'll win more often than you lose and you can take comfort from the thought that if you play the percentages, one day everything will fall nicely into place and you'll land the big one.

Recommended reading: "The definitive Guide to Betting on Sport", Racing Post Expert Series, edited by Bruce Millington, ISBN 1-904317-36-7 (available on-line from the Racing Post website).

Top