A full and clear explanation of each column is given:
Now shows BetFair market odds against each prediction. This allows potential good value bets to be easily identified. Of course the predictions need some interpretation. Victor does not know about sporting news, such as a player being injured or a match played at a neutral ground. This information needs to be taken into account. Find out more about what the figures mean
Betting Tips are suggested. A betting strategy has been devised using Victor, it shows a good profit on the last two seasons. Suggested bets appear in the Tips column.
The figures are complied using a Neural Network, the Network has a number of input terminals, a number of output terminals and some artificial Neurons connected to inputs and outputs. Numbers may be entered on the Input terminals and read from the Output terminals. The Network can be trained to associate a pattern of numbers in the inputs to another pattern on the outputs. In fact it can associate a huge number of different input patterns to output patterns. What’s more if the Network sees an input pattern it’s never encountered before it can make a very good guess at the corresponding output.
How does this help us? We construct a Network with three outputs. Those outputs represent the outcome of a soccer match:
We now add our inputs, we add one input for each feature of a Soccer match we think might be relevant in determining the outcome. For example ‘Number of matches won this season by home team’ would certainly be useful but ‘Strip colour of away team’ would not be useful. We then present to the Network the (useful) input parameters along with the result of the Match. After seeing examples of this soccer data the network learns to associate particular sets of inputs as being more likely to lead to a Home win (or Away win, or Draw). In detail the inputs used by Victor are:
When the network is trained ten years worth of soccer results from the main four English divisions are presented to the network one thousand times. It takes about twenty minutes to do this.
Victor the Predictor is an English soccer odds predictor. Victor makes his predictions using a Machine Learning technique called Neural Network. The Network is trained on the last 10 years of soccer results and no statistical analysis or human opinion is involved. Whilst Victor's predictions are usually in good agreement with Betfair markets, things happen in the real world which Victor knows nothing about, for example a key player is injured. This will impact the betting, so a large difference between a prediction and a market doesn't necessarily represent a good betting opportunity. Victor's performance tends to improve as the season advances, as each new weeks' results forces adjustments to his neural network. However, towards the end of the season as cups/promotions/relegations loom, the predictions become more difficult again. Victor finds much more consistency and predictability in the Premiership than in the other divisions. Find out more about How the figures are produced. What's new in Verison 2.00. Find out more about What the figures mean.
Neural Networks built: Sat, 27 Jun 2009, 06:21. BetFair Markets scanned: Wed, 17 Jun 2009, 11:15