A: The models were developed using 80,000 example cases. Once trained their performance was tested for a period covering ten months and 1256 races (over 12,000 new horse performances). The three-star bets found during the latter test period amounted to more that 900. However we hope that the Value Finder offers more than just three-star bets. It allows the user to identify overpriced horses throughout the day and hence could produce many more bets than just those highlighted.

Q: Is the structure of the model fixed or dynamic?

A: The structure of the model is fixed, so it will not evolve race by race. The networks are retrained periodically using the latest data available however.

Q: Do you provide historical data in excel format?

Data snapshots will be taken of the basic price/value data on a daily basis for users to download. This will be available in the near future.

Q: Will you provide a bet placement facility?

A: This is something we are considering but we need to get permission from the main betting exchanges before we can make such a utility available to users of our site.

Q: Does the neural network use the current price in its calculations?

A: Not explicitly. The current price is displayed in the information grid, and it is used to monitor whether the horse has drifted or contracted during the day which forms an input into the model, however the price itself is not used on the model.

Q: Just before the start of the race, I noticed that “Carbon Hoofprint” had an “Adv Ratio” of 1.25 with a 3 Star Rating. However, a horse in the same race, “Lordship” had a higher “Adv Ratio” of 1.38 but only a 2 Star Rating. Is there an explanation for this……?

A: The advantage ratio is simply the value price compared to the actual price available. In your example the horse with the highest adv ratio is, in theory, the best value. However the star rating takes into account reliability as well as the ratio. In some instances there are fewer historical example cases in the neighbourhood of the vector produced by the attributes of the runner under examination and for these we do not give three stars.

Q: I am very impressed with this service. I am sure it will be very successful. However I am a little unsure of what staking plan to use. For instance, what do I do when there is more than one three star selection in a race?

A: The star ratings are a guide to those runners with similar profiles that have performed well in the past. So if there are two *** selections in a race they should be treated as equal and both backed, if you are following the star ratings. Other members tend to make their own judgements based purely on the advantage ratio and stake proportionally. Others use the two and three star selections, essentially there are many ways to use the value finder, but I have found following the *** selections the easiest and most reliable way to make a profit.

Q: A runner is shown as 3 star say 2 hours before the race and a bet is made at that time. Can it subsequently change to a 2 or even 1 star bet during the next 2 hours, if so would the fact the bet was struck at the 3 star value mean it would still be a valid 3 star bet for my recording purposes?

A: The value of each runner is determined by the price at which it is available. So a horse may be a value at lunchtime but not subsequently due to price movements of that, or the other, runners in the race. Whether you consider it to be value for your recording purposes is really up to you. We have only monitored the performance of the star bets at off time though we are aware of some members who simply feel that a value bet is a value bet at whichever time the bet is placed.