We’re pleased to announce version 2.0 of our betting algorithm. This shows significant improvements in the key performance statistics used to measure the quality of our betting system.

At the same time we’re introducing a one number descriptive rating statistic for our betting system to enable comparisons over time and across leagues.

Value betting
Our previous algorithm backed or layed teams based on whether the market odds were over or under priced. However the amount staked was not dependent on how mis-priced the market odds were. Our latest algorithm bets more when odds are mis-priced by a larger amount and less when odds are close to our own internally generated odds.

This allows us to take advantage of times when the market appears to have badly mis-priced odds.

Remember, our tips include a staking system so you don’t need to be concerned with how much to bet once you’ve decided how much cash to allocate to your betting fund.

Rating system
There are many statistics that provide different ways to describe the performance of a betting system. These most commonly include yield and returns.

Since any one statistic is deficient in some way but too many stats do not make for an easy comparison, we have created just one combined statistic which is the product of our 3 most important descriptive statistics. The 3 included elements are yield, return and r-squared. We’re referring to it as the Algobet rating. The pros and cons of each individual statistic are described below.

Yield is a useful indicator of quality but lacks information on potential profits. A high yield shows a betting system is not just backing favorites, but it may just be placing one bet per month hence generating low returns.

Return does give an indication of potential profits but does not give an indication of quality, unlike yield. We use average monthly returns.

Finally, to minimise risk of variation in returns we use an r-squared measure. This measures the variation of our betting fund against its long run return. The ideal is a figure of 1 which is analogous to receiving a fixed x% each and every month.

To see details of the new algorithms in place see here. The Bundesliga and EPL are released ready for the next matches and the La Liga and Ligue 1 models will follow.