Analytic approach

Glos writes:

I currently work with predictions available at Bettify and DecTech to inform my football trading. Both sites offer estimates of the percentage chance of each of the various scores in the correct score market becoming the final result. These estimates in turn can be used to determine the likelihood of results in other markets such as match odds, overs/unders, btts etc.

Bettify and DecTech use different methodologies which are to some extent described in each site’s literature. Sadly, at the beginning of this season DecTech decided to reduce the number of leagues about which it ‘gives away’ free data, and it now appears that they are selling their ‘predictor’ to a third party. I don’t know whether or in what form the new owner will continue to provide a service but am not optimistic about prospects for continuation of a free service.

Bettify offers data on a wider range of leagues although that site, too, has reduced the number of leagues covered. I find its predictions are more emphatic than DecTech’s, which tend more circumspect. Over the last 4 years or so I’ve been collating Bettify’s and DecTech’s predictions on a spreadsheet, together with the pre-match odds associated with each correct score. This allows me to calculate value by assessing the extent to which the available odds align with Bettify and DecTech’s estimate of the chances of the event’s occurrence.

It will probably not be a surprise to seasoned traders that the option of backing the ‘most likely’ score-line is not a winning strategy in most leagues, where true value is found in outcomes that are less likely but in respect of which the exchange odds available are inflated to a point where a value-bet at long odds is a much better choice than backing the favourite. Other times, value can be found in short-priced wagers such as a lay of AuQ or O3.5 in a market where there’s a heavily-backed favourite. I find there’s an edge to be found by using hard data to inform trades, but each league is different in terms of whether likely scores or value scores, both, or neither, tend to be more successful.

At the present time (April to June ’15) I’m posting single V-Bs (or packets of V-Bs) on a variety of leagues, Euro and international matches in advance of the kick-off so the TF.eu members can evaluate their success. There’ve been 49 V-B packets posted and they’ve made a total of 7% RoI (before commission) evaluated on a set & go basis, which is below my target window of 15-25%. I use set and go as my evaluation measure because I want to measure the success of my bets at forming the basis of profitable trades, rather than my own trading. When I have enough data I propose to try to improve this by concentrating on the competitions that have proved to be the most profitable.

Bingo adds: **Glos` output of suggested trades is phenomenal. Of 350 posts on the Forum some 95% are actual ideas for trades most with analysis and staking levels advised!**

For some time now I have been using two analytics sites, namely Decision Technology and Bettify, to inform my CS bet selection. Both sites use ratings databases to produce a percentage chance of each of the 17 possible scores in the CS market being the actual final result of specific matches. I use a spreadsheet to record the % chance of each score. I then calculate the implied fair odds, enter the betfair odds and compare the two to identify value using the following formula:

(betfair price -1 / implied fair odds -1) -1

to produce a value estimate which can be positive (good value and back opportunity) or negative (poor value or lay opportunity) for each correct score that has a 3% or more likelihood. I've recorded the results over around 900 matches across a variety of European leagues. Overall the best value score identified from DecTech would have produced a RoI of 28% and the same figure from Bettify data would have produced 42% RoI.