Yet another Over/Under Goals System: 200 bets, +5% yield

[aldebaran]

Hello all, Having been inspired over some time be other successful systems posted to the forum (Grex, Bairn1, jtw.....) I would like to post selections from my over/under 2.5 goals system over the course of the season. This is a couple of seasons in the making and I think that there is a good likelihood that this will turn some kind of long-term profit, or at least won't lose too much! I will be backing the selections with real money but, and here is the DISCLAIMER, there is of course no guarantee of any kind of profit and losses are possible....anyone following this does so at their own risk and should only do so with money they can afford to lose etc Right, that out of the way and on to the system. I will not be revealing the actual selection criteria but will provide an overview here of the approach that I have taken in developing this. Apologies if this is a bit long-winded but I want to try to set the stall at clearly at the outset. The system is based on a fairly simple performance rating for both home and away teams. Contrary to what might seem the obvious approach, prior results and goals scored and/or conceded are not considered. Instead the idea is that the metric I have developed reflects the probability of goals being scored based on the two teams approaches to the game. The other primary criteria is the bookmakers odds. The thinking behind this is that the bookies have a pretty good model of the probability of goals being scored which underpins how they price the match. I believe that a review of the historic data backs this up. However, in any given odds range there are those matches that the bookies forecast correctly and those that they don't. This system seeks to find a criteria which can identify the subset of matches in an odds range that have a higher probability of occurring than the implied average probability for all those games in the odds range - remember, the bookies are interested in making a profit across all the markets and as such I propose that their models are fairly generalised and can afford inaccuracies. Hopefully I have developed a way to exploit this and hopefully this also makes some kind of sense! There is no fancy mathematical model behind the system and it is instead rather clunky in nature, by which I mean that I have applied the performance rating across a number of leagues and odds ranges to find if and where there are areas of consistent out-performance compared to the odds, i.e. finding value. In effect this is a manual Excel exercise in doing something probably much more clever and analytical but that is beyond my ken! I have used historic data from footballdata.co.uk and have used 5 seasons to calibrate the "model" and then a further 2 seasons to validate. I initially looked at the top 5 or 6 leagues in Europe plus all professional leagues in England and whilst there does seem to be some long term value in most of them, I have settled on the 3 leagues which have shown the strongest performance season after season and, crucially as far as I'm concerned, have demonstrated the least variability between winning and losing streaks over the calibration period. The leagues to be used are the English Premier League and Championship and Serie A in Italy. In the calibration and validation seasons none of these leagues has posted more than a single losing season and combined they have shown profit in each season. Yield does vary but on average it sits in the 10 to 12% range. I expect there to be typically 3 to 5 selections per league for each round of matches. The system starts at Round 9 of each league, once sufficient games have been played for meaningful data to be available. Finally, the odds used to develop the system are market average over/under 2.5. I will be using Pinnacle odds for the system as this is my preferred bookmaker and the odds will in most cases be somewhat better than those that the system is calibrated against. I think that is about it except that I have just remembered to also say that there is a specific group of matches in Serie A where there appears to be value on the Unders but as the 2.5 goals odds are relatively high (circa 2.4 and above) and in the interests of reducing variability I have opted to follow Under 3.5 in these cases. I will be logging selections and running yield using a theoretical starting bank of £200 and level stakes at 5% (£10) of this amount. Expect first selections for the weekend later in the week (Championship only for the next couple of weeks before the EPL and Serie A have racked up enough matches). Thanks for reading, hope some of you will find this interesting.

[Grex]

Re: Yet another Over/Under Goals system..... Very interesting, Aldebaran..... I wish you the best of luck :hope When I was running my Overs system, I also came up with preferred leagues of the English Premiership, Championship and Italy Serie A. The best performer was Serie A, with the Championship a definite under-achiever. I will follow this with interest :ok