v-zero Posted October 14, 2010 Share Posted October 14, 2010 I've been working on a statistically based formula for working out the probabilities of football matches after six games in a season have passed. I will be placing flat bets on the best value propositions each week per game. If a game contains no value, then there will be no bets, however most games do contain a value bet... This week's premiership probabilities: Following these probabilities I have placed one point, which is £5 with my current bank, bet on each of the following: Arsenal @ 1.32 Bolton @ 2.28 Fulham@ 3.05 Man Utd @ 1.29 Newcastle @ 1.88 Wolves @ 2.32 Chelsea @ 1.74 Everton @ 2.5 Blackpool and Man City to Draw @ 4.2 Blackburn and Sunderland to Draw @ 3.35 Other than testing with these bets, I will also be looking closely at how the results each week pan out. As far as favourite results to occur by my probabilities there should be five home wins, three away wins and two draws this weekend. I hope this will interest at least somebody, and it should help me consider the value of my work. Cheers! :cheers EDIT: If anybody is interested in probabilities for other leagues, I can adapt the system to them. Quote Link to comment Share on other sites More sharing options...
Lardonio Posted October 15, 2010 Share Posted October 15, 2010 Re: Adjusted double-poisson football, betting on value Good luck. I can't fail to notice that you've picked 80% home wins and 20% draws, which could suggest that your system is a bit biased towards home wins, which goal-based systems often are. So far we've had 48,6% home wins, 27,4% draws and 24% away wins in Premier League. Of course that does not mean you can't make a profit on your selections. Good luck! Quote Link to comment Share on other sites More sharing options...
v-zero Posted October 15, 2010 Author Share Posted October 15, 2010 Re: Adjusted double-poisson football, betting on value Good luck. I can't fail to notice that you've picked 80% home wins and 20% draws, which could suggest that your system is a bit biased towards home wins, which goal-based systems often are. So far we've had 48,6% home wins, 27,4% draws and 24% away wins in Premier League. Of course that does not mean you can't make a profit on your selections. Good luck! I'm not sure where you have your stats from, but I have 41% home wins, 23% away wins and 36% draws... This is off what my model would predict for the season so far, but I wouldn't say that is unusual. It take some time for the full-table statistics to settle down, and this weekend there happen to be a lot of strong teams playing at home, so I would expect greater than the current average of 41% home wins this weekend. My model isn't regression fit in the usual manner, as this can give skewed results in my experience. For instance, it would predict for last season that the premiership would have seen 45% home wins, 30% away wins, and 25% draws, however there were actually 51% home wins, 24% away wins and 25% draws. For the season before that it would predict 50% home wins, 24% away wins and 26% draws, whereas the actual results were 46% home wins, 29% away wins, 26% draws. It seems that statistically premiership football doesn't settle over the course of an entire season, however if you consider the two seasons and average them, the model gets 47.5% home wins where the real result was 48.5%, away wins predict at 27% where the real result was 26.5%, and draws predict at 25.5% draws where the real result was 25.5%... It seems clear that as the statistics settle we reach an accurate model, or at least I hope so. :hope EDIT: Considering the probabilities for this weekend the model would expect five home wins, two draws and three away wins, it is just the case that the home odds and draws offer the value on these matches. Quote Link to comment Share on other sites More sharing options...
v-zero Posted October 17, 2010 Author Share Posted October 17, 2010 Re: Adjusted double-poisson football, betting on value A very strange day in the premiership yesterday, but one odd weekend is not statistically significant, I'll be keeping a running tally of how the probabilities are comparing to results all season. Quote Link to comment Share on other sites More sharing options...
jimjohn Posted October 22, 2010 Share Posted October 22, 2010 Re: Adjusted double-poisson football, betting on value Hi v-zero, I'm looking at value using a simple poisson calc. The best results from what I've seen so far are on overs. I'm waiting for 6 home and away figures before betting on any so it's a few weeks away in the Prem but start this week in the other English leagues. As for 1x2 I think it might be just lays to start with, but I'll see what the results come up with. Good luck, jimjohn. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.