Jump to content
** April Poker League Result : 1st Like2Fish, 2nd McG, 3rd andybell666 **

A.I. Football ~ 27% yield


Recommended Posts

During the closed season I've been tinkering with my A.I. (Neural Nets) program to see if I could come up with a reasonably workable solution for a selection process for the footy. Well here it is Dont know if its any good but will soon find out. :loon I will be backing with real money as I do in my AI horses thread but dont recommend anyone else doing so. All bets will be 10 pts fixed stake (no fixed bank, I will decide when im losing too much :lol ) Todays selections ipswich.......... a @ 3.00 vc s/hmpton........ x @ 3.5 vc walsall............ h @2.6 bfred wycomb.......... h @ 2.5 bfred gillingham........ x @ 3.6 vc shrewsbury...... x @ 3.75 vc :ok

Link to comment
Share on other sites

  • Replies 111
  • Created
  • Last Reply

Top Posters In This Topic

Re: A.I. Football ipswich.......... a @ 3.00 vc.....won s/hmpton........ x @ 3.5 vc......lost walsall............ h @2.6 Bfred...won wycomb.......... h @ 2.5 Bfred..lost gillingham........ x @ 3.6 vc......lost shrewsbury...... x @ 3.75 vc....won not a bad start staked 60 won 3 lost 3 sr 50% return 90.35 profit 33.50 yield 50.58% return should read 93.50

Link to comment
Share on other sites

Re: A.I. Football Hi davidc I dont know how au fait you are with A.I. (neural net) but it doesn't really do back-testing. Its more of a predictive sort of function (you throw he numbers at it , it gives you an answer) end of. That's what I like about AI it doesn't dream up false prices(so called value),it works purely on the available prices and the factors that work within those available prices. I said not a bad start as its a long season and I'm not the sort of person who gets too carried away. :ok

Link to comment
Share on other sites

Re: A.I. Football

Hi davidc I dont know how au fait you are with A.I. (neural net) but it doesn't really do back-testing. Its more of a predictive sort of function (you throw he numbers at it , it gives you an answer) end of. That's what I like about AI it doesn't dream up false prices(so called value),it works purely on the available prices and the factors that work within those available prices. I said not a bad start as its a long season and I'm not the sort of person who gets too carried away. :ok
Hi, don't mind admitting I am unfamiliar with A.I. (neural net) could you possibly tell me is it your own programme or something you have purchased? Pretty impressive start though. Al
Link to comment
Share on other sites

Re: A.I. Football

Hi davidc I dont know how au fait you are with A.I. (neural net) but it doesn't really do back-testing. Its more of a predictive sort of function (you throw he numbers at it , it gives you an answer) end of. That's what I like about AI it doesn't dream up false prices(so called value),it works purely on the available prices and the factors that work within those available prices. I said not a bad start as its a long season and I'm not the sort of person who gets too carried away. :ok
Great start :cheers But if your AI, doesn´t dream up value, does it then dream up non value? I don`t get it? Good luck, I will follow your AI with interest.
Link to comment
Share on other sites

Re: A.I. Football

But if your AI, doesn´t dream up value, does it then dream up non value? I don`t get it?
Value is an opinion,computers dont have opinions they just deal with facts in the best possible way they can. So if a team is a 2/1 shot the prog doesnt dream up a price of 3/1 and call it value it works with the 2/1 and gives the best options overall compared with the rest of the teams.
are these easy to create or do you need to be a good programmer? Al
you can buy off the shelf NNets for excel quite cheap which are easy to use HERE
Link to comment
Share on other sites

Re: A.I. Football Value is not an opinion you dream up. Value is subjektive, yes, but whether you use AI, spreadsheets or pen and paper. It is you (or the proggrammer) that decides which parameters to use when calculating. We use systems, to secure we stay true (objektive) to the parameters we have chosen as important, so basically there is no difference between, AI and a system based on pen and paper. The only difference could be, that pen and paper know when it finds the value it is seeking, and the AI just point out the best possibilities? Anyway I find AI exiting and I hope you succeed.

Link to comment
Share on other sites

Re: A.I. Football It is nice somebody finally introduced neural networks (NN) to this forum. Thank you, Monkeys! I’ve been looking at NN for a while considering if it is worth an effort to learn this “modern’ technique. What I’ve found that NN employs the same classic statistical methods (regressions, Markov chains, etc.) for its “learning” and in fact it is the same “old” statistics just wrapped up in a new paper. But nevertheless it is no doubt a giant step forward in betting approach. So, Monkeys, no joke, could you please shed some light on what kind of criteria you chose for NN inputs. And it is particularly interesting if you’ve used time series analysis and if yes, what kind of data is basis for it? Anyway, will follow this tread with interest! All the best!

Link to comment
Share on other sites

Re: A.I. Football I tried using NN backpropagation (time series analysis) but as you say gives out similar to regression which is average to say the least. Opted for generic/genetic optimization which starts with the full set of factors and tests them against each other rather than pattern finding,then like natural selection weeds out the weakest eliminates it and then retests the others again renewing the process until the strongest factors are left.

Link to comment
Share on other sites

Re: A.I. Football

... Opted for generic/genetic optimization which starts with the full set of factors and tests them against each other rather than pattern finding,then like natural selection weeds out the weakest eliminates it and then retests the others again renewing the process until the strongest factors are left.
Still sounds like multivariate regression with best subsets selection method... Anyway, I wonder what kind of output you get: is it binomial kind (yes/no, win/no_win), or probability prediction (0.45 win, 0.35 lose, 0.20 draw)????
Link to comment
Share on other sites

Re: A.I. Football Ok, Monkeys, thank you for answering, your output is probability. If it is not w/d/l, then it could be anything like probability of win, goals difference or whatever. I fully understant and appreciate your unvilingness to share details. But what I dont understand is why you completely dismiss "value" in your previous posts? Because if your system predicts probability then it is inevitable to compare this "predicted probability" with "implied probability" reflected by the odds. It makes no sense for me to have a system producing probability and at the same time ignoring the concept of "value". (I am not a "value" fanatic, so please try to stay calm in your answer:)

Link to comment
Share on other sites

Re: A.I. Football Value in betting terminology is an abstract concept It can neither be proved nor disproved. Yet 'value' is too often used by the 'look at the size of my cock ive had a 16/1 winner ' or the 'in the long run i'll be profitable' fanatics to try and impress, neither of which I have any time for so, I dismiss it as early as possible in my threads so as to stop them coming on and trying to educate me in the art of the bleeding obvious ;)

Link to comment
Share on other sites

Re: A.I. Football

Ok' date=' If it is not w/d/l, then it could be anything like probability of win, goals difference or whatever.[/quote'] Did you read the link on Quarbs You have the answer already in your post on bookmakers, When you eat an orange you dont put it all in your mouth at once do you ;) . I didnt dismiss value, I said totaly different If it doesnt dream them up where must they be ?
Link to comment
Share on other sites

Re: A.I. Football ........ I tried Neural Networks a few years back, I had various models and bought various pieces of software. Unfortunately I found that what you got out at the end of the day was not worth what was fed in initially. After all there are no values or anything like that, just a set of variables fed in by whoever is running the NN. It seeks to look at patterns and remember various situations, but these are a lot of work for very little reward. I do hope you find something here though. Good luck in your attempts. Cheers. Merlin.

Link to comment
Share on other sites

Re: A.I. Football

5 for this week 50pts staked Aston Villa v Wolves x 3.4pp :clap Newcastle v Fulham x 3.4 Sky West Brom v Stoke x 3.4 Sky Accrington v Burton x 3.5 Sky Oxford v Aldershot x 3.4 Bfred :clap
staked 210 won 9 lost 12 return 301.5 profit 91.5 yield 43.57% :ok
Link to comment
Share on other sites

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.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...