Page 1 of 2 1 2 LastLast
Results 1 to 20 of 24

Thread: Poisson In-Play Model

  1. #1
    Veteran jamiedavies02's Avatar
    Join Date
    Jul 2010
    Location
    Wigan
    Posts
    247
    Thanks: 37
    Thanked 49 Times in 41 Posts

    Default Poisson In-Play Model

    Hi

    I have created a spreadsheet using Sporting Indexs Total Goals and Supremacy markets and applies 'Poisson' to them to determine 'true prices' for various markets.

    I am looking to adapt this model to In-Play but need help with the following:

    - Once a goal has been scored, I can not simply input the new quotes offered by Sporting Index as my model will still return a value for impossible scores. I need to erradicate this whilst still keeping my prices to 100% and I am unsure of how to encorporate this.

    Any help appreciated.

    Also, another question, what sort of percentage should I be looking at for draw inflation?

    Thanks

  2. #2
    Forum Guru doubleback's Avatar
    Join Date
    Oct 2009
    Posts
    3,365
    Thanks: 3,294
    Thanked 4,994 Times in 2,061 Posts

    Default

    Quote Originally Posted by jamiedavies02 View Post
    I need to erradicate this whilst still keeping my prices to 100% and I am unsure of how to encorporate this.
    I assume that you have made a matrix to calculate the probabilities of different scorelines and are working in Excel.

    After a goal has been scored, I would suggest making the probability of impossible scorelines 0. This will make the total probability <1, whereas it should =1.

    Next, make another matrix, adjacent to the one you're working on and link the respective scorelines from your original matrix to your new matrix.

    Then, in the new matrix, divide the value of the scoreline probability by the total probability from your original matrix, which should be <1.

    When the totals of all the scoreline probabilities in the new matrix are added together, they will =1.


    Quote Originally Posted by jamiedavies02 View Post
    Also, another question, what sort of percentage should I be looking at for draw inflation?
    Try the diagonally inflated model ... Link
    Last edited by doubleback; 25th November 2012 at 07:08 PM.

  3. The Following 2 Users Say Thank You to doubleback For This Useful Post:
    custard (25th November 2012), Drifter (27th November 2012)
  4. #3
    Forum Guru custard's Avatar
    Join Date
    Apr 2011
    Posts
    2,980
    Thanks: 739
    Thanked 2,733 Times in 1,257 Posts

    Default

    Your link is excellent, however, I think a thorough understanding of the material contained here will lead to a far easier understanding of the complexities mentioned there.

    http://en.wikipedia.org/wiki/Poisson_distribution


    This should make things easier for those who did not study matrices in their youth.

  5. The Following User Says Thank You to custard For This Useful Post:
    Drifter (27th November 2012)
  6. #4
    Forum Guru PC0845's Avatar
    Join Date
    May 2010
    Location
    Waterford Ireland
    Posts
    830
    Thanks: 516
    Thanked 606 Times in 255 Posts

    Default

    Thanks Custard, that really simplifies it.

  7. #5
    Veteran jamiedavies02's Avatar
    Join Date
    Jul 2010
    Location
    Wigan
    Posts
    247
    Thanks: 37
    Thanked 49 Times in 41 Posts

    Default

    Quote Originally Posted by doubleback View Post
    I assume that you have made a matrix to calculate the probabilities of different scorelines and are working in Excel.

    After a goal has been scored, I would suggest making the probability of impossible scorelines 0. This will make the total probability <1, whereas it should =1.

    Next, make another matrix, adjacent to the one you're working on and link the respective scorelines from your original matrix to your new matrix.

    Then, in the new matrix, divide the value of the scoreline probability by the total probability from your original matrix, which should be <1.

    When the totals of all the scoreline probabilities in the new matrix are added together, they will =1.
    Thanks, this has helped massively.

    Any chance of a shortcut to a rough figure for draw inflation - the 2 links are way over my head!!

  8. #6
    Forum Guru custard's Avatar
    Join Date
    Apr 2011
    Posts
    2,980
    Thanks: 739
    Thanked 2,733 Times in 1,257 Posts

    Default

    Quote Originally Posted by PC0845 View Post
    Thanks Custard, that really simplifies it.
    No problem ! I can still remember feeling angry that I had to learn all about matrices when I would have no practical use for them. I had to do mathematics as a first year subject in my Chemistry degree course but, of course, that was many years ago. What possible use could there be for matrices in a steelworks ?

    I'm sure you can easily imagine how glad I am now that I was made to study them. I don't know how I could manage without them these days.

  9. The Following User Says Thank You to custard For This Useful Post:
    PC0845 (25th November 2012)
  10. #7
    Forum Guru doubleback's Avatar
    Join Date
    Oct 2009
    Posts
    3,365
    Thanks: 3,294
    Thanked 4,994 Times in 2,061 Posts

    Default

    Quote Originally Posted by jamiedavies02 View Post
    Any chance of a shortcut to a rough figure for draw inflation - the 2 links are way over my head!!
    There is not a straight forward answer, I'm afraid. When I found that paper I had the same thoughts as you. With gritted teeth, I diligently went through the paper educating myself along the way, armed with the mantra 'nothing is unknowable'. After months of devoted concentration, I attained full comprehension. Alas, it's not something I am going to share lightly. I could give a full explanation, but that would take the fun out of it. Go on a journey, my son, use perseverance and enlightenment will surely come.

  11. The Following User Says Thank You to doubleback For This Useful Post:
    lcredd (27th November 2012)
  12. #8
    Forum Guru custard's Avatar
    Join Date
    Apr 2011
    Posts
    2,980
    Thanks: 739
    Thanked 2,733 Times in 1,257 Posts

    Default

    I agree. It's something which one has to do for oneself, not just for the added knowledge it provides, but for the warm feeling one gets knowing that the model is as good as it could be. Even more, just look at the money you save by not having to pay strangers for this sort of information. Make your own model and if works you will be chuffed !

    It's thanks, in part, to my model that I didn't back the Over 2.5 goals markets in today's televised games. It didn't tell me that both games would be 0 - 0 but it was clear that there would few goals. I never back Unders but I saved losing on Overs.

  13. #9
    Forum Addict Drifter's Avatar
    Join Date
    Dec 2009
    Posts
    436
    Thanks: 338
    Thanked 360 Times in 288 Posts

    Default

    Interesting topic. Something I always wanted to look at but haven't yet. With the lack of Horse Racing in the winter this is something I can look at for the future.

    I agree that coming up with your own strategy/model will be more satisfying and hopefully more profitable. I just hope I can find one.

  14. #10
    Forum Guru custard's Avatar
    Join Date
    Apr 2011
    Posts
    2,980
    Thanks: 739
    Thanked 2,733 Times in 1,257 Posts

    Default

    I'm more than satisfied with this one



  15. #11
    Forum Guru doubleback's Avatar
    Join Date
    Oct 2009
    Posts
    3,365
    Thanks: 3,294
    Thanked 4,994 Times in 2,061 Posts

    Default

    Another, so called 'lady', that wears a wig ...

  16. #12
    Forum Guru custard's Avatar
    Join Date
    Apr 2011
    Posts
    2,980
    Thanks: 739
    Thanked 2,733 Times in 1,257 Posts

    Default

    Wigs ? You've lost me.

  17. #13
    Forum Guru doubleback's Avatar
    Join Date
    Oct 2009
    Posts
    3,365
    Thanks: 3,294
    Thanked 4,994 Times in 2,061 Posts

    Default

    It was a reference to your penchant for ladyboys, who may wear a wig, and Naomi's battle with baldness along with her masculine temperament!

  18. #14
    Forum Guru custard's Avatar
    Join Date
    Apr 2011
    Posts
    2,980
    Thanks: 739
    Thanked 2,733 Times in 1,257 Posts

    Default

    I have seen thousands of ladyboys and am certain that none has worn a wig. Only female impersonators, like Danny La Rue wear them. We knew an entertainer in Thailand who impersonated Shirley Bassey and Tina Turner and he wore wigs. He was totally straight, super fit like a kick boxer, and it was impossible to recognise him out of his stage costumes.

    No penchant on my part just a totally open mind.

  19. #15
    Veteran jamiedavies02's Avatar
    Join Date
    Jul 2010
    Location
    Wigan
    Posts
    247
    Thanks: 37
    Thanked 49 Times in 41 Posts

    Default

    Getting back on topic (well, sort of), Ihave another question;

    I have produced a model the predicts every teams points at the end of the season, how I arrive at this figure isn't important. However it needs a little tweaking, the way it works is by predicting what % of their future available points a team will acheive and adds this to the teams current points total. At the moment if you add up all the teams points it totals more points than are achievable over a season, it isn't a million miles out but I'd like to get it as realistic as possible. What sort of reduction factor could I apply?

    Would it be something like find the maximum points available, divide my total points by this and then multiply teams points by this figure?

  20. #16
    Director of Automated Solutions 1/4 of a millionaire fund's Avatar
    Join Date
    Oct 2009
    Posts
    1,034
    Thanks: 118
    Thanked 897 Times in 450 Posts

    Default

    I guess alot depends on what you want to do with the predicted points, are you betting on the spreads or just trying to form your own league table for relative positions of teams within the league. It's never going to add up perfectly because your figures are based on probabilities rather than having your table worked out by predicted results.

  21. #17
    Novice
    Join Date
    Jul 2012
    Posts
    6
    Thanks: 2
    Thanked 0 Times in 0 Posts

    Default

    Quote Originally Posted by doubleback View Post
    I assume that you have made a matrix to calculate the probabilities of different scorelines and are working in Excel.

    After a goal has been scored, I would suggest making the probability of impossible scorelines 0. This will make the total probability <1, whereas it should =1.

    Next, make another matrix, adjacent to the one you're working on and link the respective scorelines from your original matrix to your new matrix.

    Then, in the new matrix, divide the value of the scoreline probability by the total probability from your original matrix, which should be <1.

    When the totals of all the scoreline probabilities in the new matrix are added together, they will =1.




    Try the diagonally inflated model ... Link
    Hi. This link is not working. I have i question. how find value bet with new matrix after all calculations? Is it possible?

  22. #18
    Forum Guru doubleback's Avatar
    Join Date
    Oct 2009
    Posts
    3,365
    Thanks: 3,294
    Thanked 4,994 Times in 2,061 Posts

    Default

    Here is a new link to the same thing ...

    http://scholar.google.co.uk/scholar_...=0CB8QgAMoADAA

    dude67 who posts here created a nifty calculator which includes standard Poisson, bivariate Poisson and bivariate Poisson with diagonal inflation with a choice of either a Geometric distribution or a Bernoulli function.

    http://www.thebetpro.com/calculator/

  23. #19
    Novice
    Join Date
    Jul 2012
    Posts
    6
    Thanks: 2
    Thanked 0 Times in 0 Posts

    Default

    Quote Originally Posted by doubleback View Post
    Here is a new link to the same thing ...

    http://scholar.google.co.uk/scholar_...=0CB8QgAMoADAA

    dude67 who posts here created a nifty calculator which includes standard Poisson, bivariate Poisson and bivariate Poisson with diagonal inflation with a choice of either a Geometric distribution or a Bernoulli function.

    http://www.thebetpro.com/calculator/
    www.thebetpro.com is not working

  24. #20
    Forum Guru doubleback's Avatar
    Join Date
    Oct 2009
    Posts
    3,365
    Thanks: 3,294
    Thanked 4,994 Times in 2,061 Posts

    Default

    Try sending a message on here to dude67 to ask wtf is going on!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

Talented Mavericks Limited
26 Buttercup Close
Corby
Northants
England
NN18 8LB

UK Company # 04872848
UK VAT # 887366268

Facebook TwitterYouTube



Betfair    Betdaq    VISA  Mastercard  American Express  Dinners Club  Discover  PayPal  Skrill