1. 11-01-11 11:19 AM #1
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,044
    Betpoints: 7292

    What would be one sigma away from 50%

    I am working on a model that tries to predict NFL totals on certain variables. In its 3rd week of testing it is currently hitting 59%. I am wondering what would be one sigma away from average 50%/50% on a distribution chart.
    Reply With Quote Reply With Quote
  2. 11-01-11 12:38 PM #2
    tukkk
    ★★★★★
    tukkk's Avatar Become A Pro!
    Join Date: 10-04-10
    Posts: 391
    Betpoints: 1

    the easiest way to get a rought estimate is to take your number of bets and square root that. this way you will get the sigma away from 50% bets won number
    this has many flaws, but will do for starters
    the other way involves edge and adding variances of all bets
    Reply With Quote Reply With Quote
  3. 11-01-11 01:28 PM #3
    TomG
    TomG's Avatar Become A Pro!
    Join Date: 10-29-07
    Posts: 500
    Betpoints: 1

    standard deviation of binomial distribution
    Reply With Quote Reply With Quote
  4. 11-01-11 10:30 PM #4
    Wrecktangle
    Wrecktangle's Avatar Become A Pro!
    Join Date: 03-01-09
    Posts: 1,524
    Betpoints: 3209

    Quote Originally Posted by Waterstpub87 View Post
    I am working on a model that tries to predict NFL totals on certain variables. In its 3rd week of testing it is currently hitting 59%. I am wondering what would be one sigma away from average 50%/50% on a distribution chart.
    You also need to know how many winners (at least). Number of losers is also nice to have so we can get your rounding error out of the "59%" figure if the number of winners happens to be a large number.
    Reply With Quote Reply With Quote
  5. 11-01-11 10:33 PM #5
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,044
    Betpoints: 7292

    Its currently a very small sample size cause I have yet to back test it against the last couple seasons. Currently it stands at 23 w 16 l
    so 23/39=.589
    Reply With Quote Reply With Quote
  6. 11-02-11 02:50 AM #6
    tukkk
    ★★★★★
    tukkk's Avatar Become A Pro!
    Join Date: 10-04-10
    Posts: 391
    Betpoints: 1

    if to use 50/50 like you mentioned, you should have won 19,5 bets out of 39 with a sigma of 6,245
    Reply With Quote Reply With Quote
  7. 11-02-11 12:17 PM #7
    Justin7
    Justin7's Avatar Become A Pro!
    Join Date: 07-31-06
    Posts: 8,577
    Betpoints: 1506

    Why don't you backtest your model?
    Reply With Quote Reply With Quote
  8. 11-02-11 01:57 PM #8
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,044
    Betpoints: 7292

    Quote Originally Posted by Justin7 View Post
    Why don't you backtest your model?
    I'm going to. I don't have time at the moment. I was copy pasting data from NFL.com and the like to get the base data and applying an equation. Its going to be a several month project.
    Reply With Quote Reply With Quote
  9. 11-02-11 02:14 PM #9
    cyberbabble
    cyberbabble's Avatar Become A Pro!
    Join Date: 08-30-10
    Posts: 772
    Betpoints: 3347

    I assume you have Excel. Stanford Wongs old website has a spreadsheet that you might try. The site is not active anymore, but the spreadsheet still works. I have no connection with him or his site.

    sharpsportsbetting.com > prop tools (left bottom) > download spreadsheet

    The sheet has a calculation for "rarity of a won/loss record".

    It is easy to use and get an idea of how big the sample size needs to be.

    A 50/50 handicapper would get your W/L record or better about 20% of the time just on randomness of the results.

    Good enough to continue, not good enough to get real excited yet.

    Good luck.
    Reply With Quote Reply With Quote
Top
« Previous Thread | Handicapper Think Tank | Next Thread »