1. #1
    Checkerboard
    Checkerboard's Avatar SBR PRO
    Join Date: 05-15-06
    Posts: 7,791
    Betpoints: 14361

    How Much Clock Does it Take to Back a Theory's Winning Percentage?

    Sample Theory: Team loses outright at home with negative three turnovers or worse. Take any points as home dog following week.

    The thinking is the team will focus on ball protection in the
    second week, especially at home (amends). Also, with a
    minimum -3 turnover ratio the previous week, the score will
    have been misleading as to this team's capabilities when not turning over. Public opinion going against this team would,
    in theory, lead to pointspread value.

    Now that I have a theory, I'll chart it's performance going
    back in time. If it turns up a decent winning percentage, I'll
    start playing it every time this scenario transpires from then
    on.

    My question is this:

    How Much Clock Does it Take to Back a Theory's Winning Percentage?

  2. #2
    Cyclone
    Cyclone's Avatar Become A Pro!
    Join Date: 07-20-06
    Posts: 141
    Betpoints: 18

    In my experience, you have to check at least 50 games to see if any particular theory will work.

  3. #3
    Dark Horse
    Deus Ex Machina
    Dark Horse's Avatar Become A Pro!
    Join Date: 12-14-05
    Posts: 13,764

    You're talking about sample size. Everybody can come up with a 8-1 trend, but nobody would expect it to go 80-10.

    This problem can be solved by using the Z-factor. Take the wins minus the losses, and divide by the square root of the sample size. If the Z-factor is above 2.5, you're in business.

    Take the above 8-1 example.
    8-1, divided by square root of 8+1. That is 7 divided by 3, or 2.5. (Note that 8-2 would produce 6 divided by square root of 10 = 1.90, so with this sample size things are still shaky).

    How about a 60-30 record?
    Z-factor is 60-30, divided by square root of 90. This is 3.16.

    One more. 234-166 wins and losses. Z-factor is 68 divided by square root of 400 = 3.4.

    The underlying idea is that a Z-factor of 2 has a 5% chance of being the result of luck (as in being totally random). A Z-factor of 3 only has a 1% chance of that.

    Hope that helps.

  4. #4
    MercJr
    MercJr's Avatar Become A Pro!
    Join Date: 09-05-06
    Posts: 26

    good info Dark Horse, thanks.

  5. #5
    Checkerboard
    Checkerboard's Avatar SBR PRO
    Join Date: 05-15-06
    Posts: 7,791
    Betpoints: 14361

    Thanks for this equation Dark Horse.

  6. #6
    Justin7
    Justin7's Avatar Become A Pro!
    Join Date: 07-31-06
    Posts: 8,577
    Betpoints: 1506

    There are a lot of winning angles based on turnovers. Many are valid due to one common reason: turnovers occur randomly (there is less than a 20% correlation in turnover predictability. You can almost assume they are all random).

  7. #7
    Checkerboard
    Checkerboard's Avatar SBR PRO
    Join Date: 05-15-06
    Posts: 7,791
    Betpoints: 14361

    Sample Theory at Top of Thread is Hypothetical . . .

    Quote Originally Posted by Justin7
    There are a lot of winning angles based on turnovers. Many are valid due to one common reason: turnovers occur randomly (there is less than a 20% correlation in turnover predictability. You can almost assume they are all random).
    Thanks Justin7 . . . The theory at the top of the thread's
    hypothetical . . . I'm only using it as an example of a potential
    theory. Once one has any theory though, how much clock time do you think they ought to run it through before they put stock in it? Your thoughts would be appreciated.

    For all interested in this topic, I found Dark Horse's url references above worth checking out . . .

Top