1. #1
    genyes
    genyes's Avatar Become A Pro!
    Join Date: 01-14-06
    Posts: 75

    Z Factor and money lines

    Can the Z Factor be calculated for a set of results based on differing money line wagers?

    If not, how can a set of money-line results be evaluated for statistical significance? Thanks.

  2. #2
    Justin7
    Justin7's Avatar Become A Pro!
    Join Date: 07-31-06
    Posts: 8,577
    Betpoints: 1506

    You can compute standard deviations the normal way with moneylines. Use standard mean squared... But you can't simplify it like you can on 50/50 bets.

  3. #3
    genyes
    genyes's Avatar Become A Pro!
    Join Date: 01-14-06
    Posts: 75

    Quote Originally Posted by Justin7 View Post
    You can compute standard deviations the normal way with moneylines. Use standard mean squared... But you can't simplify it like you can on 50/50 bets.
    Thanks for the response. But my understanding of statistics is limited. Is there a formula that can be used to determine the statistical significance of, for example, the following MLB results?:

    913 wins 1247 losses for a net profit of +28 units

    Thanks.

  4. #4
    Wheell
    Wheell's Avatar Become A Pro!
    Join Date: 01-11-07
    Posts: 1,380
    Betpoints: 1022

    Statistically the standard deviation for that sample is about 23 wins. The player is about 12 wins ahead of the expected value. In other words, statistical significance is a ways away.

  5. #5
    Data
    Data's Avatar Become A Pro!
    Join Date: 11-27-07
    Posts: 2,236

    Wheel, these are not 50/50 bets.

    genyes, see this thread:
    http://www.sportsbookreview.com/forum/handicappe...ce-needed.html

  6. #6
    Wheell
    Wheell's Avatar Become A Pro!
    Join Date: 01-11-07
    Posts: 1,380
    Betpoints: 1022

    really data, he went 913 and 1247 and ended up 28 units and these AREN'T 50-50 bets? BS, I don't believe you.

  7. #7
    Data
    Data's Avatar Become A Pro!
    Join Date: 11-27-07
    Posts: 2,236

    Wheell, your calculations are based on two wrong assumptions both of which are true for 50/50 no-vig bets:
    1) the lines are the same for all the bets,
    2) the squared deviations from the mean always equal 1.

  8. #8
    Wheell
    Wheell's Avatar Become A Pro!
    Join Date: 01-11-07
    Posts: 1,380
    Betpoints: 1022

    Data: I know that the bets are varied. They pretty much have to be given the information we have. as for the squared deviations, my answer was an approximate. The fact is that while his results were good (+28 is better than -28), his results are not statistically significant to the point where he can draw any conclusions from them. Assuming all bets are the same size makes it MORE likely that his results are significant. the more the bets vary in size, the more likely it is that he has a fluke result. Admittedly, the fact that he is betting on dogs also increases variance.

  9. #9
    Wheell
    Wheell's Avatar Become A Pro!
    Join Date: 01-11-07
    Posts: 1,380
    Betpoints: 1022

    Or to put it in simpler terms, the quick math can give a false positive, but no false negatives, ergo when it comes back negative no more analysis is needed.

  10. #10
    genyes
    genyes's Avatar Become A Pro!
    Join Date: 01-14-06
    Posts: 75

    Quote Originally Posted by Wheell View Post
    Statistically the standard deviation for that sample is about 23 wins. The player is about 12 wins ahead of the expected value. In other words, statistical significance is a ways away.
    I think we're getting there. But guys please keep it simple for me. First of all, the unstated assumption is that all bets (winning and losing) were made for the same amount of money and all of the plays were dogs of varying prices.

    Two questions Wheel: What is the formula for figuring standard deviation and what is the formula for figuring expected value?

    I'm going to go to the suggested page on statistics and check it out.

    But if you could lay out for me --if possible--the formulas for answering my original question I would appreciate it.

  11. #11
    Data
    Data's Avatar Become A Pro!
    Join Date: 11-27-07
    Posts: 2,236

    Quote Originally Posted by Wheell View Post
    Data: I know that the bets are varied. They pretty much have to be given the information we have. as for the squared deviations, my answer was an approximate. The fact is that while his results were good (+28 is better than -28), his results are not statistically significant to the point where he can draw any conclusions from them. Assuming all bets are the same size makes it MORE likely that his results are significant. the more the bets vary in size, the more likely it is that he has a fluke result.
    True.

    Admittedly, the fact that he is betting on dogs also increases variance.
    Exactly, if we assume 1 unit bets and identical lines on all bets, it comes down to this:
    st.dev.=square root((913+1247)*(1247+28)/913)=55
    Last edited by Data; 08-21-08 at 03:09 PM.

  12. #12
    reno cool
    the meaning of harm
    reno cool's Avatar Become A Pro!
    Join Date: 07-02-08
    Posts: 3,567

    Quote Originally Posted by Wheell View Post
    Statistically the standard deviation for that sample is about 23 wins. The player is about 12 wins ahead of the expected value. In other words, statistical significance is a ways away.
    Why is he only 12 wins ahead of exp value? Isnt 1 unit=1game
    what about vig?
    If he can just beat the vig after 2000 games that seems significant. What am I missing?

  13. #13
    Data
    Data's Avatar Become A Pro!
    Join Date: 11-27-07
    Posts: 2,236

    Quote Originally Posted by reno cool View Post
    Why is he only 12 wins ahead of exp value? Isnt 1 unit=1game
    Had he won 901 and lost 1259 betting at +139.65 he would have almost broken even.

    what about vig?
    No vig assumed.

    If he can just beat the vig after 2000 games that seems significant. What am I missing?
    As it was pointed out to you earlier (not in this thread) many words used in this forum are not just words but statistical terms. "Significant" is one of those words:
    http://en.wikipedia.org/wiki/Statistical_significance
    Last edited by Data; 08-21-08 at 04:05 PM.

  14. #14
    reno cool
    the meaning of harm
    reno cool's Avatar Become A Pro!
    Join Date: 07-02-08
    Posts: 3,567

    aren't there different degrees of significance?

    2000 games beating the vig at those odds, wouldn't mean anything?

    how can the vig not be involved in sample given?

  15. #15
    Rufus
    Rufus's Avatar Become A Pro!
    Join Date: 03-28-08
    Posts: 107
    Betpoints: 384

    Of course, but you won't find ANYTHING significant within one standard deviation of expectation. A commonly used level of significance is 5%, which corresponds to roughly two standard deviations (1.96 actually) of a normally distributed variable.

Top