• Repeats and Dodgy Statistics (Hi MJ!)

    From nigel@21:1/5 to Marmaduke Jinks on Mon Dec 11 13:25:36 2023
    The correlation betweem my system's predictions for number of repeats
    and the actual results has now hit -0.20389041942593, and that's after
    44 draws.

    I 'cheated' and used my software's TTest function with the following parameters:
    Array of predictions
    Array of results
    2-tailed test (since we're looking for oddities in either direction)
    Pairs (since each prediction and result forms a discrete pair)

    It calculated the probability as 0.64430697390173.

    I'm not sure what that means, but my best guess is that it reckons
    there's a 64% chance that something untoward is going on ie not
    statistically significant.

    Nigel



    Marmaduke Jinks wrote:

    Congrats for keeping it up ;-)

    Did you learn much? Did you find a bias?

    MJ
    (I have post here, can't post further down the thread)



    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From nigel@21:1/5 to Ion Saliu on Wed Dec 13 15:31:43 2023
    Ion Saliu wrote:

    <snip>

    I second that. Bravo, Doollow! Keep it going, axiomates, Doollow & Martsopolleahk! Being alive is of utmost importance — second only to
    good health.


    Thank you for the kind words - more than I deserve after all my teasing.
    You're a better man than I am, Gunga Din!

    I'm like a senescent ugly duckling, paddling away below the surface ever
    more slowly, in the hope it's not too late to become a swan.

    I noticed something strange a while back but wrote it off as a
    coincidence. Then something similar happened, then again and again. I've started monitoring it formally because if it holds up, it's weird,
    wonderful and mind-blowing. I've been here before only to have my hopes
    dashed but it's nice to have a bit of optimism to cling to in these
    troubled times.

    I wish you and your family a prosperous New Year.

    Nigel

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Marmaduke Jinks@21:1/5 to All on Thu Dec 14 08:30:42 2023
    "nigel" <[email protected]> wrote in message news:[email protected]...


    Ion Saliu wrote:

    <snip>

    I second that. Bravo, Doollow! Keep it going, axiomates, Doollow &
    Martsopolleahk! Being alive is of utmost importance - second only to
    good health.


    Thank you for the kind words - more than I deserve after all my teasing. You're a better man than I am, Gunga Din!

    I'm like a senescent ugly duckling, paddling away below the surface ever
    more slowly, in the hope it's not too late to become a swan.

    I noticed something strange a while back but wrote it off as a
    coincidence. Then something similar happened, then again and again. I've started monitoring it formally because if it holds up, it's weird,
    wonderful and mind-blowing. I've been here before only to have my hopes dashed but it's nice to have a bit of optimism to cling to in these
    troubled times.

    I wish you and your family a prosperous New Year.

    Nigel

    Festive Greetings to all in the group :-)

    MJ

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From nigel@21:1/5 to nigel on Tue Dec 19 12:50:19 2023
    Did I say dodgy? Most of my software's built-in functions ignore blank
    rows. It seems the TTest function is different. So the result I posted
    below is wrong, though fortunately not by much.

    nigel wrote:

    The correlation betweem my system's predictions for number of repeats
    and the actual results has now hit -0.20389041942593, and that's after
    44 draws.

    I 'cheated' and used my software's TTest function with the following parameters:
    Array of predictions
    Array of results
    2-tailed test (since we're looking for oddities in either direction)
    Pairs (since each prediction and result forms a discrete pair)

    It calculated the probability as 0.64430697390173.

    I'm not sure what that means, but my best guess is that it reckons
    there's a 64% chance that something untoward is going on ie not
    statistically significant.

    Nigel

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From nigel@21:1/5 to nigel on Sun Dec 24 15:59:32 2023
    Yesterday my syste predicted 2 repeats in the New York Lottery and we
    got two repeats. That didn't do my dodgy statistics much good :-)

    Happy Christmas to those who care, Happy Holidays to those who don't.

    nigel wrote:

    Did I say dodgy? Most of my software's built-in functions ignore blank
    rows. It seems the TTest function is different. So the result I posted
    below is wrong, though fortunately not by much.

    nigel wrote:

    The correlation betweem my system's predictions for number of repeats
    and the actual results has now hit -0.20389041942593, and that's after
    44 draws.

    I 'cheated' and used my software's TTest function with the following
    parameters:
    Array of predictions
    Array of results
    2-tailed test (since we're looking for oddities in either direction)
    Pairs (since each prediction and result forms a discrete pair)

    It calculated the probability as 0.64430697390173.

    I'm not sure what that means, but my best guess is that it reckons
    there's a 64% chance that something untoward is going on ie not
    statistically significant.

    Nigel



    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)