BGonline.org Forums

Probability one play is best from RO of equal plays

Posted By: Maik Stiebler
Date: Friday, 24 November 2017, at 5:04 p.m.

In Response To: Probability one play is best from RO of equal plays (Bob Koca)

What I think XG does is

- d:= equity difference of play A and play B estimated from the rollout

- jsd:= joint standard deviation of equities of play A and play B estimated from the rollout

- output [Phi^(-1)](d/jsd), where Phi is the cumulative of the standard normal distribution (or similar Student's t-distribution, shouldn't matter for large enough N)

As d/jsd is approximately standard normal distributed (again, at least for large N), the output is approximately uniformly ]0,1[-distributed. I think.

Messages In This Thread

• The Curious Case of Benjamin's Button
  Nack Ballard -- Thursday, 16 November 2017, at 9:08 a.m.
  • The Curious Case of Benjamin's Button
    scotty -- Thursday, 16 November 2017, at 12:59 p.m.
  • The Curious Case of Benjamin's Button -- SOLUTIONS
    Nack Ballard -- Sunday, 19 November 2017, at 11:46 p.m.
    • The Curious Case of Benjamin's Button -- SOLUTIONS
      mamabear -- Monday, 20 November 2017, at 12:37 a.m.
      • The Curious Case of Benjamin's Button -- SOLUTIONS
        Nack Ballard -- Monday, 20 November 2017, at 4:26 a.m.
        • The Curious Case of Benjamin's Button -- SOLUTIONS
          Philippe Michel -- Monday, 20 November 2017, at 10:22 p.m.
          • The Curious Case of Benjamin's Button -- SOLUTIONS
            Maik Stiebler -- Tuesday, 21 November 2017, at 7:31 a.m.
    • The Curious Case of Benjamin's Button -- SOLUTIONS
      Neil Robins -- Tuesday, 21 November 2017, at 8:04 p.m.
    • The Curious Case of Benjamin's Button -- SOLUTIONS
      Bob Koca -- Thursday, 23 November 2017, at 12:50 a.m.
      • The Curious Case of Benjamin's Button -- SOLUTIONS
        Nack Ballard -- Thursday, 23 November 2017, at 3:13 a.m.
        • The Curious Case of Benjamin's Button -- SOLUTIONS
          Bob Koca -- Thursday, 23 November 2017, at 2:30 p.m.
          • The Curious Case of Benjamin's Button -- SOLUTIONS
            Nack Ballard -- Thursday, 23 November 2017, at 6:40 p.m.
    • stronger RO gives expected result
      Bob Koca -- Thursday, 23 November 2017, at 3:35 p.m.
      • stronger RO gives expected result (362k trials)
        Nack Ballard -- Friday, 24 November 2017, at 1:30 a.m.
        • Another strange thing
          Bob Koca -- Friday, 24 November 2017, at 1:50 a.m.
        • stronger RO gives expected result (362k trials)
          Maik Stiebler -- Friday, 24 November 2017, at 11:14 a.m.
          • Probability one play is best from RO of equal plays
            Bob Koca -- Friday, 24 November 2017, at 4:35 p.m.
            • Probability one play is best from RO of equal plays
              Maik Stiebler -- Friday, 24 November 2017, at 5:04 p.m.
              • Oops
                Maik Stiebler -- Friday, 24 November 2017, at 5:28 p.m.
              • Probability one play is best from RO of equal plays
                Maik Stiebler -- Friday, 24 November 2017, at 6:19 p.m.
              • Probability one play is best from RO of equal plays
                Bob Koca -- Sunday, 26 November 2017, at 2:02 a.m.
      • stronger RO gives expected result
        Maik Stiebler -- Friday, 24 November 2017, at 10:43 a.m.
        • stronger RO gives expected result
          Bob Koca -- Friday, 24 November 2017, at 4:29 p.m.
          • stronger RO gives expected result
            Maik Stiebler -- Friday, 24 November 2017, at 7:15 p.m.