BGonline.org Forums

Error rate differences

Posted By: Maik Stiebler
Date: Saturday, 13 November 2010, at 8:49 a.m.

In Response To: Error rate differences (Timothy Chow)

In fact, it seems to me that we get the same conclusion using any measure of the quality of the moves [and any valid measure of luck] that ha[ve] the property that when you sum it over all the moves [and rolls] in the match, it yields 1 for the winner and 0 for the loser.

That's right, and it has some very interesting consequences. I wonder why there is no Zare article about it.

A practically relevant example would be:

- measure of luck: difference between XG-3ply rollout value of the position after and before the roll

- measure of the quality of moves: difference between XG-3ply rollout value of the position after your move and that after XG-3ply's move

As a consequence, analysis with XG 3-ply on first pass, full 3-ply rollouts of all differences, enable outplays, is very close to an unbiased measure of skill, and I say "very close" only because I am not sure that everything is implemented as needed. In particular, I suspect that cube decisions would not be handled consistently (if the actual play is no double, 3-ply's choice is double/take and 3-ply rollout says double/pass, for example, you should be dinged for the difference between the RO value of no double vs double/take, not for no double vs. double/pass).

All this stuff is known and used in the poker AI community (DIVAT). Every other year I gently try to spread the word in the BG community (it was time again), but to no avail.

Messages In This Thread

• Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
  happyjuggler0 -- Thursday, 11 November 2010, at 5:53 a.m.
  • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
    Michael Petch -- Thursday, 11 November 2010, at 6:19 a.m.
    • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
      higonefive -- Thursday, 11 November 2010, at 6:37 a.m.
      • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
        Michael Petch -- Thursday, 11 November 2010, at 6:45 a.m.
        • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
          higonefive -- Thursday, 11 November 2010, at 7:13 a.m.
          • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
            Michael Petch -- Thursday, 11 November 2010, at 7:49 a.m.
            • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
              higonefive -- Thursday, 11 November 2010, at 9:19 a.m.
              • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
                Michael Petch -- Thursday, 11 November 2010, at 9:49 a.m.
      • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
        Rich Munitz -- Thursday, 11 November 2010, at 6:47 p.m.
        • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
          Fabrice Liardet -- Friday, 12 November 2010, at 10:56 a.m.
    • American football analogy?
      Chuck Bower -- Thursday, 11 November 2010, at 6:42 p.m.
      • American football analogy?
        Rich Munitz -- Thursday, 11 November 2010, at 7:05 p.m.
      • American football analogy?
        Michael Petch -- Thursday, 11 November 2010, at 9:01 p.m.
  • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
    mtuhtan -- Thursday, 11 November 2010, at 6:24 a.m.
  • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
    Maik Stiebler -- Thursday, 11 November 2010, at 6:51 a.m.
  • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
    Christian Munk-Christensen -- Thursday, 11 November 2010, at 8:21 a.m.
    • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
      Michael Petch -- Thursday, 11 November 2010, at 9:04 a.m.
    • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
      rew -- Thursday, 11 November 2010, at 1:55 p.m.
    • the old
      garyo -- Friday, 12 November 2010, at 6:32 p.m.
  • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
    phil simborg -- Thursday, 11 November 2010, at 1:53 p.m.
  • Error rate differences
    Timothy Chow -- Thursday, 11 November 2010, at 3:25 p.m.
    • Error rate differences
      Maik Stiebler -- Thursday, 11 November 2010, at 8:37 p.m.
      • Error rate differences
        Timothy Chow -- Friday, 12 November 2010, at 1:06 a.m.
        • Error rate differences
          Chuck Bower -- Friday, 12 November 2010, at 2:51 a.m.
        • Error rate differences
          Maik Stiebler -- Friday, 12 November 2010, at 6:05 a.m.
          • Error rate differences
            Fabrice Liardet -- Friday, 12 November 2010, at 10:48 a.m.
            • Error rate differences
              Timothy Chow -- Friday, 12 November 2010, at 6:15 p.m.
              • Error rate differences
                Maik Stiebler -- Friday, 12 November 2010, at 10:32 p.m.
                • Error rate differences
                  Timothy Chow -- Friday, 12 November 2010, at 11:44 p.m.
                  • Error rate differences
                    Maik Stiebler -- Saturday, 13 November 2010, at 8:49 a.m.
            • Error rate differences
              Maik Stiebler -- Friday, 12 November 2010, at 10:42 p.m.
              • Error rate differences
                Fabrice Liardet -- Sunday, 14 November 2010, at 7:35 p.m.
                • Error rate differences
                  Maik Stiebler -- Sunday, 14 November 2010, at 7:46 p.m.
  • Who is the better BG player: the one who has the lowest error rate, or the one who somehow maximizes the difference between his errror rate and that of his opponent?
    Steve Mellen -- Thursday, 11 November 2010, at 10:00 p.m.