BGonline.org Forums

2 Rollouts: Hit or play safe?

Posted By: Maik Stiebler
Date: Thursday, 9 December 2010, at 3:45 p.m.

In Response To: 2 Rollouts: Hit or play safe? (Timothy Chow)

Does the Jacoby rule make any difference to the centered-cube checker play?

As Casper wrote, the Jacoby rule makes a difference. But according to my analysis it's not quite enough to make 8/6 8/5 the better play with centered cube without Jacoby.

(Warning, unstructered thoughts follow:) Obviously, the reason for the hit without Jacoby is that it increases the value of cube access. I tried to find that increased value based on the fact that if there were a linear mapping from equities with opponent holding the cube (OC) to equities with centered cube (CC), the order of the moves could not change. Thus I did a linear regression with the equities after all replies after both 8/6 8/5 and 18/16* 18/15 to find an approximate linear mapping between OC and CC equities, in order to be able to point to those replies where the deviation from the approximation is largest. Unfortunately I still can't make sense of all of this :-). Anyway, the result was that the opponent's dances are quite cube valuable, even though they are market losers. Interestingly, the replies that hurt most (2s after the hit), deviate in the same direction, which is a bit hard to rationalize. Basically it looks like a convex function mapping OC to CC would be a better approximation to the data than a linear function. The typical moderate equities after 8/6 8/5 lie deep in the belly of that convex function (though maybe there also is a positional, rather than technical, reason, why the cube is relatively less valuable in these, but the equities after the moderate replies to the hit also seem to be affected by the convexity to some degree). I was reminded of a thing Douglas Zare wrote in his second article on the Janowski formulas on Gammonvillage: With access to the cube, it is often right to make a larger variance play, as long as it doesn't lead to large market losers. The explanation that Zare gives is not quite applicable to the situation here, though.

Messages In This Thread

• Hit or play safe?
  Flaccus -- Tuesday, 7 December 2010, at 7:45 p.m.
  • Hit or play safe?
    ah_clem -- Tuesday, 7 December 2010, at 9:14 p.m.
    • Hit or play safe?
      Sebastiaan Kuijs -- Thursday, 9 December 2010, at 8:53 a.m.
  • Hit or play safe?
    Michael Nielsen -- Tuesday, 7 December 2010, at 9:31 p.m.
  • Hit or play safe?
    Timothy Chow -- Tuesday, 7 December 2010, at 10:01 p.m.
  • 2 Rollouts: Hit or play safe?
    Flaccus -- Wednesday, 8 December 2010, at 4:49 p.m.
    • 2 Rollouts: Hit or play safe?
      neilkaz -- Wednesday, 8 December 2010, at 5:08 p.m.
    • 2 Rollouts: Hit or play safe?
      Timothy Chow -- Wednesday, 8 December 2010, at 7:01 p.m.
      • 2 Rollouts: Hit or play safe?
        Maik Stiebler -- Thursday, 9 December 2010, at 3:45 p.m.
    • 2 Rollouts: Hit or play safe?
      Casper van der Tak -- Wednesday, 8 December 2010, at 11:33 p.m.