BGonline.org Forums

Adjusted match equity

Posted By: Rich Munitz
Date: Tuesday, 16 November 2010, at 6:05 p.m.

In Response To: Adjusted match equity (Dimitar Draganov)

What you seek cannot be done.

Either a MET is calculated from scratch using base assumptions of gammon rates and game winning chances, or it is calculated via rollouts like the Rockwell-Kazaross table or G11.

You can't take a table and adjust it because those tables already have assumptions about game winning chances baked in. You must calculate a new table from scratch, using the single game win chances associated with the applicable skill difference rather than 50/50. However, if you had access to the rollout data for the RK table, you could do better. You'd know the win/G/BG rates at the current score for equal skill and could adjust that for unequal skill. That is of course imperfect since playing according to score is itself a skill-based thing. Of course using skill-adjusted single game winning chances is also imperfect because cube usage is also skill-based.

In the end, you would get a MET that would predict winning chances at all equal scores. You could then compare that with what the Elo formula predicts.

I suppose an alternate strategy would be to fill out the 0-away boundaries of the MET (i.e. all 0% and 100%). You could then fill in the n-away, n-away values as predicted by the Elo formula. The RK-MET will have 50% for all of these scores. You could then imagine linearly skewing the RK table for each of the m-away,n-away and the n-away,m-away values between the known n-away,n-away and 0-away,n-away (and n-away,0-away) values.

Messages In This Thread

• Adjusted match equity
  Dimitar Draganov -- Tuesday, 16 November 2010, at 1:30 p.m.
  • Adjusted match equity
    Fabrice Liardet -- Tuesday, 16 November 2010, at 3:34 p.m.
    • Adjusted match equity
      Dimitar Draganov -- Tuesday, 16 November 2010, at 4:01 p.m.
      • Adjusted match equity
        Fabrice Liardet -- Thursday, 18 November 2010, at 8:55 a.m.
  • Adjusted match equity
    Petter Bengtsson -- Tuesday, 16 November 2010, at 4:45 p.m.
    • Adjusted match equity
      higonefive -- Tuesday, 16 November 2010, at 5:00 p.m.
    • Adjusted match equity
      Gregg Cattanach -- Thursday, 18 November 2010, at 2:02 p.m.
      • Adjusted match equity
        Maik Stiebler -- Friday, 19 November 2010, at 11:16 p.m.
  • Adjusted match equity
    Rich Munitz -- Tuesday, 16 November 2010, at 6:05 p.m.
    • Adjusted match equity
      Dimitar Draganov -- Tuesday, 16 November 2010, at 7:13 p.m.
    • Adjusted match equity
      Fabrice Liardet -- Thursday, 18 November 2010, at 8:58 a.m.
  • Adjusted match equity
    Fabrice Liardet -- Thursday, 18 November 2010, at 2:52 p.m.