| |
BGonline.org Forums
Adjustment for GNU errors
Posted By: David Rockwell In Response To: Make sure the bot is responding properly vs 63R at score *NM* (neilkaz)
Date: Saturday, 2 January 2010, at 3:19 p.m.
This was a worthwhile exercise. Thanks for suggesting it Neil!
I will list the move GNU plays erroniously to the left of the Nacbrac. This is GS for the opener and GG for the responder. In my haste to get back to my paying job, I have been posting a lot of Nacbrac errors lately. Please point out anything that looks inaccurate. I double checked the calculation, so it should be good.
63R
63R-31H [S H18] - 5k /// 63R-51S [$ S24] - 11k /// 63R-54D [X D3.3] - 15k /// 63R-63S [D S13] - 10k /// 63R-65S [H S12] - 5k
These errors inflate the final rollout result by .0039. This is a very large error based on my experience doing these validations. I've only seen one that was larger. And using Nack's 14:20 ratio posted yesterday, this would be the largest I have seen.
63S
63S-32H [$ Z11 H11] - 5k /// 63S-51K [H K3.7] - 15k /// 63S-62D [X D0.1] - 15k
These errors inflate the final rollout result by .0008, I have made an unscientific estimate that the mean error for GNU 2 ply is .0010 for money. So, this is a normal sized error. The mode is zero of course.
So, my rollout shows:
63.s [R S5.4] - 15k
Making the adjustments for the above errors, this should be:
63.s [R S2.3] - 15k
R is still best by .0023, but certainly TCTC.
As an aside, I often hear of S being played rather than R because it complicates the game. I find in general that the responses to R are more difficult and less familiar than S. As Nack pointed out yesterday, they may seem more difficult because they are less familiar. In any event, I sometimes play R when leading a match for this reason. I like to get the gambit equity back as soon as possible.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.