| |
BGonline.org Forums
Question about 4 ply XG cube rollouts
Posted By: David Rockwell
Date: Thursday, 31 December 2009, at 3:34 p.m.
I am not sure I can articulate this question well. Here goes. I am counting ply as Snowie and XG count, not as GNU counts.
Doubling is inherently a 3 ply decision. You have an option to double now or to wait and consider a double again next turn. When a rollout is done with 4 ply cube, how is the fourth ply used? Is the fourth ply used to add precision to the equities from the two relevent sets of positions (1 ply and 3 ply)? Or is the value in waiting measured at the end point, 3 rolls from the original position? The former would be correct. The latter would be mathematically incorrect.
For example, imagine a position which is D/P at a particular 3-ply node. This must be valued in the rollout as 1.0. The 21 individual components of this position may have equities which are a mixture of values both less than 1.0 and equal to or greater than 1.0. If the four ply rollout summed the individual equities and used something over 1.0 in the rollout, this would be a mistake. Or if the four ply rollout added the components and got something less than 1.0 (e.g. from a mixture of equities < 1.0 and equal to 1.0), this would also be a mistake. The only correct compilation would be exactly 1.0 reflecting the cube decision at the point in time that the cube decision is actually made.
Are we confident the four ply cube rollouts are working as they should?
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.