| |
BGonline.org Forums
Simple bear off redouble *** Correction and Supplementary Info"
Posted By: Rick Janowski In Response To: Simple bear off redouble (Rick Janowski)
Date: Saturday, 21 April 2012, at 2:10 p.m.
Othello Itikawa kindly and correctly drew my attention to a typographical error on paragraph 3 of my posting:
“The peculiar cube situation where there can only be one redouble in the remainder of the game (this assumes the maximum cube level has been capped to the level after the redouble) . For gammonless games the redouble point probability may be approximated by the expression, RD(x.0) = 1/(2+0.667x), again where x is the cube efficiency coefficient (a value between 0 and 1, typically about 2/3).”
The formula should have read: RD(x.0) = 1/(2-0.667x). Apologies for any confusion caused.
Fortunately, this was just a typo rather than a mathematical blunder. The conclusion that redoubles which give dead cubes to the opponent are generally 7% earlier than normal money redoubles remains valid. Moreover, this generalisation has been verified as being reasonable by a number of rollouts carried out by Dmitri and me for a wide range of race lengths. Between EPCs of 40 and 130 the measured difference was within the range 6.5% and 7.2%. Outside this EPC range, the use of the 7% adjustment should generally lead to negligible errors apart from perhaps in very late bearoff positions.
The rollout exercise also considered what the early redouble would correspond to in EPC difference terms. Generally, cube capped money redoubles should occur about 3-4 pips (EPC) earlier than normal money redoubles.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.