| |
BGonline.org Forums
XG Error analysis question
Posted By: Daniel Murphy In Response To: XG Error analysis question (joe freedman)
Date: Sunday, 18 September 2011, at 3:55 p.m.
Put another way, is it really a blunder to miscalculate game winning chances by 0.32%?
Here, no -- that's why Take is not a blunder.
But why is No Double a blunder?
Because Player 1 can obtain 84.18% MWC with Double/Pass.
What can Player 1 gain by not doubling? A few sequences make Player 1 Too Good. But probably the only such sequences are 6-6, 5-5 or 4-4 followed by most misses. 5-4 pointing ... fan, for example -> Double/Pass.
What does Player 1 risk by not doubling?
- (a) He rolls 6-6 5-5 or 4-4, gets hit 41/1296 times and loses most of these
- (b) He rolls 2-1 or 3-1 and Player 2 rolls 6-6 -> No double
- (c) He rolls 4-1 played 6/2 3/2 -> No double and similarly with
- (d) 3-3 played 6/off 5/2(2) followed by most rolls -> Double/Take
In all other sequences, Player 1 will double Player 2 out.
And the sum of all that is that Player 1's MWC is, I think, about 1.5% higher after Double/Pass than after No Double.
That said, however, we need to remember that the rollout's W/G/BG breakdowns are precise for Double/Take and only an estimate for No Double. The D/T breakdown is precise because Player 2 has an automatic redouble to 8. The ND breakdown is largely an estimate, because most of the time Player 1 will double out next roll. When he does, the rollout ends. So a large proportion of the ND breakdown is comprised of the bot's evaluations at the time of D/P. How accurate are those evaluations? Try out some positions where Player 1 doubles out next roll, and see.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.