| |
|---|---|
| 64R-41 | |
GnuBG .. d[$ W20.5 S33] 5k
GnuBG .. d[$ W5.8 S28 U31] 20k 5kSU
[For those interested, I'll also point out that the 95% CIs reported by Gnu for the two plays in the 5k rollout are .0033*1.96 = .00647, and .0037*1.96 = .00725. The 95% CI for the margin between the plays would therefore be sqrt (.00647^2 + .00725^2) = .0097. And in my 20k reroll, the 95% CI for the margin between plays is sqrt((.0018*1.96)^2 + (.0017*1.96)^2) = .00485.
With one margin having a CI of .0097 and the other .00485, the 95% CI for the difference in margins is sqrt (.0097^2 + .00485^2) = .0108. The actual difference is = (.0205 - .0058) = .0147. This degree of deviation (between 2.5 and 3 SD) borderline-defies 99.5% confidence.
That is, the chances of the $ vs W margins in the above two DMP rollouts (one 5k and the other 20k) being at least this far apart are 1/200, but such outlier results seem to happen significantly more often than that, even accounting for all the non- or less outlying margin-pairs (such as $ vs S, and W vs S here) within all DMP rollout comparisons I've observed.]
A related position is 65R-41. The RO on this site has it at
GnuBG .. [$ S11] 5k
But where is W? I would think its at least as strong with 65R-41 as with 64R-41. It's on my Gnu queue to reroll 65R-41 at least 15k with $, S and W.
What I've been leading up is: Given that you are a W fan, you might want to consider adding 64R-41 at dmp and 65R-51 at dmp to your XG queue.
Key: $ = Slot (13/9 6/5), W = Wild (24/20 6/5), S = Split (24/23 13/9), U = Up (24/23 24/20). Error sizes in rollout strings are in thousandths. A "d" before the brackets denotes dmp. After the brackets the number of trials (in thousands, rounded down) is listed. For example, "d[$ W5.8 S28 U31] 20k 5kSU" means that at dmp, $ is best and W is -.0058 after 20000+ trials (and S is -.028 and U is -.031 after 5000+ trials).
Nack
| |
|---|---|
| 64R-41 | |
| |
|---|---|
| 65R-41 | |
