There are some opening game whores who DO waste their time memorizing and learning things they are never going to see. Some even waste their time on ATS replies for these openings.
Unimaginable. :)
Early game study is paying off, David. You did impressively well with your answers.
I'll include a diagram for each reply, put your comments in italics and my return comments in non-italics. Rollout results will be indented, with an explanation of nacbrac notation (for the general readership) at the end of the post.
32Z replies [money]:
| |
|---|---|
| 32Z-21 | |
Support of H (Hit, 6/3*) or S (Split, 24/23 13/11) depends on the bot, but it's very close in any case:
- Snowie .. [S H9 V40 U54] 31k 10kV 5kU
GnuBG .. [H S7] 46k
XG 3-ply [S H3] 62k
| |
|---|---|
| 32Z-32 | |
By "agg" (bot aggregate), the related 52S-32 position (with Opp having a fourth 8pt checker instead of one on the 10pt) is [H S13 K15 X21], whereas 32Z-32 is [K X8 H12 S49], broken down by bot below. K (Kill, 6/3*/1) becomes better than H (Hit, 13/11 6/3*) because the game plan of keeping Opp as off balance as possible decreases the chances of her making a point with her 10pt checker in the near future, and it increases the chances of hitting that 10pt checker in the ensuing scramble (which can take the form of a blitz, semi-blitz or multiple hit exchanges).
- Snowie .. [K X11 H21 S49 Z84] 31k 5kSZ
GnuBG .. [K X3 H6] 46k
XG 3-ply [K H8 X9] 62k
| |
|---|---|
| 32Z-41 | |
True. U (Up, 24/23 24/20) has rolled and still does roll out best. You are also correct that the sexy-looking S (Split, 24/23 13/11) and H (Hit, 8/3*) are highly competitive -- within .01. A fourth play, T (sTack, 13/8) isn't far behind. Agg is [U S9 H10 T14]; individual bots are:
- Snowie .. [U S13 T17 H18] 31k
GnuBG .. [U S1 H3 T11] 46k 15kT
XG 3-ply [U H10 T12 S14] 62k
| |
|---|---|
| 32Z-43 | |
Right on target. By agg, 52S-43 is [H X4], whereas 32Z-43 is [X H10], showing a gain of .014 for X (hit-and-split, 24/20 6/3*) versus H (hit, 13/9 6/3*). The 3 duplication is a contributing factor, but it's largely about the value of threatening an outfield blot. By bot:
- Snowie .. [X H12] 31k
GnuBG .. [X H8] 46k
XG 3-ply [X H10] 62k
| |
|---|---|
| 32Z-51 | |
Yup, Opp's 10pt builder is remarkably friendly to placing a checker on the 18pt. It's too close to call, and XG hasn't weighed in, but at this point the nod by a whisker goes to R (Run, 24/18) over S (Split, 24/23 13/8):
- Snowie .. [R S1 X35] 31k
GnuBG .. [R=S X18] 46k 6kX
| |
|---|---|
| 32Z-52 | |
Yes, K (Kill, 6/3*/1*) does seem distributionally imbalanced, as do X (hit-and-split, 24/22 8/3*) and H (Hit, 13/11 8/3*), but by agg they are only .004, .008 and .017 behind S (Split, 24/22 13/8), the move you chose.
- Snowie .. [S X5 K7 H22 D35 x47] 20k 10kH 5kDx
GnuBG .. [S K1 X12 H13 D26 x38] 46k 15kXH 5kDx
| |
|---|---|
| 32Z-65 | |
That's correct, unless you go out of your way to also count 32U-65 and 31S-65. :) You didn't fall for my trap. Agg for 32Z-65 is [R S9]. Still, by comparing that to 32S-65 [R S67], where S (Split, 24/18 13/8) is a blunder instead of being within .01 of R (Run, 24/13), one can dramatically see the huge effect of the 10pt builder.
- Snowie .. [R S10] 31k
GnuBG .. [R S8] 15k
I play 32Z just for fun every once in a while. The problem with it is that even though few players know how to respond, so many rolls have multiple reasonable choices which are very close. There's no real hope of getting the gambit equity back even with gifts like 64R which I received the last time I gave it a try. And, those four outfield shots are killers.
To be clear, it's four vs two, a difference of only two outfield shots, assuming one compares opening 32Z to 32S. But I'm in agreement. 32Z-64R is only -.032 but 64 is rolled only one time in 18, for an average recoup of only .002. Even if Opp plays 32Z-64P (which has been opined), the recoup is less than .004. Given the diversity of close plays, Opp would have to be very imaginative with other rolls to reimburse the entire .015 you're gambiting.
For the balance of the otherwise unwarranted gambit, you'd have to justify opening 32Z in some other way, such as throwing Opp off his game; i.e., he might decide to adjust his normal checker play or cube somehow in order to best take advantage of a player (you) whose style is unorthodox enough to sport opening 32Z -- and therefore who knows what might be next -- you might (from his point of view) even be drunk/drugged or on tilt. Remember that Opp might believe you're theoretically giving away a lot more than .015. This is a potential benefit you won't attain, for example, with commonly played openings of 41$, 54D, 52D, 32D and 51$, which theoretically sacrifice a similar amount of equity (more than .010).
***************************************************
Nactation: D = Down, H = Hit, K = Kill (hit twice), P = Point, R = Run, S = Split, T = sTack/Tower, U = Up, V = Variant-Up, X = hit-and-split, Z = reverse split. The best plays for each position are also translated into specific traditional notation within each segment.
Nacbracs: Error sizes are in thousandths. Number of trials ("k") appear after the brackets and are in thousands, rounded down. For example, "[R S10] 31k" means that R is best and S is -.010 after 31000+ trials.
"Agg" is an abbreviation for "bot aggregate" (an average of rollout results from at least two but usually three bots).
Nack