Great analysis, Ian.
You asked about the related positions in which Blue has made a different inside point. For two of them, you stated that the GnuBG evals are: 42P-61P-41 [S D36] \e, and 31P-61P-41 [S D25] \e. Where you wrote "D," I'm guessing you meant "U," but for the second result either Gnu 4-ply eval is way off or you accidentally reversed the order of the two plays.
I have a short rollout of 31P-61P-41 [E U17 N25 S40] /5**3 (from Stick), which puts U ahead of S by .023. I do not have (or can't find) rollouts for the other three positions.
Isolating just the S (Split, 24/23 13/9) and U (Up, 24/23 24/20) plays, we have:
- 64P-61P-41 [S U15] "&e
53P-61P-41 [S U6] "&e -- [U S1] "<=75, [U S13] "<10
42P-61P-41 [S U40] "&e
31P-61P-41 [U S11] "&e -- [U S23] /3
We can see from the rollout data (from the two of the positions we have it for) that XGR++ undervalues U relative to S, but in order to compare apples to apples, I'll key off the XGR++ evaluations (in black).
Most of the apparent inflation of the 42P-61P-41 S/U margin can be explained by the fact that making the 4pt blocks 53 from hitting. If White could hit, her (XGR++ eval) equity would be +.160, whereas her best legal play (53D) has an equity of -.266. That's a swing of .426, divide by 18 creates a prophylactic value of nearly .024. (Also, owning the 4pt enhances the swing on double 3s which fails to POH on Blue's 20pt if he hangs back, though that effect works out to be less than another .004.)
Finally, note that Blue's opening 31P has blocked 52 (a move later), and his 53P has likewise blocked 52, and playing 41U takes full advantage of White's unsplittable 4 or 2. Playing 41U is not as efficient in conjunction with 64P and 42P, because White is happy to use her (blocked) 51 or 53 to hit on her 5pt.
For reference, the four positions under discussion are diagrammed below.
|
| ||||||||||||||||
|
| ||||||||||||||||