After looking at XGR++ I thought I was on to what Stick was talking about. i.e. at 3a Na the doubling window is very very small and you go from not good enough to too good in the blink of an eye. Here's what XGR++ says:
Analyzed in XG Roller++
Player Winning Chances: 77.26% (G:56.15% B:12.82%)
Opponent Winning Chances: 22.74% (G:1.68% B:0.10%)Cubeless Equities: No Double=+1.098, Double=+1.459 Cubeful Equities:
No double: +0.990 (-0.010)
Double/Take: +1.020 (+0.020)
Double/Pass: +1.000 Best Cube action: Double / Pass
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
So, according to XGR++ we're right in the middle of this vanishingly small doubling window. Full rollout tells another tale:
Even though blue has an overwhelming superiority (~75% GWC 55% gammons) the potential re-whip makes it too dangerous to ship the cube here. All white has to do is recube and those gammon wins are meaningless. I think it's easy to see this as a hold; it's not so easy to see it as a take, but Neil explained it well (even if I had to read his post three times before getting it).
                        |
| ||||
| XGID=-a---BC-CB--ca---aac-bbaBC:0:0:-1:00:0:8:0:11:10 | |||||
| White on roll, cube action? | |||||
| Analyzed in Rollout | No double | Double/Take |
| Player Winning Chances: | 75.78% (G:56.17% B:18.09%) | 75.03% (G:56.31% B:37.26%) |
| Opponent Winning Chances: | 24.22% (G:1.71% B:0.13%) | 24.97% (G:1.73% B:0.10%) |
| Cubeless Equities | +1.120 | +1.380 |
| Cubeful Equities | ||
| No double: | +0.946 | ±0.016 (+0.930..+0.962) |
| Double/Take: | +0.848 (-0.098) | ±0.041 (+0.807..+0.889) |
 Double/Pass: | +1.000 (+0.054) | |
| Best Cube action: No double / Take | ||
| Percentage of wrong pass needed to make the double decision right: 39.2% | ||
| Rollout details | ||
| 1296 Games rolled with Variance Reduction. Moves: 3-ply, cube decisions: XG Roller | ||
| Double Decision confidence: | 100.0% | |
| Take Decision confidence: | 100.0% | |
| Duration: 1 hour 21 minutes | ||
eXtreme Gammon Version: 2.10, MET: Kazaross XG2