I've long wondered if sometimes "sandbagging" a play to improve the cube efficiency might be right in some positions, and this looks like two clear examples.
Of course, they only arose after botched cube decisions and if the cube is in the wrong place then it's not a "normal" position.
I"ve included XGR++ results below for the cube decisions, and the missed cubes are whoppers. I'm wondering if there might be examples with optimal or near optimal play.
                      |
| ||||
| XGID=-BBBaCC---C-a-------ccbcb-:0:0:1:00:0:0:1:0:10 | |||||
| Blue on roll, cube action? | |||||
| Analyzed in XG Roller++ | No double | Double/Take |
| Player Winning Chances: | 80.95% (G:5.47% B:0.04%) | 81.38% (G:5.90% B:0.04%) |
| Opponent Winning Chances: | 19.05% (G:0.46% B:0.00%) | 18.62% (G:0.75% B:0.01%) |
| Cubeless Equities | +0.669 | +1.359 |
| Cubeful Equities | ||
| No double: | +0.749 (-0.251) | |
| Double/Take: | +1.249 (+0.249) | |
 Double/Pass: | +1.000 | |
| Best Cube action: Double / Pass | ||
eXtreme Gammon Version: 2.10
    |
| ||||
| XGID=-BaDCAB-BA-----a---abbcbc-:0:0:1:00:0:0:1:0:10 | |||||
| Blue on roll, cube action? | |||||
| Analyzed in XG Roller++ | No double | Double/Take |
| Player Winning Chances: | 73.85% (G:1.54% B:0.02%) | 74.40% (G:2.32% B:0.03%) |
| Opponent Winning Chances: | 26.15% (G:1.23% B:0.03%) | 25.60% (G:2.46% B:0.06%) |
| Cubeless Equities | +0.480 | +0.973 |
| Cubeful Equities | ||
| No double: | +0.698 (-0.149) | |
| Double/Take: | +0.847 | |
| Double/Pass: | +1.000 (+0.153) | |
| Best Cube action: Double / Take | ||
eXtreme Gammon Version: 2.10