I've never tried to refine the rule as I always know it when it comes up. As a couple of others have mentioned and Nack gave enough examples I don't have to delve into my opening material that was the same thing, make the 4pt. You will get into situations like this one where the other options are appealing, even damn appealing. They seem to accomplish many things at once and you will talk yourself into some other play. Don't do it, make the 4pt!
This is like trying to find your cubing window as the leader in a contact position at -3 -6 or more. Sure, it could happen one in a million times but even when you do luckbox into finding it you'll have made the correct decision by an insignificant amount, .020 or less. The other 999,999 you decide to stray from the rule you'll have made an error to a massive blunder as Nack showed with his examples and as I've shown with countless -3 -X cubes in the past.
                           |
| |||||||||||||||
| XGID=---d-BD-C-C-bBb--b-cb-A---:0:0:1:11:8:2:0:13:10 | ||||||||||||||||
 to play 11 | ||||||||||||||||
is Player 2| 1. | Rollout1 | 6/4(2) | eq: +0.667 | |||
| ||||||
| 2. | Rollout1 | 22/21 10/9(3) | eq: +0.661 (-0.006) | |||
| ||||||
| 3. | Rollout2 | 10/9(3) 6/5 | eq: +0.638 (-0.029) | |||
| ||||||
| 4. | Rollout3 | 10/7 8/7 | eq: +0.595 (-0.072) | |||
| ||||||
| 1 15552 Games rolled with Variance Reduction. Dice Seed: 1296 Moves and cube decisions: 3 ply 2 5184 Games rolled with Variance Reduction. Dice Seed: 1296 Moves and cube decisions: 3 ply 3 1296 Games rolled with Variance Reduction. Dice Seed: 1296 Moves and cube decisions: 3 ply | ||||||