I adjusted the position a little bit to one where I could get EPC directly from the databases. One can ask GNU for a one sided RO but it seems the bearin technique is faulty and they can be large by even more than a pip.
I pushed both sides up 4 pips and the win chance stayed about the same.
Here blue has a somewhat low variability position and a win chance of 72.4% (original position Igor skillfully estimated is 72.5% and XG++ gave as 72.500% I guess Igor wasn't confident to the three decimal points). Blue's wastage is 13.7 which is the same as the reference position I used. Here the 6's play a lot better but the 5's play worse. The wastage is probably about the same as in Phil's position so looks like I overestimated the EPC by about 1 pip.
                       |
| ||||
| XGID=-CBC-B-BA---B---bb-ccbba--:0:0:1:00:0:0:3:0:10 | |||||
| Blue on roll, cube action? | |||||
| Analyzed in Rollout | No double | Double/Take |
| Player Winning Chances: | 72.429% (G:0.000% B:0.000%) | 72.381% (G:0.000% B:0.000%) |
| Opponent Winning Chances: | 27.571% (G:0.000% B:0.000%) | 27.619% (G:0.000% B:0.000%) |
| Cubeless Equities | +0.4486 | +0.8952 |
| Cubeful Equities | ||
| No double: | +0.7006 (-0.0205) | ±0.0034 (+0.6972..+0.7040) |
| Double/Take: | +0.7210 | ±0.0041 (+0.7169..+0.7251) |
| Double/Pass: | +1.0000 (+0.2790) | |
| Best Cube action: Double / Take | ||
| Rollout details | ||
| 1296 Games rolled with Variance Reduction. Dice Seed: 60367227 Moves: 3-ply, cube decisions: XG Roller | ||
| Double Decision confidence: | 100.0% | |
| Take Decision confidence: | 100.0% | |
| Duration: 1 minute 09 seconds | ||
eXtreme Gammon Version: 2.10
This next position has an EPC that is different by only .04 (which is worth about 0.1% win chance normally) but the win chance is down all the way to 70.6%. The other 1.7% is due to the variability being greater. The correct adjustment should have been 1 pip and not 0.5 pips.
  |
| ||||
| XGID=---CBCEA-A------bb-ccbba--:0:0:1:00:0:0:3:0:10 | |||||
| Blue on roll, cube action? | |||||
| Analyzed in Rollout | No double | Double/Take |
| Player Winning Chances: | 70.680% (G:0.000% B:0.000%) | 70.649% (G:0.000% B:0.000%) |
| Opponent Winning Chances: | 29.320% (G:0.000% B:0.000%) | 29.351% (G:0.000% B:0.000%) |
| Cubeless Equities | +0.4136 | +0.8260 |
| Cubeful Equities | ||
| No double: | +0.6385 | ±0.0037 (+0.6347..+0.6422) |
| Double/Take: | +0.6330 (-0.0055) | ±0.0044 (+0.6285..+0.6374) |
| Double/Pass: | +1.0000 (+0.3615) | |
| Best Cube action: No double / Take | ||
| Percentage of wrong pass needed to make the double decision right: 1.5% | ||
| Rollout details | ||
| 1296 Games rolled with Variance Reduction. Dice Seed: 60367227 Moves: 3-ply, cube decisions: XG Roller | ||
| Double Decision confidence: | 96.9% | |
| Take Decision confidence: | 100.0% | |
| Duration: 1 minute 07 seconds | ||
eXtreme Gammon Version: 2.10
Overall, for Phil's position I should have been seeing the lead as 6.5 pips instead of 5 pips.