"My tables assume a take point which is 60% towards the Dead TP. 30% - (30%-23%) x .6 = 26%."
That works for for a longish race for money play (78% is 60% of the way from 75% to 80%) but how much did you test it for various match play scores? In particular I think you will be dropping too easily if your recube has a low give point. For example if 3a doubles 4a the dead and live values are 35.17% and 21.07%. Your guideline gives an effective takepoint of 26.71% which is off by a lot. It seems to be only about 24.3 or 24.4% which is needed.
|
| |||
| XGID=---BCBCCAA-----aabbccba---:0:0:1:00:1:0:0:4:10 | ||||
 on roll, cube action? | ||||
| Analyzed in Rollout | |
| No Double | |
| Player Winning Chances: | 75.74% (G: 0.00% B: 0.00%) |
| Opponent Winning Chances: | 24.26% (G: 0.00% B: 0.00%) |
| Double/Take | |
| Player Winning Chances: | 75.68% (G: 0.00% B: 0.00%) |
| Opponent Winning Chances: | 24.32% (G: 0.00% B: 0.00%) |
| Cubeless Equities | |
| No Double: | +0.515 |
| Double: | +1.532 |
| Cubeful Equities | |
| No Double: | +0.835 (-0.165) |
| Double/Take: | +1.005 (+0.005) |
| Double/Drop: | +1.000 |
| Best Cube action: Double / Drop | |
| Rollout details | |
| 1296 Games rolled with Variance Reduction. Dice Seed: 4 Moves and cube decisions: 3 ply | |
| Confidence No Double: | ± 0.003 (+0.832...+0.838) |
| Confidence Double: | ± 0.005 (+1.000...+1.009) |
| Double Decision confidence: | 100.0% |
| Take Decision confidence: | 96.7% |
| Duration: 1 minute 34 seconds | |
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross
Also realize that it should be tailored to the individual position. For example as races become shorter there is less cube power. A holding game with a prime ready also has less cube power.
"The below rollout shows a GWC of 24.5% and a small drop (.039)."
No and No. Get the GWC from the DT and not the no double. 1.05 take equity means taking is a .050 error.
"However, XG 3 ply believed that the GWC was 26.22% and this would be a take. The take point seems to be 26% for this problem"
The 26.22% gives a very solid take and you only decrement it by .22% to get what you think is the minimum %age needed. If there are no gammons every extra 1% of loss chances adjusts the take equity by about .04. The 3 ply gave a take by about .078 so adjust down to about 24.3% and not to just 26%. Also you would rather look at a rollout here. The 3 ply is just sort of guessing how much the recube power will be. I made the position a little better to get a borderline choice and it suggest the takepoint for this type of position is very close to 25%.
|
| |||
| XGID=aBBBCCC---------------b---:0:0:-1:00:0:0:0:3:10 | ||||
 on roll, cube action? | ||||
| Analyzed in Rollout | |
| No Double | |
| Player Winning Chances: | 74.88% (G: 0.00% B: 0.00%) |
| Opponent Winning Chances: | 25.12% (G: 0.00% B: 0.00%) |
| Double/Take | |
| Player Winning Chances: | 74.98% (G: 0.00% B: 0.00%) |
| Opponent Winning Chances: | 25.02% (G: 0.00% B: 0.00%) |
| Cubeless Equities | |
| No Double: | +0.498 |
| Double: | +1.268 |
| Cubeful Equities | |
| No Double: | +0.925 (-0.075) |
| Double/Take: | +1.001 (+0.001) |
| Double/Drop: | +1.000 |
| Best Cube action: Double / Drop | |
| Rollout details | |
| 1296 Games rolled with Variance Reduction. Dice Seed: 4 Moves and cube decisions: 3 ply | |
| Confidence No Double: | ± 0.020 (+0.905...+0.944) |
| Confidence Double: | ± 0.026 (+0.975...+1.027) |
| Double Decision confidence: | 100.0% |
| Take Decision confidence: | 52.1% |
| Duration: 27.1 seconds | |
eXtreme Gammon Version: 1.21, MET: Rockwell-Kazaross