Like John says, when the race is close it's usually right to keep contact. The issue we have is what's close and what's enough to break contact? It's simple to estimate our win % here if we virtually break contact by playing 9, 8. We have a 5 pip lead but are not on roll. It's accepted that being on roll is worth 4 pips. Each pip in a straight race is worth roughly 2% so all that talk would lead us to thinking by clearing the 12pt we'd be ~52% favorites. We see per XG that if we keep contact we win 55.5% here. Does that mean if we were up two more pips we should break contact and if so what kind of general rule can we come up with for these situations?
        
 
          
 | | White is Player 2
score: 0
pip: 107 | Unlimited Game
Jacoby Beaver | pip: 109
score: 0
Blue is Player 1 |
|
| XGID=---A-BEbCB--B----acccaaa--:0:0:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | Rollout1 | 8/4 6/3 | eq: +0.165 |
| Player:
Opponent: | 55.49% (G:0.99% B:0.03%)
44.51% (G:1.08% B:0.03%) | Conf.: ± 0.008 (+0.157...+0.172) - [100.0%]
Duration: 20.2 seconds |
|
| 2. | Rollout1 | 12/9 12/8 | eq: +0.073 (-0.091) |
| Player:
Opponent: | 52.20% (G:0.05% B:0.00%)
47.80% (G:0.05% B:0.00%) | Conf.: ± 0.008 (+0.065...+0.081) - [0.0%]
Duration: 9.5 seconds |
|
| |
1 1296 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
|
Variant with 2 pips more difference
No surprise the above idea was faulty. We have to take into consideration that changing the pip count also increases the win % of the play keeping the contact.
 
| | White is Player 2
score: 0
pip: 109 | Unlimited Game
Jacoby Beaver | pip: 109
score: 0
Blue is Player 1 |
|
| XGID=---A-BEbCB--B----adcbaaa--:0:0:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | Rollout1 | 8/4 6/3 | eq: +0.254 |
| Player:
Opponent: | 57.92% (G:0.86% B:0.03%)
42.08% (G:0.89% B:0.04%) | Conf.: ± 0.008 (+0.245...+0.262) - [100.0%]
Duration: 23.2 seconds |
|
| 2. | Rollout1 | 12/9 12/8 | eq: +0.183 (-0.071) |
| Player:
Opponent: | 56.10% (G:0.05% B:0.00%)
43.90% (G:0.03% B:0.00%) | Conf.: ± 0.007 (+0.176...+0.190) - [0.0%]
Duration: 11.5 seconds |
|
| |
1 1296 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
|
Variant 3 with ~10% difference!
In this example after the roll we're up 10 pips which is almost 10% (but not on roll of course) and it's still correct to keep contact. I'm not going to try to refine a rule at this point but it shouldn't be hard to look at many similar positions and figure out the general break point (for eg up 12% after the roll) for breaking contact. It will of course likely differ some due to the length of the race also and other minor factors.
| | White is Player 2
score: 0
pip: 112 | Unlimited Game
Jacoby Beaver | pip: 109
score: 0
Blue is Player 1 |
|
| XGID=---A-BEbCB--B--a-accbaaa--:0:0:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | Rollout1 | 8/4 6/3 | eq: +0.363 |
| Player:
Opponent: | 61.37% (G:0.82% B:0.05%)
38.63% (G:0.83% B:0.02%) | Conf.: ± 0.009 (+0.354...+0.372) - [100.0%]
Duration: 26.2 seconds |
|
| 2. | Rollout1 | 12/9 12/8 | eq: +0.343 (-0.020) |
| Player:
Opponent: | 61.18% (G:0.08% B:0.00%)
38.82% (G:0.05% B:0.00%) | Conf.: ± 0.008 (+0.334...+0.351) - [0.0%]
Duration: 14.3 seconds |
|
| |
1 1296 Games rolled with Variance Reduction.
Moves: 3-ply, cube decisions: XG Roller
|