I am posting the below position with rollout results to request
analysis help. I redoubled and was dinged for a whopper. After a
rollout the only significant statistic was gnubg giving O a much higher
gammon chance. I struggle to see why O is favored here. I have a 19
pip advantage, there are only three rolls which can hurt me (11, 21,
33), and if I enter will be bearing off quite a few checkers before O
ever enters his home board. Does it have something to do with the
score?

------------------

GNU Backgammon Position ID: 9nYDwAD2tocAAA
Match ID : UQngAAAAIAAE
+13-14-15-16-17-18------19-20-21-22-23-24-+ O: gnubg
| | | O O O O O | 0 points
| | | O O O O O |
| | | O O |
| | | O |
| | | |
v| |BAR| | 7 point match
| | | |
| | | X X |
| | | X X |
| O | | X X X X X | On roll
| X O | | X X X X X | 4 points
+12-11-10--9--8--7-------6--5--4--3--2--1-+ X: badgolferman (Cube:
2)
Pip counts : O 87, X 68

Cube analysis
Rollout cubeless equity +0.792 (Money: +0.793)

Cubeful equities:
1. No double +0.882
2. Double, pass +1.000 (+0.118)
3. Double, take +0.766 (-0.116)
Proper cube action: No redouble, take (49.4%)

Rollout details:
Player badgolferman owns 2-cube:
0.896 0.001 0.000 - 0.104 0.001 0.000 CL +0.792 CF +0.882
[0.000 0.001 0.000 - 0.000 0.000 0.000 CL 0.000 CF 0.001]
Player gnubg owns 4-cube:
0.897 0.009 0.000 - 0.103 0.001 0.000 CL +1.180 CF +0.766
[0.000 0.002 0.000 - 0.000 0.000 0.000 CL 0.001 CF 0.002]
Truncated cubeful rollout (depth 10) with variance reduction
1296 games, Mersenne Twister dice gen. with seed 405516982 and
quasi-random dice
Play: supremo 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

On 7/29/2021 9:58 AM, badgolferman wrote:
> I am posting the below position with rollout results to request
> analysis help. I redoubled and was dinged for a whopper. After a
> rollout the only significant statistic was gnubg giving O a much higher
> gammon chance. I struggle to see why O is favored here. I have a 19
> pip advantage, there are only three rolls which can hurt me (11, 21,
> 33), and if I enter will be bearing off quite a few checkers before O
> ever enters his home board. Does it have something to do with the
> score?

Yes, the score is hugely important. O will redouble immediately and
play for the match, and needs only about 9% winning chances to take.
It might seem that O doesn't have that, but in addition to the bad
sequences you mentioned in which you get hit, you could also lose
the race. For example, if you roll one of your worst racing rolls,
31, then the computer estimates you have an 11% chance of losing the
race.

Having said that, I'll note that eXtremeGammon actually recommends
redoubling, so I'm a little surprised at the GNU rollout results.
See below. Maybe you're using a different match equity table.

Let me also add that in practice, at this score, I will redouble to 4
only if I'm confident that either my opponent will pass, or should pass.
If my opponent has, say, 10% winning chances and takes, I'll be really
annoyed if I lose the game and thereby the match. On the other hand,
if I lose my market, I'm not too upset to enter the Crawford game with
my opponent needing to win 7 points before I win one game. Yes, it is
theoretically true that redoubling at exactly the right moment should
increase my match-winning chances, but even the computer will usually
say that the missed double is only a small technical error, and the
psychological aspects are important in practice. Many players get
demoralized when trailing by that much and will play poorly.

XGID=--BDBBDb--A--------bcbdb--:1:1:1:00:4:0:0:7:10

X:Player 1 O:Player 2
Score is X:4 O:0 7 pt.(s) match.
+13-14-15-16-17-18------19-20-21-22-23-24-+
| | | O O O O O |
| | | O O O O O |
| | | O O |
| | | O |
| | | |
| |BAR| |
| | | |
| | | X X |
| | | X X | +---+
| O | | X X X X X | | 2 |
| X O | | X X X X X | +---+
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 68 O: 87 X-O: 4-0/7
Cube: 2, X own cube
X on roll, cube action

Analyzed in Rollout
No redouble
Player Winning Chances: 89.63% (G:0.08% B:0.00%)
Opponent Winning Chances: 10.37% (G:0.09% B:0.00%)
Redouble/Take
Player Winning Chances: 89.73% (G:0.07% B:0.00%)
Opponent Winning Chances: 10.27% (G:0.08% B:0.00%)

Cubeful Equities:
No redouble: +0.895 (-0.028)
Redouble/Take: +0.923
Redouble/Pass: +1.000 (+0.077)

Best Cube action: Redouble / Take

Rollout:
1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 4-ply, cube decisions: XG Roller+
Search interval: Large
Confidence No Double: ± 0.002 (+0.893..+0.896)
Confidence Double: ± 0.002 (+0.921..+0.925)

eXtreme Gammon Version: 2.19.207.pre-release, MET: Kazaross XG2

---
Tim Chow

On Friday, July 30, 2021 at 4:54:14 AM UTC+1, Tim Chow wrote:
> On 7/29/2021 9:58 AM, badgolferman wrote:
> > I am posting the below position with rollout results to request
> > analysis help. I redoubled and was dinged for a whopper. After a
> > rollout the only significant statistic was gnubg giving O a much higher
> > gammon chance. I struggle to see why O is favored here. I have a 19
> > pip advantage, there are only three rolls which can hurt me (11, 21,
> > 33), and if I enter will be bearing off quite a few checkers before O
> > ever enters his home board. Does it have something to do with the
> > score?
> Yes, the score is hugely important. O will redouble immediately and
> play for the match, and needs only about 9% winning chances to take.
> It might seem that O doesn't have that, but in addition to the bad
> sequences you mentioned in which you get hit, you could also lose
> the race. For example, if you roll one of your worst racing rolls,
> 31, then the computer estimates you have an 11% chance of losing the
> race.
>
> Having said that, I'll note that eXtremeGammon actually recommends
> redoubling, so I'm a little surprised at the GNU rollout results.
> See below. Maybe you're using a different match equity table.
>
> Let me also add that in practice, at this score, I will redouble to 4
> only if I'm confident that either my opponent will pass, or should pass.
> If my opponent has, say, 10% winning chances and takes, I'll be really
> annoyed if I lose the game and thereby the match. On the other hand,
> if I lose my market, I'm not too upset to enter the Crawford game with
> my opponent needing to win 7 points before I win one game. Yes, it is
> theoretically true that redoubling at exactly the right moment should
> increase my match-winning chances, but even the computer will usually
> say that the missed double is only a small technical error, and the
> psychological aspects are important in practice. Many players get
> demoralized when trailing by that much and will play poorly.

Suppose that your opponent beats you with a bad take, how would you feel then?
I'd feel completely peachy about it -- in fact, I might buy a peach if I was in London
or some other major European city in summer, where they have good produce in open markets.

Paul

On 7/30/2021 6:11 AM, peps...@gmail.com wrote:
> Suppose that your opponent beats you with a bad take,
> how would you feel then? I'd feel completely peachy about it

I can't say I'd feel peachy about it, but I'd feel better
about it than if my opponent were to beat me with a good take.

---
Tim Chow

Timothy Chow wrote:

>Yes, the score is hugely important. O will redouble immediately and
>play for the match, and needs only about 9% winning chances to take.
>It might seem that O doesn't have that, but in addition to the bad
>sequences you mentioned in which you get hit, you could also lose
>the race. For example, if you roll one of your worst racing rolls,
>31, then the computer estimates you have an 11% chance of losing the
>race.

It seems impossible to me for someone to estimate the difference
between 9, 10, 11 % chance of winning OTB.

>Having said that, I'll note that eXtremeGammon actually recommends
>redoubling, so I'm a little surprised at the GNU rollout results.
>See below. Maybe you're using a different match equity table.

I don't know how to verify this. Regardless though, should there be
such a difference between gnubg calling this double a whopper and xg
calling it good?

On 7/30/2021 9:29 AM, badgolferman wrote:
> Timothy Chow wrote:
>> Having said that, I'll note that eXtremeGammon actually recommends
>> redoubling, so I'm a little surprised at the GNU rollout results.
>> See below. Maybe you're using a different match equity table.
>
> I don't know how to verify this. Regardless though, should there be
> such a difference between gnubg calling this double a whopper and xg
> calling it good?

You can go to Analyze -> Match equity table to see what the name of
your MET is. Also, what version of GNU are you using? (Go to
Help -> About GNU Backgammon -> Build info.)

When I roll it out using my version of GNU, I get similar results to
what I get with XG.

---
Tim Chow

Timothy Chow wrote:

>On 7/30/2021 9:29 AM, badgolferman wrote:
>>Timothy Chow wrote:
>>>Having said that, I'll note that eXtremeGammon actually recommends
>>>redoubling, so I'm a little surprised at the GNU rollout results.
>>>See below. Maybe you're using a different match equity table.
>>
>>I don't know how to verify this. Regardless though, should there be
>>such a difference between gnubg calling this double a whopper and xg
>>calling it good?
>
>You can go to Analyze -> Match equity table to see what the name of
>your MET is. Also, what version of GNU are you using? (Go to
>Help -> About GNU Backgammon -> Build info.)

met/zadeh.xml

>When I roll it out using my version of GNU, I get similar results to
>what I get with XG.

1.06.002-mingw 32-bit 20180802

I also did a Hint of the position this time and got this:

Cube analysis
2-ply cubeless equity +0.789 (Money: +0.790)
0.895 0.001 0.000 - 0.105 0.001 0.000
Cubeful equities:
1. No double +0.889
2. Double, pass +1.000 (+0.111)
3. Double, take +0.751 (-0.138)
Proper cube action: No redouble, take (55.4%)

Not sure if this is an anomaly and should be reported?

On 8/1/2021 7:34 AM, badgolferman wrote:
> met/zadeh.xml

That's the explanation. The Zadeh MET dates from 1977. Click on
the "Load Table" button and load some other table (one with the
name "Kazaross" in it if there is one).

---
Tim Chow

Timothy Chow wrote:

>On 8/1/2021 7:34 AM, badgolferman wrote:
>>met/zadeh.xml
>
>That's the explanation. The Zadeh MET dates from 1977. Click on
>the "Load Table" button and load some other table (one with the
>name "Kazaross" in it if there is one).
>

Thanks! I changed it to Kazaross xg2 and now receive the same results
as you.