Two Plus Two Forums
Sports and Games
Chess and Other Board Games
Backgammon Forum hosted by Bill Robertie.
slotting in the homeboard
12-14-2014
, 12:03 PM
This thread is meant for positions in which slotting in the homeboard is first choice. Everytime I will encounter a unique one I will add it to this thread. Everyone is welcome to collaborate.
How much equity you guess is won by slotting.
How much equity you guess is won by slotting.
12-14-2014
, 05:58 PM
Yes, it is not easy to hide QF's.
And how would you play this one?
And how would you play this one?
12-14-2014
, 08:23 PM
12-15-2014
, 03:35 PM
The first rollout was made for no Jacoby, no beavers. The second rollout uses the Jacoby rule (with the same seed). It generated a similar result, with a somewhat smaller margin (slotting, in fourth place, trails the top play by only 0.009).
A more important consideration might be my choice of Supremo checker play for the rollout. Supremo uses the same 2-ply settings as World Class, except it uses a much larger move filter. On each move in a rollout, Supremo checks the top 16 plays that are within 0.32 of best using 2-ply evals. This can be very important in containment positions, because bots are notoriously weak in playing them. World class checks only the top 8 plays that are within 0.16 of best.
My 5k XG rollouts also have slotting ranked as a small error. The XG rollout using the Jacoby rule has fourth-best slotting behind by 0.016.
White - Pips 55
Black - Pips 127
Black to Play 4-1
Rollout #1 – No Jacoby, No Beavers
GnuBg ID: Ph8AALRtRgAGAw:cAkGAAAAAAAE
Code:
1. Rollout 19/15 9/8 Eq.: +0.293877
0.642928 0.000000 0.000000 - 0.357072 0.173430 0.019293 CL +0.093133 CF +0.293877
[0.001384 0.000000 0.000000 - 0.001384 0.001384 0.000540 CL 0.003244 CF 0.007582]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
2. Rollout 19/18 19/15 Eq.: +0.293644 ( -0.000233)
0.642356 0.000000 0.000000 - 0.357644 0.168572 0.016330 CL +0.099809 CF +0.293644
[0.001366 0.000000 0.000000 - 0.001366 0.001329 0.000510 CL 0.003214 CF 0.007409]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
3. Rollout 19/14 Eq.: +0.287390 ( -0.006487)
0.642083 0.000000 0.000000 - 0.357917 0.167501 0.016260 CL +0.100405 CF +0.287390
[0.001408 0.000000 0.000000 - 0.001408 0.001335 0.000465 CL 0.003288 CF 0.007550]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
4. Rollout 19/18 9/5 Eq.: +0.274278 ( -0.019599)
0.643044 0.000000 0.000000 - 0.356956 0.193109 0.024657 CL +0.068322 CF +0.274278
[0.001349 0.000000 0.000000 - 0.001349 0.001438 0.000678 CL 0.003409 CF 0.008160]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
Rollout #2 – Using Jacoby Rule and Beavers
GnuBg ID: Ph8AALRtRgAGAw:cAkGAAAAAAAA
Code:
1. Rollout 19/15 9/8 Eq.: +0.301868
0.642305 0.000000 0.000000 - 0.357695 0.172450 0.019663 CL +0.092498 CF +0.301868
[0.001362 0.000000 0.000000 - 0.001362 0.001348 0.000466 CL 0.003180 CF 0.007496]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
2. Rollout 19/14 Eq.: +0.300662 ( -0.001206)
0.642098 0.000000 0.000000 - 0.357902 0.166549 0.016428 CL +0.101220 CF +0.300662
[0.001371 0.000000 0.000000 - 0.001371 0.001259 0.000405 CL 0.003190 CF 0.007500]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
3. Rollout 19/18 19/15 Eq.: +0.300606 ( -0.001262)
0.641945 0.000000 0.000000 - 0.358055 0.169456 0.017018 CL +0.097417 CF +0.300606
[0.001337 0.000000 0.000000 - 0.001337 0.001327 0.000414 CL 0.003214 CF 0.007409]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
4. Rollout 19/18 9/5 Eq.: +0.292579 ( -0.009289)
0.643263 0.000000 0.000000 - 0.356737 0.193158 0.025161 CL +0.068206 CF +0.292579
[0.001353 0.000000 0.000000 - 0.001353 0.001421 0.000653 CL 0.003401 CF 0.008689]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 766411571 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]
- Select the plays you want to show in the GnuBg Move Panel. (Click the first, then hold the Shift key and click the last.)
- Click the copy button in the Move Panel. This opens the “Move details” window.
- Press Ctrl+A for “Select All,” followed by Ctrl+C for copy. This copies the move details to the Windows clipboard.
- Switch to the TwoPlusTwo Reply window, and click the # button to insert CODE tags in your message.
- Press Ctrl+V to paste in the move details you copied from GnuBg.
12-15-2014
, 04:42 PM
You post this while I have been toiling doing research on these kind of positions. It is really frustrating, because I have been requesting for XG rollouts some messages ago.
In anyway, we are not so infallible in our play as the bot, so those results could be more practical for us vogues. From now on I refer to a full rollout when it concerns an untruncated rollout consisting of 1296 trials at worldclass level in an unlimited match.
Okay, now back to business.
Getting 2 checkers of white's stacks in the last position, so with 6 in the bearoff, gives a difference of .02, and still another 2 white checkers in the bearoff making a total of 8 gives a difference of .05. So the correlation between slotting and bornoff checkers is rather small.
With an open 4-point the results will be about the same as with the 5-point.
If instead of the 5-point the 6-point is open, the 51 dice combination will not lead to an extra blot, so the equity is somewhat less. The outcome for the following position with a full rollout:
18/13 0.286 and 17/16 10/6 0.274
The outcome for the following position with a full rollout: 14/9 0.595 and 14/13 7/3 0.575
This means that when there is an open 3-point you are alright to slot in case the two checkers are at the outerboard.
Things change drastically with an open 2-point. Only combinations 25 and 26 give white the opportunity to get out of jail, so the best tactic is to wait until white throws a two, and then try to hit him on his head.
Rule of thumb for these positions:
Slot always when the 4, 5 or 6-point is open, when the 3-point is open slot in case the two checkers are away from white's homeboard, and don't slot with an open 2-point.
Also for those points I assume that until 10 born-off checkers the results will not vary significantly.
In anyway, we are not so infallible in our play as the bot, so those results could be more practical for us vogues. From now on I refer to a full rollout when it concerns an untruncated rollout consisting of 1296 trials at worldclass level in an unlimited match.
Okay, now back to business.
Getting 2 checkers of white's stacks in the last position, so with 6 in the bearoff, gives a difference of .02, and still another 2 white checkers in the bearoff making a total of 8 gives a difference of .05. So the correlation between slotting and bornoff checkers is rather small.
With an open 4-point the results will be about the same as with the 5-point.
If instead of the 5-point the 6-point is open, the 51 dice combination will not lead to an extra blot, so the equity is somewhat less. The outcome for the following position with a full rollout:
18/13 0.286 and 17/16 10/6 0.274
The outcome for the following position with a full rollout: 14/9 0.595 and 14/13 7/3 0.575
This means that when there is an open 3-point you are alright to slot in case the two checkers are at the outerboard.
Things change drastically with an open 2-point. Only combinations 25 and 26 give white the opportunity to get out of jail, so the best tactic is to wait until white throws a two, and then try to hit him on his head.
Rule of thumb for these positions:
Slot always when the 4, 5 or 6-point is open, when the 3-point is open slot in case the two checkers are away from white's homeboard, and don't slot with an open 2-point.
Also for those points I assume that until 10 born-off checkers the results will not vary significantly.
12-17-2014
, 02:39 AM
White - Pips 55
Black - Pips 108
Black to Play 4-1
XGID=aBBBB-B-AA-A---------e-eB-:0:0:1:41:0:0:0:0:10
All signs point towards slotting. Any play that slots the 5pt is miles ahead of any play that does not. We’re talking about whopper territory.
The smaller risk is to slot. Even if hit, Black should be able to enter quickly. Usually, he will get a shot or two as White comes around the board.
As far as the downside is concerned, often it does not matter whether Black slots or not. Either way, it is going to be bad news for him if White enters. That being the case, this is probably a good time to assume White will dance. As previously noted, Black will probably survive being hit, and may be in a good position to harrass White as he heads for home.
My 20k XG rollout follows.
Mike
3-ply XG2 Rollout -- Unlimited Game, No Jacoby, No Beavers
Black - Pips 108
Black to Play 4-1
XGID=aBBBB-B-AA-A---------e-eB-:0:0:1:41:0:0:0:0:10
All signs point towards slotting. Any play that slots the 5pt is miles ahead of any play that does not. We’re talking about whopper territory.
- Black trails by a zillion. He cannot hope to sneak home while White is on the bar. He must make the 5pt in order to have any chance to win.
- Black’s builders are in position. After slotting, he will have a double-shot to cover.
- White has only a two-point board. In case he is hit, Black will usually have little trouble entering.
- Black’s gap is high in his board. If White enters, it will be hard to keep him bottled up. If the gap were lower down, this might be a tougher decision.
The smaller risk is to slot. Even if hit, Black should be able to enter quickly. Usually, he will get a shot or two as White comes around the board.
As far as the downside is concerned, often it does not matter whether Black slots or not. Either way, it is going to be bad news for him if White enters. That being the case, this is probably a good time to assume White will dance. As previously noted, Black will probably survive being hit, and may be in a good position to harrass White as he heads for home.
My 20k XG rollout follows.
Mike
3-ply XG2 Rollout -- Unlimited Game, No Jacoby, No Beavers
Code:
XGID=aBBBB-B-AA-A---------e-eB-:0:0:1:41:0:0:0:0:10
X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
Pip count X: 108 O: 55 X-O: 0-0
Cube: 1
X to play 41
1. Rollout¹ 11/10 9/5 eq:+0.5747
Player: 71.61% (G:0.00% B:0.00%)
Opponent: 28.39% (G:11.90% B:1.32%)
Confidence: ±0.0023 (+0.5724..+0.5770) - [97.6%]
Duration: 5 minutes 51 seconds
2. Rollout¹ 9/5 8/7 eq:+0.5716 (-0.0031)
Player: 71.18% (G:0.00% B:0.00%)
Opponent: 28.82% (G:12.52% B:1.48%)
Confidence: ±0.0020 (+0.5696..+0.5736) - [2.4%]
Duration: 5 minutes 01 second
3. Rollout² 9/5 6/5 eq:+0.5009 (-0.0738)
Player: 68.68% (G:0.00% B:0.00%)
Opponent: 31.32% (G:13.98% B:1.56%)
Confidence: ±0.0050 (+0.4959..+0.5059) - [0.0%]
Duration: 1 minute 37 seconds
4. Rollout² 24/20 8/7 eq:+0.4767 (-0.0980)
Player: 69.06% (G:0.00% B:0.00%)
Opponent: 30.94% (G:12.93% B:0.71%)
Confidence: ±0.0071 (+0.4697..+0.4838) - [0.0%]
Duration: 3 minutes 17 seconds
5. Rollout² 24/19 eq:+0.4696 (-0.1051)
Player: 68.85% (G:0.00% B:0.00%)
Opponent: 31.15% (G:13.13% B:0.75%)
Confidence: ±0.0070 (+0.4627..+0.4766) - [0.0%]
Duration: 3 minutes 02 seconds
6. Rollout² 24/20 11/10 eq:+0.4026 (-0.1721)
Player: 67.53% (G:0.00% B:0.00%)
Opponent: 32.47% (G:12.98% B:0.76%)
Confidence: ±0.0054 (+0.3972..+0.4080) - [0.0%]
Duration: 2 minutes 54 seconds
¹ 20736 Games rolled with Variance Reduction.
Dice Seed: 67516426
Moves: 3-ply, cube decisions: XG Roller
² 5184 Games rolled with Variance Reduction.
Dice Seed: 67516426
Moves: 3-ply, cube decisions: XG Roller
Rollout by Taper_Mike
2014-Dec-16
eXtreme Gammon Version: 2.10.199.2658
Flashcard PositionID 000935.xgp
Last edited by Taper_Mike; 12-17-2014 at 02:45 AM.
12-17-2014
, 08:01 AM
Thanks for your tip about the display of the rollout. I will bring it into practice now.
I mentioned that in these positions that the number of checkers born off has little effect on the slotting strategy. The following position shows that the previous results roughly hold just when white has got the 2 and 4 points.
The equity difference of the slotting move is in red, and is negligible. (For the few who don't know, there is a horizontal bar downunder).
I mentioned that in these positions that the number of checkers born off has little effect on the slotting strategy. The following position shows that the previous results roughly hold just when white has got the 2 and 4 points.
White - Pips 37. Match Score 0/64
Black - Pips 127. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 127. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 19/14 Eq.: -0,712
0,381 0,000 0,000 - 0,619 0,205 0,060 CL -0,504 CF -0,712
[0,001 0,000 0,000 - 0,001 0,001 0,001 CL 0,003 CF 0,010]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 767116748 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
2. Rollout 19/18 9/5 Eq.: -0,725 ( -0,013)
0,382 0,000 0,000 - 0,618 0,213 0,069 CL -0,520 CF -0,725
[0,002 0,000 0,000 - 0,002 0,001 0,001 CL 0,003 CF 0,011]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 767116748 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
3. Rollout 19/15 9/8 Eq.: -0,730 ( -0,017)
0,380 0,000 0,000 - 0,620 0,211 0,059 CL -0,512 CF -0,730
[0,001 0,000 0,000 - 0,001 0,001 0,001 CL 0,003 CF 0,010]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 767116748 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
12-17-2014
, 09:06 AM
Slotting gets it's value mainly by the fact that there is a gap between white's homeboard points. But what if white occupies points 2 and 3? All rollouts fulfill the following criteria:
open 6 point:
Slotting is okay when both checkers are out of white's homeboard.
open 5-point
As with open 6 point.
open 4-point
It is not worth slotting anymore.
Code:
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 767026616 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
White - Pips 50. Match Score 0/64
Black - Pips 115. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 115. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 13/12 10/6 Eq.: +0,337
0,656 0,000 0,000 - 0,344 0,169 0,019 CL +0,123 CF +0,337
2. Rollout 24/20 13/12 Eq.: +0,330 ( -0,007)
0,641 0,000 0,000 - 0,359 0,173 0,011 CL +0,098 CF +0,330
3. Rollout 14/10 13/12 Eq.: +0,289 ( -0,048)
0,629 0,000 0,000 - 0,371 0,155 0,011 CL +0,091 CF +0,289
open 5-point
White - Pips 50. Match Score 0/64
Black - Pips 116. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 116. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 14/10 13/12 Eq.: +0,282
0,625 0,000 0,000 - 0,375 0,166 0,014 CL +0,069 CF +0,282
2. Rollout 13/12 9/5 Eq.: +0,272 ( -0,009)
0,632 0,000 0,000 - 0,368 0,185 0,018 CL +0,060 CF +0,272
3. Rollout 24/20 13/12 Eq.: +0,267 ( -0,014)
0,619 0,000 0,000 - 0,381 0,177 0,011 CL +0,049 CF +0,267
open 4-point
White - Pips 50. Match Score 0/64
Black - Pips 108. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 108. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 24/20 12/11 Eq.: +0,421
0,650 0,000 0,000 - 0,350 0,114 0,005 CL +0,181 CF +0,421
2. Rollout 12/11 8/4 Eq.: +0,376 ( -0,044)
0,661 0,000 0,000 - 0,339 0,141 0,011 CL +0,169 CF +0,376
12-17-2014
, 09:58 AM
I still like to simplify the conclusions of the above post. The next position shows that in the above case it is okay to slot with an open 6-point (and 5-point) when one checker gives coverage:
White - Pips 50. Match Score 0/64
Black - Pips 119. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 119. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 24/20 19/18 Eq.: +0,224
2. Rollout 12/11 10/6 Eq.: +0,213 ( -0,011)
0,626 0,000 0,000 - 0,374 0,192 0,023 CL +0,036 CF +0,21
3. Rollout 19/15 12/11 Eq.: +0,185 ( -0,038)
0,604 0,000 0,000 - 0,396 0,200 0,019 CL -0,013 CF +0,185
[0,001 0,000 0,000 - 0,001 0,001 0,000 CL 0,003 CF 0,007]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 767026616 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
12-17-2014
, 10:06 AM
The next position shows the value of slotting with white's 2 and 3-point when there is good coverage. In this case it is exactly 0.1 equity gain.
White - Pips 40. Match Score 0/64
Black - Pips 109. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 109. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 11/10 9/5 Eq.: -0,295
0,446 0,000 0,000 - 0,554 0,169 0,041 CL -0,320 CF -0,295
2. Rollout 24/20 11/10 Eq.: -0,395 ( -0,100)
0,422 0,000 0,000 - 0,578 0,175 0,025 CL -0,356 CF -0,395
12-17-2014
, 04:38 PM
Now I am going to look what happens if white occupies the 3 and 4-point.
Open 6-point
Slotting seems to be right, but it is only apparent.
open 5-point
This shows that the 1 should be played 24/23, but in most cases we don't throw a 1, so that's why I take 54 again:
Not slotting is slightly superior, because it brings a second checker pointing on the open gap. Compare with 19/10 in the first example, where the two checkers are on one point, so not diversified.
open 4-point
In this case bringing a second pointer is so much superior (I stopped the rollout prematurely), that I am going to find out in one of my next posts, whether this always is a better strategy. There also could be a bug.
My conclusion for now is that in these cases it is better to wait with slotting until another pointer has arrived.
Code:
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 767259618 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
White - Pips 46. Match Score 0/64
Black - Pips 126. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 126. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 19/14 10/6 Eq.: -0,356
0,457 0,000 0,000 - 0,543 0,217 0,061 CL -0,365 CF -0,356
2. Rollout 19/10 Eq.: -0,383 ( -0,027)
0,438 0,000 0,000 - 0,562 0,215 0,039 CL -0,380 CF -0,383
open 5-point
White - Pips 46. Match Score 0/64
Black - Pips 119. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 119. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. Rollout 24/23 9/5 Eq.: -0,178
0,503 0,000 0,000 - 0,497 0,166 0,044 CL -0,206 CF -0,178
2. Rollout 11/10 9/5 Eq.: -0,229 ( -0,050)
0,488 0,000 0,000 - 0,512 0,175 0,045 CL -0,245 CF -0,229
3. Rollout 24/19 Eq.: -0,298 ( -0,119)
0,463 0,000 0,000 - 0,537 0,204 0,037 CL -0,316 CF -0,298
White - Pips 46. Match Score 0/64
Black - Pips 127. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 127. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 19/10 Eq.: -0,315
0,463 0,000 0,000 - 0,537 0,197 0,047 CL -0,319 CF -0,315
2. Rollout 19/14 9/5 Eq.: -0,335 ( -0,020)
0,466 0,000 0,000 - 0,534 0,207 0,055 CL -0,332 CF -0,335
open 4-point
White - Pips 46. Match Score 0/64
Black - Pips 128. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 128. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 19/10 Eq.: -0,079
0,525 0,000 0,000 - 0,475 0,171 0,044 CL -0,166 CF -0,079
2. Rollout 24/19 8/4 Eq.: -0,351 ( -0,272)
0,471 0,000 0,000 - 0,529 0,238 0,040 CL -0,337 CF -0,351
3. Rollout 24/20 24/19 Eq.: -0,479 ( -0,400)
0,450 0,000 0,000 - 0,550 0,283 0,017 CL -0,403 CF -0,479
White - Pips 46. Match Score 0/64
Black - Pips 128. Match Score 0/64
Black to Play 4-3
Created with www.BGdiagram.com
Black - Pips 128. Match Score 0/64
Black to Play 4-3
Created with www.BGdiagram.com
Code:
1. Rollout 19/12 Eq.: -0,187
0,500 0,000 0,000 - 0,500 0,184 0,044 CL -0,228 CF -0,187
2. Rollout 19/16 8/4 Eq.: -0,226 ( -0,038)
0,496 0,000 0,000 - 0,504 0,194 0,053 CL -0,255 CF -0,226
3. Rollout 24/17 Eq.: -0,388 ( -0,201)
0,447 0,000 0,000 - 0,553 0,256 0,041 CL -0,405 CF -0,388
12-18-2014
, 10:42 AM
Code:
Full cubeful rollout with var.redn.
42 games, Mersenne Twister dice gen. with seed 767159485 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0,16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
White - Pips 46. Match Score 0/64
Black - Pips 128. Match Score 0/64
Black to Play 6-4
Created with www.BGdiagram.com
Black - Pips 128. Match Score 0/64
Black to Play 6-4
Created with www.BGdiagram.com
Code:
1. Rollout 19/9 Eq.: -0,167
0,503 0,000 0,000 - 0,497 0,176 0,043 CL -0,215 CF -0,167
2. Rollout 19/13 8/4 Eq.: -0,179 ( -0,012)
0,505 0,000 0,000 - 0,495 0,184 0,050 CL -0,225 CF -0,179
3. Rollout 24/14 Eq.: -0,181 ( -0,015)
0,481 0,000 0,000 - 0,519 0,190 0,029 CL -0,259 CF -0,181
4. Rollout 24/18 8/4 Eq.: -0,248 ( -0,081)
0,490 0,000 0,000 - 0,510 0,222 0,033 CL -0,276 CF -0,248
open 3-point
White - Pips 46. Match Score 0/64
Black - Pips 129. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 129. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 19/14 7/3 Eq.: -0,157
0,519 0,000 0,000 - 0,481 0,170 0,044 CL -0,177 CF -0,157
2. Rollout 19/10 Eq.: -0,166 ( -0,009)
0,507 0,000 0,000 - 0,493 0,166 0,032 CL -0,186 CF -0,166
White - Pips 46. Match Score 0/64
Black - Pips 127. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 127. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 18/9 Eq.: -0,099
0,529 0,000 0,000 - 0,471 0,156 0,035 CL -0,135 CF -0,099
2. Rollout 18/13 7/3 Eq.: -0,146 ( -0,047)
0,519 0,000 0,000 - 0,481 0,168 0,043 CL -0,174 CF -0,146
again 65 is thrown, in order to get a third checker in direct range.
White - Pips 46. Match Score 0/64
Black - Pips 118. Match Score 0/64
Black to Play 6-5
Created with www.BGdiagram.com
Black - Pips 118. Match Score 0/64
Black to Play 6-5
Created with www.BGdiagram.com
Code:
1. Rollout 24/18 7/2 Eq.: +0,051
0,553 0,000 0,000 - 0,447 0,114 0,014 CL -0,023 CF +0,051
2. Rollout 19/13 7/2 Eq.: +0,004 ( -0,046)
0,542 0,000 0,000 - 0,458 0,113 0,023 CL -0,053 CF +0,004
3. Rollout 19/8 Eq.: -0,008 ( -0,059)
0,532 0,000 0,000 - 0,468 0,101 0,017 CL -0,054 CF -0,008
open 1-point : black should try to prime
Last edited by yogiman; 12-18-2014 at 10:48 AM.
12-18-2014
, 11:55 AM
Coming back to the position at the bottom of post #14. I remarked that slotting is not productive anymore. There is only a place for a second white blot if black moves from the 24-point. Even the two outer points (5&6) in black's homeboard decrease the necessity of recycling a black checker. Probably better to get a builder on those points.
However, there should be a little correction, because 3 pointing checkers do make it profitable, though slightly:
But in case of the open 6 or 5-point there is enough gain in slotting when there is a second checker in direct range:
So to keep matters simple in the case of white's 2,3 homeboard, slot only with an open 5 or 6-point, and only when there is at least one checker backing it.
However, there should be a little correction, because 3 pointing checkers do make it profitable, though slightly:
White - Pips 50. Match Score 0/64
Black - Pips 87. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 87. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 12/7 8/4 Eq.: +0,589
0,712 0,000 0,000 - 0,288 0,078 0,003 CL +0,343 CF +0,589
2. Rollout 17/8 Eq.: +0,566 ( -0,023)
0,681 0,000 0,000 - 0,319 0,020 0,001 CL +0,341 CF +0,566
White - Pips 50. Match Score 0/64
Black - Pips 97. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 97. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 14/9 10/6 Eq.: +0,438
0,655 0,000 0,000 - 0,345 0,145 0,006 CL +0,159 CF +0,438
2. Rollout 17/12 14/10 Eq.: +0,408 ( -0,030)
0,651 0,000 0,000 - 0,349 0,083 0,002 CL +0,215 CF +0,408
12-19-2014
, 06:46 AM
Returning to the first position of this thread, I have been very incomplete.
Open 6-point
Getting a second pointer is far superior. That's due to those checkers in white's outfield. On the one hand black will have a lot of opportunity of making a point in direct range next turn, on the other hand if white enters next turn there is a lot of chance he will be hit.
open 5-point
The possibility of white throwing 51 makes a lot of difference. Slotting is not very much worse than making a direct point. 16/7, three checkers in a row gives white a good chance to hit upon entering.
The following position shows that slotting is almost as good as bringing pointers. And actually we are not entirely sure, for the weakness of rollouts is that after our alternative play the bot will go on according to it's own conditioning.
Open 6-point
White - Pips 43. Match Score 0/64
Black - Pips 114. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 114. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 19/15 17/12 Eq.: -0,222
0,458 0,000 0,000 - 0,542 0,184 0,016 CL -0,285 CF -0,222
2. Rollout 19/15 11/6 Eq.: -0,329 ( -0,107)
0,444 0,000 0,000 - 0,556 0,223 0,019 CL -0,356 CF -0,329
open 5-point
White - Pips 43. Match Score 0/64
Black - Pips 110. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 110. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 17/13 16/11 Eq.: -0,054
0,507 0,000 0,000 - 0,493 0,129 0,016 CL -0,132 CF -0,054
2. Rollout 16/12 10/5 Eq.: -0,093 ( -0,038)
0,502 0,000 0,000 - 0,498 0,154 0,030 CL -0,182 CF -0,093
3. Rollout 16/7 Eq.: -0,135 ( -0,081)
0,485 0,000 0,000 - 0,515 0,159 0,023 CL -0,214 CF -0,135
The following position shows that slotting is almost as good as bringing pointers. And actually we are not entirely sure, for the weakness of rollouts is that after our alternative play the bot will go on according to it's own conditioning.
White - Pips 43. Match Score 0/64
Black - Pips 112. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 112. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. Rollout 16/11 13/9 Eq.: -0,115
0,495 0,000 0,000 - 0,505 0,158 0,026 CL -0,195 CF -0,115
2. Rollout 16/12 10/5 Eq.: -0,129 ( -0,013)
0,493 0,000 0,000 - 0,507 0,167 0,027 CL -0,208 CF -0,129
12-19-2014
, 03:45 PM
I skip the open 3 and 4-point, because their results are not very different from the position with an open 2-point.
Slotting is just as good as lenghtening the prime.
The position in post #16 is unrealistic with those 2 checkers on the 24 point. So let's look what is the difference in the real world
Slotting has plunged 1.5 in value compared to post #16. That's because only double 5 will give white a reasonable way out, and then still can be hit in his own homeboard. At the same time black is increasing his chances considerably to fill up the gap.
White - Pips 43. Match Score 0/64
Black - Pips 99. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Black - Pips 99. Match Score 0/64
Black to Play 5-4
Created with www.BGdiagram.com
Code:
1. 13/8 12/8 Eq.: +0,122
0,543 0,000 0,000 - 0,457 0,056 0,004 CL +0,027 CF +0,122
2. 15/11 7/2 Eq.: +0,115 ( -0,006)
0,550 0,000 0,000 - 0,450 0,087 0,007 CL +0,005 CF +0,115
3. 15/11 12/7 Eq.: +0,108 ( -0,014)
0,534 0,000 0,000 - 0,466 0,050 0,004 CL +0,015 CF +0,108
The position in post #16 is unrealistic with those 2 checkers on the 24 point. So let's look what is the difference in the real world
White - Pips 40. Match Score 0/64
Black - Pips 88. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Black - Pips 88. Match Score 0/64
Black to Play 4-1
Created with www.BGdiagram.com
Code:
1. 14/10 13/12 Eq.: -0,186
0,440 0,000 0,000 - 0,560 0,105 0,006 CL -0,232 CF -0,186
2. 13/12 9/5 Eq.: -0,234 ( -0,048)
0,447 0,000 0,000 - 0,553 0,156 0,017 CL -0,279 CF -0,234