Search This Blog

January 16, 2013

texmex whoppers

Below are blunders from an online match against Texmex

Score is 7 away 7 away.  32 to play.  Split or not?

is texmex

score: 0
pip: 128
                         
7 point match
                          pip: 133
score: 0

is Chutzpah
XGID=-B-BbBB-B---aC--abbe---bB-:0:0:1:32:0:0:0:7:10
to play 32

1.XG Roller++24/22 24/21eq: -0.147
Player:
Opponent:
48.29% (G:11.71% B:0.32%)
51.71% (G:18.78% B:0.48%)
2.XG Roller++24/22 13/10eq: -0.215 (-0.068)
Player:
Opponent:
45.97% (G:12.55% B:0.40%)
54.03% (G:20.65% B:1.01%)
3.XG Roller++24/21 8/6eq: -0.249 (-0.102)
Player:
Opponent:
44.64% (G:11.05% B:0.32%)
55.36% (G:18.96% B:0.86%)
4.XG Roller+24/21 13/11eq: -0.092 (+0.054)
Player:
Opponent:
49.15% (G:12.40% B:0.40%)
50.85% (G:17.17% B:0.76%)
5.XG Roller+13/8eq: -0.232 (-0.085)
Player:
Opponent:
44.86% (G:10.42% B:0.25%)
55.14% (G:15.15% B:0.56%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
13/8.  I didn't split. On the logic that texmex has a lame prime going and is just itching to attack me.

However, any form of splitting is better than my play.  The best version is 24/21, 13/11 which splits to the highest point and duplicates 52 and 43 as a bonus. 

Why split?  Not sure.   Here are some possible reasons:

  • I have him outboarded.  Therefore contact favors me.
  • It is bad for me when he makes the 4 point whether or not my man is there.
  • If I do get pointed on, I have some return shots from the bar, plus some anchoring chances (like 54 or 32)
  • I am short on timing.  If I don't split now, I probably never will.  This seems to be my best chance.
I looked at a variation where I improved the opponent's position.  I moved his spares from the six point and closed his 5 point.  Splitting is still a whopper even in that position.

Another variation gives texmex the 5 point and cleans up all blots.  (Move 2 from six point to close 5 point, and move man from 9 point to midpoint).  Splitting is even more correct.

So even as the opponent's position is getting stronger, splitting is getting more and more correct.  Not sure what's going on here.  Perhaps the above reasons (outboarded, sucks when he makes it anyway, if not now, when?) make this a last ditch now or never try for a game.

****************
Later in the same game.  I own the cube and have 52 to play.  Duplicate 4's or least outside shots?

is texmex

score: 0
pip: 72
                         
7 point match
                          pip: 71
score: 0

is Chutzpah
XGID=-ECBC-------a----bbb-Bccb-:0:0:1:52:0:0:0:7:10
to play 52

1.XG Roller++21/14eq: -0.708
Player:
Opponent:
30.37% (G:0.95% B:0.01%)
69.63% (G:6.19% B:0.11%)
2.XG Roller++21/16 4/2eq: -0.781 (-0.074)
Player:
Opponent:
29.50% (G:0.94% B:0.01%)
70.50% (G:7.26% B:0.13%)
3.2-ply21/16 3/1eq: -1.000 (-0.292)
Player:
Opponent:
24.24% (G:0.84% B:0.00%)
75.76% (G:11.16% B:0.20%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
I went for duplicating 4's with 21/16, 4/2.

Correct was running all the way with 21/14.

On closer inspection, it appears that my play doesn't duplicate anything, and actually diversifies his 4's.

After 21/16, 4/2 ...

41 picks and passes clean.  The other way picks and passes on the inner board, giving me a few combo returns from the bar.
42 also picks and passes clean.  The other way picks and passes on the inner board and I get a few combos from the bar.
43 is strong either way.
44 points on me and hits a second checker.  The other way 44 is really good but at least doesn't pick up a second man.
45 now becomes a pick and pass whereas the other way 45 leaves a direct shot.
46 also becomes pick and pass while the other way it plays clean.

I guess the lesson is look at how the numbers play.  It 'feels' like I am duplicating 4's athough I am actually making all 4's really good numbers.

21/14 gives good 2's but gives me shots on 45 as some compensation.  Also the race is close and this brings the man closer to home. 

*************
I trail 7 away 5 away and trying to scramble home after getting hit from a 1/3 backgame.  32 to play.

is texmex

score: 2
pip: 173
                         
7 point match
                          pip: 56
score: 0

is Chutzpah
XGID=-bGbFb-Aa--A--------bb-bb-:1:-1:1:32:0:2:0:7:10
to play 32

1.XG Roller+11/9 7/4eq: +0.133
Player:
Opponent:
50.40% (G:27.63% B:2.81%)
49.60% (G:5.30% B:0.11%)
2.XG Roller+11/8* 4/2eq: -0.359 (-0.492)
Player:
Opponent:
36.93% (G:20.14% B:1.81%)
63.07% (G:18.86% B:0.44%)
3.3-ply7/2eq: -0.049 (-0.182)
Player:
Opponent:
45.73% (G:25.12% B:2.60%)
54.27% (G:8.15% B:0.20%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
This one is so obvious there really is no point in analyzing it.

I instantly played 11/8*, 4/2 temporizing by putting him on the bar.  But let's count the return shots.

61, 62, 63, 64, 65, 66
55, 54, 53, 52, 51,
43, 41
33, 32, 31
21

Everything except 42 and 44 which dance.  And it leaves two blots!!

11/9, 7/4 walks into a triple shot but cleans up a blot.  'One blot theory' is a good rule of thumb, but is completely obvious here where it actually leaves less shots.  55, 52, 32 miss.

The reason to post the problem is to remind myself to slow down.  Playing too fast, making the first move I see gets me into trouble more often than I care to admit.  I estimate that by always pausing to consider a second checker play option I would probably improve my PR by at least .2

****************
I lead 3 away 5 away, have already doubled and have 32 to play.

is texmex

score: 2
pip: 86
                         
7 point match
                          pip: 86
score: 4

is Chutzpah
XGID=-B-BbBCBB-B--------bbbbcb-:1:-1:1:32:4:2:0:7:10
to play 32

1.XG Roller++8/6 8/5eq: +0.220
Player:
Opponent:
69.62% (G:2.94% B:0.04%)
30.38% (G:0.99% B:0.01%)
2.XG Roller++6/1eq: +0.198 (-0.022)
Player:
Opponent:
68.98% (G:2.34% B:0.03%)
31.02% (G:1.24% B:0.01%)
3.XG Roller++10/8 10/7eq: +0.179 (-0.041)
Player:
Opponent:
67.63% (G:2.19% B:0.02%)
32.37% (G:0.76% B:0.01%)

eXtreme Gammon Version: 2.02, MET: Kazaross XG2

I cleared from the rear, 10/8, 10/7. Not terrible but a illogical technical error which is worth time examining. The race is close so contact favors me here. I want to slow opponent down by blocking the big numbers (sixes here). Also I want to position myself to attack my opponent should he run with one checker. 8/6, 8/5 suits both objectives. I have 4 builders/attackers in case of 45 or 46. Boxes are blocked. And if opponent rolls 5's or 4's he has to think twice about running off the anchor. Even the lame looking 6/1 is superior to my play. While it kills a builder at least it holds the blockade. My play disengages from the rear at a time when that is simply the wrong game plan.

January 7, 2013

A42kilo: a couple more early game blunders

Here are two more early game checker blunders by me.  These are taken from an online match against A42kilo.


**************
I trail 7 away 6 away and have 61 to play.
54S-61P-21$-61 ??

is A42kilo

score: 1
pip: 155
                         
7 point match
                          pip: 160
score: 0

is Chutzpah
XGID=-a---aEBB---cDa--d-da---B-:0:0:1:61:0:1:0:7:10
to play 61

1.XG Roller++24/18 6/5* eq: -0.018
Player:
Opponent:
49.80% (G:12.23% B:0.57%)
50.20% (G:14.49% B:0.64%)
2.XG Roller++13/7 6/5* eq: -0.061 (-0.044)
Player:
Opponent:
48.30% (G:13.75% B:0.65%)
51.70% (G:15.33% B:0.88%)
3.XG Roller++7/1* 6/5* eq: -0.069 (-0.051)
Player:
Opponent:
47.51% (G:16.54% B:0.86%)
52.49% (G:16.21% B:1.19%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
I played 13/7, 6/5*.  Down from the midpoint is the wrong idea and splitting is better.

My midpoint is already unstacked and doesn't need development.  The bar is already made.  In other words, my back men or defense are in more urgent need of improvement than my offense.

In addition, opponent's 5 point blot and duplicatin of sixes works in my favor.  For example, if opponent enters with a six should he cover the 5 point of hit me?  Similarly, would opponent prefer to cover or hit me with an ace?

I've seen this idea before.  Hopefully it will stick now.

***********
Score is 6 away  and I have 62 to play.
31P-53P-65R-31P-51D-62??  [not sure on the nactation]


is A42kilo

score: 1
pip: 146
                         
7 point match
                          pip: 155
score: 1

is Chutzpah
XGID=-a-B-BC-A---eE---c-cc---B-:0:0:1:62:1:1:0:7:10
to play 62

1.XG Roller+13/5eq: -0.298
Player:
Opponent:
41.43% (G:12.44% B:0.28%)
58.57% (G:13.73% B:0.67%)
2.XG Roller+24/16eq: -0.379 (-0.081)
Player:
Opponent:
41.23% (G:9.75% B:0.28%)
58.77% (G:15.74% B:0.59%)
3.4-ply24/18 13/11eq: -0.325 (-0.027)
Player:
Opponent:
42.55% (G:10.63% B:0.32%)
57.45% (G:16.98% B:0.86%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
I ran with 24/16.  Wrong, wrong, wrong on so many levels.  13/5 is far and away the best play.

Why am I running when behind in the race?
Why am I leaving a blot in front of his stacked midpoint where he would love to hit?
Why am I leaving a blot at all when the roll plays nicely?  Meaning, 13/5 develops / unstacks the midpoint.



January 3, 2013

Deep Crawford, 2-away and 3-away scores

I saw this on bgonline.org and wanted to document it before I totally forgot it.  I am sure I will forget anyway, but at least I know where to look to find it again now.

Here is the original thread.  http://www.bgonline.org/forums/webbbs_config.pl?read=134602

*************

Neil's Rule of .6: for approximating deep Crawford equities.  By Neil Kazaross.

Start from 6 away Crawford (11.3%) and multiply by .6 to get 8-away, 10-away, etc.

Or Start from 7 away Crawford (9.3%) and multiply by .6 to get 9-away, 11-away, etc.

** Neil recommends keeping in mind that .

6 x .6 is .36 ( a bit more than a 3rd)
.6 cubed is .216  (about .22 or a bit more than a fifth)
.6 to the 4th is .1296 (.13 or about an eighth)

Examples: Neil is way more exact as he calcs this in his head.  Below is a more realistic way of how I would do it.

6-away Crawford is 11% so ...
8-away Crawford is 11 x .6 or about 6.5%  [actual 6.76]
10-away Crawford is 11 x .36 or a bit more than a third or a bit less than 4% [actual 4.07]
12-away Crawford is 11 x .22 or a bit more than a fifth or a bit more than 2% [actual 2.45]
14-away Crawford is 11 x an eighth or about 1.5% [actual 1.47]

************

Chuck's rule of .7868: for approximating deep 2-away scores.  By Chuck Bower.

Start with the deepest 2 away score you remember and multiply by .7868 to get the next one, ad infinitum down the line.

OK, OK no way I am doing .7868 so let's call this Chuck's rule of .8 - close enough for me.

* note * .8 squared is .64 or a bit less than 2/3 and .8 cubed is about 1/2

2-away 6 away Crawford is 20%
2-away 7 away Crawford is 16% [actual 16.2]
2-away 8 away is 12.8

* note * another trick for 2-away equities comes from MCG (Matt Cohn-Geier) who writes that the two away equities are close to the crawford equities, shift 2.

So using MCG...
2 away 6 away is like 4 away Crawford or 21% [actual 20]
2 away 7 away is like 5 away Crawfod or 16% [actual 16.2]
2 away 8 away is like 6 away Crawford or 11% [actual 12.5]
2 away 9 away is like 7 away Crawford or 9% [actual 9.8]


*************

Jason's Rule of 2/3: for approximating deep 3-away equities.  By Jason Lee.

Start from 3 away 6 away (28.9%) and multiply by 2/3 to get 8 away, 10 away, etc.

Or Start with 3 away 7 away (23.8%) and multply by 2/3 to get 9 away, 11 away, etc.

Examples: again doing it my way since math is not my forte. Jason is very exact and all in his head!!

3 away 6 away.  Neil's Number is 7 so leader is 71 and trailer 29. [actual 28.9]
3 away 8 away.  2/3 of 29 is about 20% [actual 19.5]
3 away 10 away.  2/3 of 20% is about 14% [actual 12.9]

3 away 7 away.  Stick's 7 away trick says we know 7 away Crawford is 9%, then add 7% for each jump up so 3 away 7 away is 23% [actual 23.8%]
3 away 9 away.  2/3 of 23 is about 16% [actual 15.98]
3 away 11 away.  2/3 of 16 or about 10% [actual 10.56]