Search This Blog

August 29, 2012

Stick's Cheat Sheet

I was watching Stick's excellent lecture on the 5 away 5away score when I noticed something ineteresting.  This video shows Stick playing practice games against XG at the 5A 5A score. 

I particularly enjoy watching Stick think out loud in real time as he plays through the game.  It is definitely worth your time to watch.  Step inside the mind of a great master!

This is a free lecture, though it will probably only be available for a week or so. Act quick!! (http://www.bgonline.org/forums/webbbs_config.pl?read=12al8780)

Anyway, Stick flashed his "cheat sheet" for match equities while he was playing.  This showed the scores of a five point match only, the most common and critical ones to know.  He suggested all tournament players should have these values memorized - know them cold.

I transcribed the sheet and share it below.

It looks like I need to toss out my own Match Equity table (http://bigmoneybill.blogspot.com/2012/01/bills-match-equity-table.html) and learn Stick's values instead.  See especially the last footnote.  Apparently Stick diverges from XG (most likely Rockwell-Kazaross) in that he assumes less value for recube vig, which means Stick's takepoints are a bit higher than in the Rockwell-Kazaross or XG tables.

While I have some memory work ahead of me, one value that confuses me is the 25.5% for 3 away 3 away.  I learned this was a very high takepoint, almost 30%.  A fast double, fast pass type score.  But XG shows about 30% for dead cube and about 23% for live cube.  And Stick of course shows 25.5%

Maybe this is just an example of XG showing takepoint of 23, add 2% for less recube vig gets you 25% - OK so Stick actually has 25.5% but close enough.  Maybe the 30% only applies to "dead cubes" like a holding game where the opponent is unlikely to get much value in holding the cube?  Meaning, he hits and wins or misses and loses, and he doesn't often catch up in the race to get a 4 cube in?  Of course, Stick knows all this and more, but I still wonder how to interpret these cheat sheet values and use at the table.

Stick's Cheat Sheet
Score Live Cube TP GV on 2 cube
5A 5A 20.0% 0.65
5A 4A 24.5% 0.75
4A 5A 19.5% 0.80
5A 3A 22.0% 0.75
3A 5A 19.0% 0.45
5A 2A 23.5% 0.70
2A 5A 17.5% 0.00
4A 4A 22.5% 1.00
4A 3A 24.5% 0.95
3A 4A 21.5% 0.45
4A 2A 18.5% 1.00
2A 4A 20.0% 0.00
3A 3A 25.5% 0.50
3A 2A 25.0% 0.50
2A 3A 26.0% 0.00
2A 2A 32.0% 0.00
Notes:
a live cube take point means recube vig is already factored in
b all takepoints are rounded to nearest 1/2 percent
c all gammon values are rounded to nearest .05
d Stick's takepoints are about 1.5 - 2% higher than XG
according to Stick, XG assumes a higher recube efficiency than is practical

August 26, 2012

65R-31P-XX

Experimenting with a new feature.  Trying to play all rolls for a given 3rd roll position.  I bought Opening Ceremony but could never get it to install and work.  So instead I will just rely on XG ++ for the evaluations.  These could be off but I am guessing they are close enough.

There is no real rhyme or reason as to why I chose this position to look at.  I guess you gotta start somewhere.

I got several of these wrong, so hopefully looking at these 3rd rolls will be a good learning exercise into the future.

**************
Money game.  Black to play all rolls.

is Player 2

score: 0
pip: 163
                         
Unlimited Game
Jacoby Beaver
                          pip: 156
score: 0

is Player 1
XGID=-b----E-C---eF---b-db---A-:0:0:1:00:0:0:3:0:10
on roll, cube action?

eXtreme Gammon Version: 2.02


General thoughts. 

We have escaped one back man and we are up in the race.  This is good.

On the other hand, we are outboarded and behind in development, which is bad.

Our best gameplan at the moment is a running game - we are behind in terms of both priming and attacking - but the game is young and things could change.

We don't want to get hit, since we would lose the only advantage we have - the race. 

What would Mochy say?

Under Mochy's scheme, I think this positio best fits Blitz vs Prime.

Mochy's advice for this game type:
  • Attack if possible
  • Escpae back man if you can't attack
  • Don't Slot

*** My initial answers are underlined followed by my reasoning. 
*** Correct answer in bold if I am wrong. 

*** More commentary follows below on my wrong answers.

65.  Run 24/13

64. 

Run 24/14.  I think we should play to our strength  and try to escape both men.  We have an added bonus that if missed sixes connect to the 8 point while most numbers one apart make an outside point.

Making the 2 point 8/2, 6/2 doesn't feel right.  While the two point is an improvement, it's not much of an improvement.  This a blitzing / attacking play which is not what we are going for here.  Also we strip the 8 point while leaving the midpoint stacked.  I don't like it.

Splitting 24/18, 13/9 is a possibility but I don't like it either.  In the analagous 2nd roll position 31P-64 running is slightly preferable to splitting.  Since we already escaped one man I think running to try and escape both makes more sense.


63.  I am guessing Split 24/18, 13/10  and Run are pretty close. 

Running 24/15 could be right. 
+ It plays to our main objective and
+ getting hit isn't the end of the world but ....
- it fails to unstack the midpoint
- even if missed we will have to use next turn cleaning up our blot, and
- even if missed we have some duplication of our own numbers, as rolls two apart both make an outside point as well as inside points.

Splitting 24/18, 13/10 could be right too.
+ Getting hit isn't the end of the world.  We don't lose too much in the race by getting hit on the bar.
+ Opponent is stripped on the 8-point and would prefer not to hit with an ace.
+ we unstack the midpoint to start out development
- we give 54 as an extra hitter to the opponent
- we leave more shots than by running

Slotting 13/4 feels like the wrong strategy.
We are behind in the priming battle - trying to escape seems more logical

All in all I like Splitting 24/18, 13/10 but I'm not sure.

62.

Split 24/18, 13/11.

Basically similar reasoning to 63 above.
With 62, we are up less in the race and get hit more which probably argues against running.
Splitting to the 11 point is a bit better spot for a builder, which is a plus, and also 54 doesn't hit this time.

So I am guessing Splitting is more clear here than with 63.

Run 24/16 is right.

61.  Make the bar point 13/7, 8/7.

54.  Two Down, 13/8, 13/9.  What else is there really?

53. Not sure about this one. 

Running 24/16 is always an idea since we are ahead in the race.  But I doubt it is right in this position since the other plays do something constructive.

Making the 3 point 8/3, 6/3 makes sense as this represents obvious improvement. 
+ Making a home board point
- Fails to unstack the midpoint

Two down 13/8, 13/10 is another possibility.
+ Develops the midpoint
- Fails to make a home board point
- Leaves 63 and 54 as (unnecessary) hitters

Making a point is for sure improvement so I will choose that.  8/3, 6/3.

52.

Two Down 13/8, 13/11 is my gut play.  Develop the midpoint and give yourself many new numbers to make a good point next roll - all for the rather small cose of a 64 hit.

Down and Up 24/22, 13/8 scoots the back man up a bit to try and escape later but there should be time to escape later.  Also, you expose yourself to 53 (55 is good for him whether you jump up or stay back)

51.

24/23, 13/8.  Scooting up with the ace isn't ideal, since we would prefer to use our sixes to escape but the alternatives just don'e make sense.  Slotting can't be right and running 24/18 is too desperate.

43.

Lots of options here.

Totally safe 13/7.  It's ugly but it preserves your assets (race lead, escaped back man) while leaving no shots and hoping for a better roll next time.

Two Down 13/9, 13/10.  This is the move you want to play if you know you will be missed next time.  You develop the midpoint and give yourself all sorts of rolls to make a new point next time.  But, you get hit on 8's (62, 53, 44) and 9's (63, 54, 33) so is the risk worth it?  Perhaps 44 and 33 shouldn't count since these are good rolls anyway, so let's say is 8 shots worth the risk? I am guessing not.

Up and Down 24/21, 13/9.  The halfway house play has some merit too.  We develop the midpoint while leaving less shots.  But stepping up into the attack zone (the 4 point) is waving a red flag to a bull - the opponent will happily attack you there.  I don't like this play.

42. Make 4 point 8/4, 6/4.

41. 

13/8.  Safe and a small improvement.  Given our race lead I think we should be conservative here.

24/23, 13/9. 
+ Unstacks the midpoint
+ diversifies rolls to make points next time
- leaves 8's as hitters
- takes away your escaping sixes next time (and into the future).

On balance, I see the wimpy 13/8 as a definite improvement while the unstack is a maybe improvement.

32.

13/8.  Similar reasoning to 41 above.  It is not the most sexy play, but it represents a small but for sure improvement.

Two Down 13/11, 13/10.  Again the play you want to make if you know you don't get hit.  But is the risk worth it?  9's (63, 54, 33) and 10's (64) hit but 33 is good already so really only six hitters.  This could be right but I still play wimpy.

Split 24/21, 13/11 doesn't make sense since you are inviting the attack for no reason.

Reverse Split 24/22, 13/10 is a bit better since at least opponent would rather attack on 4 point than 3 point, but it seems better alternatives exist so I reject this play.

31.  Make the 5-point 8/5, 6/5.

21.

24/23, 13/11 gives you the best builder but may give you some problems with sixes in the future.  On the bright side, 64, 63, 61 all make points while 62 you can still run, so only 65 is bad next time.  Also, note opponent's 64 is duplicated.

13/10 preserves the sixes but this builder is not as good as the 11 point builder, and you also give more shots to get hit.

66. 

13/7(4) takes advantage of the fact that you have 6 on the midpoint.  You leave no shots and make a new point.  Definite improvement.  i like it.

24/18, 13/7(3) makes a new point and attempts to escape - but I think you should have time to escape later. 

55.  13/3(2) seems forces.

44.  13/5(2) closing the 5-point is too good to pass up.

33.  There are other plays but I like 13/10(2), 6/3(2) which unstacks the midpoint and makes a home board point and leaves no shots.

22.  13/11(2), 6/4(2) seems pretty clear,

11.  I like 8/7(2), 6/5(2) which places tremendous pressure on the opponent to hit. But it leaves a blot so I will prefer 24/22, 6/5(2) instead.


************
Follow up on my errors, and other observations.

63.  My splitting play is a whopper with cheese, wrong by .15.  I guess the gameplan simply demands you run and hope for the best.  All the other considerations are overanalyzing.

62.  Well at least I was right the splitting with 62 was better than with 63, but only in the sense that it is less wrong.  Splitting is wrong by .08, and still bleeds red on XG.  Once again, it seems that once you are racing you gotta race and hope for the best.  Hopefully, next time I run.

53.  Making the 3 point is correct, but I was surprised that running is 2nd best, and beats two down.  This underlines for the me the importance of the running game plan, or conversely, emphasizes that because you have escaped one man you don't want to get hit.

52.  XG++ likes my play but I wanted to note that 24/22, 13/8 is very close - only .01 error.  I am starting to see a theme here.  Let's make it a rule of thumb.  When you have one man back to two men back you don't want to get hit!!  Tend towards the safe play.

43.  Two down was the right play by .03.  My totally safe play was 2nd best.  8 hits is a lot, but I guess you still want to take your risks while the game is young.

32.  My play 13/8 was right but this really confuses me.  Two down seems more logical here (at least to me) than with 43, since the risk is less (you get hit less).  Do any of my readers have ideas here?

21.  The two options 13/10 and 24/23, 13/11 are essentially tied.

66.  13/7(3), 24/18 is right by a lot.  I guess this an example of Pottle's Law which says "always run the back checker"

August 25, 2012

Lyurix

Below are selected descisons / errors from an online match against Lyurix.

**************
I trail 7 away 5 away and have 65 to play.  Do you stay or do you go?

is Lyurix

score: 2
pip: 129
                         
7 point match
                          pip: 139
score: 0

is Chutzpah
XGID=-B-BA-C---A-dC-b-cBcc---A-:0:0:1:65:0:2:0:7:10
to play 65

1.XG Roller+24/13eq: -0.186
Player:
Opponent:
42.67% (G:1.24% B:0.01%)
57.33% (G:1.48% B:0.04%)
2.XG Roller+13/8 10/4eq: -0.210 (-0.025)
Player:
Opponent:
41.82% (G:4.02% B:0.03%)
58.18% (G:4.93% B:0.18%)
3.4-ply24/18 13/8eq: -0.195 (-0.010)
Player:
Opponent:
42.18% (G:1.28% B:0.01%)
57.82% (G:1.50% B:0.05%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2

I stayed back, moving 13/8, 10/4, on the logic that I am behind in the race and my man on the ace point is usefully guarding the holes on Lyurix's inner board.  And this is true, as far it goes.

Here is my general thinking on this point (as of this moment only!!) --

  • Staying back on the ace point plays for maximum contact, and generates the most shots
  • The more open points in the home board, the more effective this lone back man is.
  • Staying back on the ace point gets you gammoned the most often.   Sometimes you can be pointed on, or hit loose and dance - in which case you could stay on the bar a long time and get G'balled.
  • The more points the opponent fills in, the more danger your lone back man is in.
  • A rule of thumb - when the opponent fills in all the points in front of the back blot it is too dangerous, and time to get the hell out of there if you can.
  • Keep an eye on the race - even if you are behind jumping up the the anchor can be still be correct if the race is a legitimate option.
 The best move in this exact position was 24/13.  The last point on the race seems to be the key factor here.  After the play I am only slightly behind in the race so racing is a legitimate option.  While I am behind, I have the midpoint contact plus the chance to get lucky and win the race.  In particular, I need to position myself so that 55 or 44 are truly excellent roles for me. 

****************
I trail 7 away 5 away and am on roll.  Cube action?

is Lyurix

score: 2
pip: 83
                         
7 point match
                          pip: 75
score: 0

is Chutzpah
XGID=-B-BBBCAC------a-b-cebb---:0:0:1:00:0:2:0:7:10
on roll, cube action?

Analyzed in 4-ply No double Double/Take
  Player Winning Chances: 73.29% (G:0.00% B:0.00%) 73.29% (G:0.00% B:0.00%)
  Opponent Winning Chances: 26.71% (G:0.00% B:0.00%) 26.71% (G:0.00% B:0.00%)
  Cubeless Equities +0.466 +0.945
Cubeful Equities
     No double:+0.750 (-0.062)
xg Double/Take:+0.812
     Double/Pass:+1.000 (+0.188)
 
Best Cube action: Double / Take
eXtreme Gammon Version: 2.02, MET: Kazaross XG2

I doubled and Lyruix took. Not too difficult. I posted it because I wanted to practice the various methods for estimating race cube actions.


Probably the easiest is just 10% + 2 pips. 

Per Doug Zare, being on roll with a 10% lead means you win about 75% of the time.  This applies for race of 50-120 pips. 

10% + 2 pips indicates the point of last take and 10% - 2 pips would be initital double point, 10% - 1 pip the redouble point.

10% of 75 is 7.5, so I can double with 5.5 pip lead and he can take up to 9.5 pips behind. 

Thus, Double / Take is the cube action.


Trice is similar. 

For races 62 pips or greater, point of last take is 10% of leader count + 1 pip,  round up.  Double within 3 pips of last take and Redouble within 2 pips.

For races of 62 pips or sammler, point of last take is (leader count - 5 ) / 7, round down.  Double within 3 pips of last take and Redouble within 2 pips.

I was always a bit confused on this, but I believe Trice also recommends using Ward adjustments for wastage.  Apply the adjustments to both sides, and then use the Trice formula to estimate cue actions.

** Ward Adjsutments **
+/- 2 pips for each extra checker on board
+ 2 pips for each extra checker on the ace point (more than 2)
+ 1 pip for each extra checker on the deuce point (more than 2)
+ 1/2 pip for each extra outside crossover
-1 pip for each additional occupied home board point

I wouldn't make any adjustments to this position, but would note that Lyurix has a better bearoff since I already have guys on my ace and his setup if more like the triangle.

So, 10% of 75 +1 is 9 pips.  Point of last take is 9 pips and I can double if I lead by six pips or more.  Double / Take.


Ballard Rule of 57.

This is just a more precise way to estimate Trice's last take for races of 57 pips or more.

Take Leader pipcount - 33, double it, and find the nearest square root.  This is point of last take.

75 - 33 = 44 X 2 = 88.  Sq rt of 88 is about 9.5.

So once again. Double / Take


Keith Count

Below image is from Phil Simborg, an excellent summary of the Keith Count method.



That said, I don't use it because I always get confused on applying the rules.

Leader: 75 + 2 = 77.  Add 1/7 or 11 = 88. 
Trailer: 83 (no adjustments). 

Therefore, No Double / Take per Keith (assuming I am doing it right - not sure)


EPC / Stick Method

Estimate EPC or effectice pip count for both sides.
EPC / 7 = # of Rolls to go
# of Rolls to go - 3 = Trailer's point of last take
Double withing 2 pip of Last Take
Redouble within 1 pip of Last take

I would have guessed wastage of 10 for me (which is wrong) or EPC of 75+10 or 85.
I would have guessed wastage of 7 for Lyurix (which is also wrong) or EPC of 83 + 7 or 90.

85 / 7 = 12, less 3 is 9.  So last take of 9 pips.
Lyurix can take with 85 + 9 = 94 pips
He has 90 which is more than 2 pips away from last take so No Double / Take using this method.

Again, not sure if I am applying this correctly.

**************
Score is 5 away 5 away and Lyurix is on roll.  Cube action?

is Lyurix

score: 2
pip: 136
                         
7 point match
                          pip: 153
score: 2

is Chutzpah
XGID=-aB---D-C---dDb--a-c-b-bAA:0:0:-1:00:2:2:0:7:10
on roll, cube action?

Analyzed in XG Roller+ No double Double/Take
  Player Winning Chances: 71.88% (G:30.47% B:1.15%) 71.63% (G:31.60% B:1.09%)
  Opponent Winning Chances: 28.12% (G:6.05% B:0.26%) 28.37% (G:6.42% B:0.29%)
  Cubeless Equities +0.672 +1.475
Cubeful Equities
     No double:+0.920 (-0.080)
     Double/Take:+1.250 (+0.250)
xg Double/Pass:+1.000
 
Best Cube action: Double / Pass
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
Lyurix doubled and I took. Oops, a double whopper with cheese. That will leave a mark.

Let's look at this a few different ways. 

First, what does PRAT suggest?

Position - I am outboarded 3 to 2 and have 2 men back to 1, with one in the air.  He has 10 men in the zone to attack with.  Large advantage Lyurix

Race - I am significantly down in the race.  Minus 17 pips and he is on roll and I am in the air.  Significant advantage Lyurix.

Threats - He has three builders working on the 5 point - I count 10 numbers that immediately make the 5 point next turn.  PLus again I have a man in the air.  Significant threats.

0 for 3 adds up to a clear drop. 

Besides this, Lyurix enjoys an elevated gammon value of .66 at this score - gammons are more valuable than for money.  This is extra incentive to drop, not that I needed any.

What do Stick's guidelines suggest?

Below are my notes from some early issues of Stick Daily Quiz Recaps - sadly these are being discontinued.

Stick Stat. Early game blitzes that are proper Double / take are about 65-68% wins and 25-28% gammon wins for the stronger side (Issue 1: DQ 4.1.11)

Caution!  Worst error in blitz is to underestimate gammon losses and take a monster pass.

Stick's Early blitz evaluation criteria: (Issue 2: DQ 4.8.11, Issue 4: DQ 4.22.11)
  • The race. I always like to note the race. It will be more important than some positions than others but it's a valuable detail to know and often plays a role in your decision.
  • How many checkers on the bar?  Big difference between no men up versus 1 man versus 2 men up.
  • Number of men in the zone. The key number that weighs in on the possible success of a blitz is 10 men in the zone.
    • If there are more than 10 men in the zone that's a lot of wood and you should be careful about taking any cubes.
    • If there are less than 10 men it's generally an easy take.
    • If there are exactly 10 take it on a position by position basis weighing in the other relevant factors.
  • Diversification of men in the zone. Clearly you'd rather have your men diversified than all stacked up on the same point.
  • Efficiency of men in the zone. Again, 10 is the magic number for minimum checkers you'd like to have in the zone but how they're put to work is important.
    • As a simple example in our position above we have 10 men in the zone in a standard looking blitz position.
    • Imagine if these 10 checkers instead made a prime from from the three point to the seven point. This would be light years better than the position we now have.
  • Status of the doubler's back checkers. If the checkers are split it's a big swing versus both of them still being on the 24 point. If one of them has already escaped it will make a big difference.
  • Additional points the person cubing has made. For each additional point the doubler has made, especailly inner board points, you have to be more and more careful if you're considering taking a cube.
    • A good rule of thumb is most (closer to all) positions where your opponent has a two point board or less are takes.
    • If he has a three point board or stronger beware. There are many exceptions to this statement of course but it's not a bad guideline.
  • Defender's offensive progress. Has the defender made any real progress on his side of the board? Any new points, esp. inner board points, will make his ability to counter punch while the attacker is trying to carry out the blitz more potent.
  • Other Tweak factors.
    • How “in the zone” are the 10 checkers? (the closer the better for the attacker)
    • What is likely to happen on the next exchange of rolls?
    • Match Score and gammon values.

So, looking at Stick's criteria - he has a 3 point board and 10 in the zone, which spells danger but not automatic drop.

Unfortunately the other factors are in Lyurix's favor:

  • I have an additional point, but it is the lame 2 point.  So this helps me some, but not too much.
  • Lyurix's 10 men in the zone are working well by aiming at the key 5 point.  As noted earlier, 10 rolls make the 5 point immediately.
  • Lyurix has already escaped one back man.  This is huge. 
  • Note Stick's stat that proper Double / take are about 65-68% wins and 25-28% gammon wins for the stronger side.  Here XG gives 72% wins and 30% Gammons - more than the guideline.  I guess we would have to infer this from the other factors.

In any case, this looks like a comfortable drop any way you look at.

**************
I trail 3 away 1 away Post crawford and have 55 to play. 

is Lyurix

score: 6
pip: 159
                         
7 point match
                          pip: 156
score: 4

is Chutzpah
XGID=-a----EaC---dFa--c-e----A-:0:0:1:55:4:6:0:7:10
to play 55

1.XG Roller+13/8(2) 6/1*(2) eq: +0.695
Player:
Opponent:
61.47% (G:21.48% B:0.46%)
38.53% (G:7.05% B:0.24%)
2.XG Roller+8/3(2) 6/1*(2) eq: +0.653 (-0.042)
Player:
Opponent:
58.79% (G:25.09% B:0.56%)
41.21% (G:8.27% B:0.37%)
3.2-ply13/3 8/3 6/1* eq: +0.539 (-0.156)
Player:
Opponent:
56.33% (G:19.24% B:0.61%)
43.67% (G:8.84% B:0.40%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2

I made the 'automatic' but incorrect play 8/3(2), 6/1*(2).

Better is 13/8(2), 6/1*(2).  This play strikes me as odd, but does crop up from time to time.  I don't fell I really understand what is going on.  Apparently you just don't want get hit so leaving a blot on the 8 point is simply wrong.

*************
DMP.  51 to play.

is Lyurix

score: 6
pip: 125
                         
7 point match
                          pip: 103
score: 6

is Chutzpah
XGID=--a-CCCBB---d-B--a-cbbb---:0:0:1:51:6:6:0:7:10
to play 51

1.XG Roller+6/1 5/4eq: +0.544
Player:
Opponent:
77.21% (G:10.41% B:0.09%)
22.79% (G:2.65% B:0.09%)
2.XG Roller+14/13 14/9eq: +0.474 (-0.070)
Player:
Opponent:
73.70% (G:12.75% B:0.35%)
26.30% (G:5.91% B:0.25%)
3.4-ply14/8eq: +0.461 (-0.083)
Player:
Opponent:
73.06% (G:14.90% B:0.43%)
26.94% (G:3.88% B:0.11%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2

I elected to Pay now, moving 14/9, 14/13 and duplicating ones.  This was the wrong idea, and the better play was the ugly but safe 6/1, 5/4.

I guess the goal is simply to come home safe and win the game.  Therefore, safety is a premium and you don't want to leave a shot if you don't have to.  Here, the roll plays, sort of ugly, but it plays.  Next time, if I roll two bigs I can simply jump past his midpoint while if I roll an ace or a duece I can play safe once more and wait for a better roll to either clear or attack his blot.

************
Still DMP.  Next roll.  How to play 41?

is Lyurix

score: 6
pip: 117
                         
7 point match
                          pip: 97
score: 6

is Chutzpah
XGID=--a-CCCBBA--bAa--aacbbb---:0:0:1:41:6:6:0:7:10
to play 41

1.XG Roller+13/9 5/4eq: +0.694
Player:
Opponent:
84.72% (G:8.28% B:0.06%)
15.28% (G:0.86% B:0.02%)
2.XG Roller+13/9 4/3eq: +0.577 (-0.117)
Player:
Opponent:
78.87% (G:10.71% B:0.32%)
21.13% (G:4.68% B:0.28%)
3.4-ply13/8eq: +0.708 (+0.014)
Player:
Opponent:
85.41% (G:15.74% B:0.37%)
14.59% (G:0.92% B:0.02%)
eXtreme Gammon Version: 2.02, MET: Kazaross XG2
I completed the six prime with 13/9, and slotted the next point with 4/3.

Same theme as prior roll applies here.  My move looks sexy but takes chances for no reason.  Sure, most likely, Lyurix misses, or he hits and my prime holds up and I won anyway - but why take the chance when you don't need to?

Better was 13/9, also completing the full prime and then simply 5/4.  Sure, the builders are a bit ugly but who cares?  All I need to do is roll home safe and win the game.