1. First game of 11 point match. I have doubled and have this 43 to play.
is Roberto score: 0 pip: 121  
11 point match  
pip: 109 score: 0 is Bill  
XGID=BbCDBBcBccbb:1:1:1:43:0:0:0:11:10  
to play 43 
1.  XG Roller+  18/11  eq: +0.082  
 
2.  XG Roller+  8/5 8/4  eq: 0.002 (0.083)  
 
3.  3ply  6/2 5/2  eq: +0.025 (0.057)  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
I went with 8/5, 8/4 on the logic of placing the blot in the best possible location assuming I am missed. Also I play with only one blot, leave less shots, and retain the anchor in case I am hit. All of these concepts have some merit but are misapplied here.
Best was making a break for it with 18/11. Compared to my play this looks especially risky. I leave two blots and expose myself to 17 immediate hits instead of 11. How can this be right?
My current thinking:
 I am up in the race so racing should be my gameplan. Breaking contact is good if I get away with it.
 If I am hit I retain a broken four prime to fight against his anchor. Roberto will still have a bit of work to do to escape my prime before he can cube me out.
 The six to hit and escape is duplicated.
2. Very next roll. Your 53 to play.
is Roberto score: 0 pip: 115  
11 point match  
pip: 102 score: 0 is Bill  
XGID=BbADDBcBbbbbb:1:1:1:53:0:0:0:11:10  
to play 53 
1.  XG Roller+  6/1 4/1  eq: +0.040  
 
2.  XG Roller+  7/4 7/2  eq: 0.005 (0.044)  
 
3.  2ply  7/4 6/1  eq: 0.128 (0.168)  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
I made the best board with 7/4, 7/2.
While a stronger board will be useful in the event I am lucky enough to score a hit, I am more likely to be squeezed and offer up the first blot. In particular I will be forced to run with the next six which could be deadly.
Best was 6/1, 4/1. The obvious concept is to save a six.
Another small point is that holding the bar makes it more difficult for Roberto to break anchor and move both men to the outfield.
3A. Disaster has struck. I have been forced out with a six. Recube action?
is Bill score: 0 pip: 77  
11 point match  
pip: 87 score: 0 is Roberto  
XGID=ACABCCAbbAbabbbb:1:1:1:00:0:0:0:11:10  
on roll, cube action? 
Analyzed in XG Roller+  No redouble  Redouble/Take 
Player Winning Chances:  70.21% (G:4.30% B:0.15%)  70.10% (G:3.87% B:0.10%) 
Opponent Winning Chances:  29.79% (G:4.24% B:0.08%)  29.90% (G:4.23% B:0.09%) 
Cubeless Equities  +0.406  +0.815 
Cubeful Equities  
No redouble:  +0.585 (0.045)  
Redouble/Take:  +0.631  
Redouble/Pass:  +1.000 (+0.369)  
Best Cube action: Redouble / Take 
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
Roberto redoubled and I dropped.
I got scared over the board and viewed my chanced pessimistically. As the computer shows, the win chances are only 70% to 30% with low gammons (less than 5%). Acutally no gammons since I gammon him the same as me.
Now in the calm of my office, can I figure out why this is a proper take?
Let's break down the rolls into categories: Hit and cover, Hit and not cover, Miss.
 Hit and cover  5 rolls (52, 62, 22)
 For simplicity I will give myself 0% win chances after this
 Hit and not cover  23 rolls (66, 65, 64, 63, 61, 55, 54, 53, 51, 42, 41, 32, 33)
 Looking into some of these rolls, Roberto is supposed to hit and lift. This is not entirely obvious and may be an additional reason to take as my checker plays are easier than Roberto's.
 The computer gives me about 25% win chances on average after these variations  good enough for 5 wins and change in 36 games.
 Miss  8 rolls (44, 43, 31, 21, 11).
 My win chances in these variations are slightly better than 50/50 on average  good enough for another 4 wins and change in 36 games.
That's 9 wins out of 36  good enough for a take already. Throw in a tiny bit for cube vig and I can get to about 10 wins in 36 (not matching the 30% per the computer but good enough to get us to take land)
Probably the most difficult part of this exercise is evaluating the win chances in the hit and not cover scenario.
3B. For reference, here is a variation where Roberto hits and lifts with 54 and then I enter with 53. What are my win chances?
is Player 2 score: 0 pip: 78  
11 point match  
pip: 84 score: 0 is Player 1  
XGID=ACABCCAbaabbbbcA:2:1:1:53:0:0:0:11:10  
to play 53 
1.  XG Roller+  Bar/20 6/3  eq: 0.076  
 
2.  XG Roller+  Bar/20 8/5  eq: 0.176 (0.100)  
 
3.  3ply  Bar/20 4/1  eq: 0.186 (0.109)  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
This seems about average. If I come in the first roll I have a fighting chance, say 3540% to win. Putting this together with above analysis ....
Roberto hits and doesn't cover on 23 rolls. Then I immediately enter 1/3 of the time (over estimate) and win those 1/3 of the time (underestimate) then as a rough calculation that I actually have a chance at over the board ... gives me about 2.5 wins out of 36 games.
3C. Another variation for reference. Robert hits and lifts with 54 and I dance. What are my win chances here?
is Player 2 score: 0 pip: 78  
11 point match  
pip: 84 score: 0 is Player 1  
XGID=ACABCCAbaabbbbcA:2:1:1:64:0:0:0:11:10  
to play 64 
1.  3ply  Cannot Move  eq: 0.589  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
This is obviously very bad but still not hopeless. If I gave myself another 20% wins (overestimate) on the 70% of time I dance (overestimate) then I get another 3 wins or so on the variations where Roberto hits and lifts and I dance. That is actually pretty significant. More than I would have thought.
Summarizing this bit of math ...
 I get about 5.5 wins when Roberto hits and lifts
 2.5 when I immediately enter and another 3 when I dance
 Add to this I am slight favorite when Roberto misses and I am at 9.5 wins  enough for a take.
 Give me something for cube vig and we can get perhaps over 10 wins.
4. I trail 10 away 9 away and am being doubled. Take or Pass?
is Bill score: 1 pip: 93  
11 point match  
pip: 107 score: 2 is Roberto  
XGID=ABCBCBaacbbbbBb:0:0:1:00:1:2:0:11:10  
on roll, cube action? 
Analyzed in Rollout  No double  Double/Take 
Player Winning Chances:  73.02% (G:10.38% B:0.27%)  73.48% (G:10.52% B:0.28%) 
Opponent Winning Chances:  26.98% (G:3.73% B:0.07%)  26.52% (G:3.89% B:0.08%) 
Cubeless Equities  +0.529  +1.079 
Cubeful Equities  
No double:  +0.899 (0.024)  ±0.006 (+0.894..+0.905) 
Double/Take:  +0.924  ±0.010 (+0.914..+0.933) 
Double/Pass:  +1.000 (+0.076)  
Best Cube action: Double / Take  
Rollout details  
1296 Games rolled with Variance Reduction. Moves: 3ply, cube decisions: XG Roller  
Double Decision confidence:  100.0%  
Take Decision confidence:  100.0%  
Duration: 3 minutes 20 seconds 
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
I dropped. My instinct or "mental reference" is a 3 point anchor against a 5 prime is always a drop, regardless of the race.
Obviously this isn't true. Computer has this as a relatively comfortable take.
 Playing around with the position, one detail that seems to matter is that Roberto has the 2 point made. Thus he won't have as many active builders to attack me when I eventually pop out with a six.
 Also I have a good board at the moment and a bit of a race lead. If do throw a six quickly Roberto will have to respect my home board. He can't just hit loose with no fear.
 Gammons are relatively low at about 5%
So my new "current" thinking is I can take a primed 3 point anchor position as long as the above conditions are true.
 I have a good board
 I have a modest race lead
 Stronger side lacks some active builders (checkers are past your anchor)
5. 8 away 8 away. Early game 54. Your move?
is Roberto score: 3 pip: 160  
11 point match  
pip: 159 score: 3 is Bill  
XGID=aaECAdDacAeA:0:0:1:54:3:3:0:11:10  
to play 54 
1.  XG Roller+  24/20 18/13  eq: 0.051  
 
2.  XG Roller+  18/13 8/4*  eq: 0.052 (0.001)  
 
3.  XG Roller+  18/9  eq: 0.140 (0.089)  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
I made the putrid 18/9. I think I knew this was bad even at the time but did this anyway. I was probably trying to hold onto my race lead.
Best by far is 18/13, 8/4*
The lesson? When you get a shitty roll in the early game, you need to hit to take away half the roll. It's great if you get away with it and if you don't the game simply continues. Both sides only have a 1 point board. getting hit isn't so bad.
Perhaps most important by not hitting I give Roberto the freedom to play his entire roll. He just has too many good things that can happen and almost no bad rolls. He can hit, make a defesive anchor or make an offensive point with almost any shake.
6. I lead 5 away 8 away, have doubled and am in the late stages of a blitz. Lift or leave the blot with this 32?
is Roberto score: 3 pip: 154  
11 point match  
pip: 84 score: 5 is Bill  
XGID=bBBBBADcAbbeAa:1:1:1:32:5:3:0:11:10  
to play 32 
1.  XG Roller++  20/18 5/2  eq: +1.622  
 
2.  XG Roller++  15/10  eq: +1.619 (0.003)  
 
3.  XG Roller++  15/13 5/2  eq: +1.618 (0.004)  
 
4.  XG Roller++  20/15  eq: +1.607 (0.015)  
 
5.  XG Roller++  20/18 6/3  eq: +1.529 (0.093)  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
This is actually a bit of a trick question since it turns out it doesn't really matter. Leaving or lifting plays are all about the same as long as you don't make a half ass play like I did.
I moved 20/18, 6/3. I can't explain it, except to say this is obviously bad. I am not sure what my thought process was here.
Generally speaking Roberto is fucked whether or not I cover, which argues for simply lifting.
On the other hand, Roberto has a one point board with a blot so there is very limited danger in going for the closeout. Which argues for leaving the blot. If leave the blot, then 15/10 seems the most logical play. This generates the most offensive, whether I get hit or not.
7. Later in the same game. As you can see I didn't manage to close Roberto out. How best to handle this 53?
is Roberto score: 3 pip: 119  
11 point match  
pip: 78 score: 5 is Bill  
XGID=BECBaAcbbcBbb:1:1:1:53:5:3:0:11:10  
to play 53 
1.  4ply  21/13  eq: 1.000  
 
2.  4ply  21/18 7/2  eq: 1.000  
 
3.  4ply  7/2 4/1  eq: 1.000  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
This is another trick question. It really doesn't matter since I should be dropping after any play.
I thought for a long time on this one before eventually making the inferior 7/2, 4/1.
Even though this is the worst play per the computer I nevertheless think this is the most practical play. I have a better chance that Robert won't double me with no blots to shoot at. As it turned out, Roberto didn't in fact double and I later pulled out a miraculous win.
8. I lead 3 away 8 away and hold a two cube. Cube action?
is Roberto score: 3 pip: 66  
11 point match  
pip: 78 score: 8 is Bill  
XGID=aABCCBBAAacccd:1:1:1:00:8:3:0:11:10  
on roll, cube action? 
Analyzed in XG Roller++  No redouble  Redouble/Take 
Player Winning Chances:  93.10% (G:4.23% B:0.06%)  93.29% (G:3.60% B:0.04%) 
Opponent Winning Chances:  6.90% (G:0.60% B:0.00%)  6.71% (G:0.77% B:0.01%) 
Cubeless Equities  +0.880  +1.310 
Cubeful Equities  
No redouble:  +0.929 (0.071)  
Redouble/Take:  +1.003 (+0.003)  
Redouble/Pass:  +1.000  
Best Cube action: Redouble / Pass 
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
Should I rewhip and play for the match from here?
Roberto's takepoint is very low, about 7%. Does he have it? How the hell do you judge what 7% looks like?
I wasn't sure so I cautiously rolled.
I violated Woolsey's law! If unsure of a take, then simply double and let your opponent worry about it. Generally speaking I was timid with the cube the entire weekend at Novi. This is one of my big takeaways from the tournament. I was wimpy both ways  doubling too late and dropping too early. Not a good combination and something for me to work on. The one and only comment Victor Ashkenazi made after my heartbreaking loss to him was that I played like a pussy with he cube, or words to that effect. Sage advice!
I ended up losing this match, even though I had significantly better the luck factor per the computer. Perhaps if I had recubed here the result would have been different.
9. I lead 1 away 5 away post crawford and have just rolled a joker boxes. How best to make use of this great roll (66)?
is Roberto score: 6 pip: 143  
11 point match  
pip: 139 score: 10 is Bill  
XGID=aaBCBCAcBacbBbb:0:0:1:66:10:6:0:11:10  
to play 66 
1.  XG Roller++  8/2*(2) 7/1*(2)  eq: +0.567  
 
2.  XG Roller++  20/14(2) 13/7(2)  eq: +0.426 (0.142)  
 
3.  3ply  20/8(2)  eq: +0.416 (0.152)  

eXtreme Gammon Version: 2.10, MET: Kazaross XG2
This was my last key decision of the match and I blew it.
I figured I didn't need a gammon to win so simply ran around the board with 20/14(2), 13/7(2). My distribution looks sexy and I shouldn't have too much trouble bringing this one home provided Robert doesn't hit this turn. Or so I thought.
All true but closing out two (or even three) men is the simplest way to win. After 8/2*(2), 7/1* (2) Roberto is really behind the 8 ball . He has 2 men on the bar against a four point board and I am shooting at a third blot.
Roberto will find it very difficult to make an anchor and even if he does, I will likely be in a favorable holding game at that point anyway, retaining large winning chances no matter what happens. The computer shows the blitz play wins about 7% more often.
This blunder probably cost me the match.
Nice analysis of your weekend.
ReplyDeleteGood positions and analysis. A few comments:
ReplyDelete1. Running off the barpoint anchor this late in the game is hard to do safely and sometimes you are even forced to leave a double shot. You're out of timing and getting away with a single direct shot is not a bad deal, so take it.
5. One thing I've noticed in your analysis is that you frequently consider only your own play and XG 3ply's top play. But you should also consider other plays that the computer thinks are close. Here, 24/20 18/13 is perfectly reasonable so you can't conclude that hitting is "best by far" and that you "need to hit".
Good tip Tim. Thanks for the comments. Hope to see you at the next tournament.
ReplyDelete