
BGonline.org Forums
Live Cube TP (Gammonless / Match)
Posted By: phil simborg
Date: Thursday, 5 June 2014, at 3:31 a.m.
In Response To: Live Cube TP (Gammonless / Match) (Jeremy Bagai)
With coaching from john o'hagan, I don't use either the dead or live cube take points (except on the last roll or a dead cube I use the dead cube take points) and I don't teach the use of the live cube take points in XG at all.
Instead I use what we call "the effective live cube take point" which is approximately 68% of the difference between the two. For money you take 25 and subtract 20 and multiply by 68 and get about 3.5 and you subtract that from 25 and that gives you 21.5, and that is what we call the ELCTP for money. This number is very accurate and useful for long races and appears to be very close to what XG uses for long races. We do the same thing for match scores but realize that we must do some additional tweaking for recube Vig, and at certain scores where you can redouble and totally wipe out gammons by making the value 0 there is even bigger recube vig and we have formulas for that as well.
The next version of XG will have a column between the live and dead cube take points that will show the ELCTP at each score.
Going to your example, at 5a/5a I use 20 percent as the TP. To decide whether to take or pass a cube I estimate the net gammons, multiply times .655, and add that number to 20 and that is the Gammon Adjusted Take Point.
John does it differently, and it is a lot more complicated, but more accurate. He takes the gammons and adds them to the dead cube TP and then subtracts cube vig which is a function of the number of gammons and the recube take point and a lot of math over the board. When he's done, for most scores, his numbers are very close to what I get, so even he concedes that my way is probably more practical and useful for most players. We also know that XG doesn't use either John's method or mine, but a blended method to come up with recube vig that is too complicated, even for John to calculate over the board.
The points is that we come very close to the right numbers and decisions in longer races and at most positions at most scores without having to do any complicated math over the board.
Now, my way can be very far off at certain scores...scores where you are trailing and you get doubled and you can get gammoned a lot, but your opponent is 3 or 4 away and you can redouble and make your gammons worth 0 and especially at scores where you can redouble and go to DMP or close to DMP. (4a/3a, 5a/3a, 6A/4A, and even 4A/4A where you can redouble and make gammons worth 0).
At those scores you can always take, no matter how many gammons there are, if your wins are greater than your MWC if you drop. (You could take and recube immediately and be better off than passing...however you wouldn't recube immediately unless you have serious market losers, which is rarely the case.) Even if your winning chances are slightly below your MWC if you drop, you can take because of cube vig and cube efficiency, not to mention the possibility of human error by your opponent on the recube. But if you have a lot of gammons and you can redouble and make it DMP and wipe out the gammon costs, then you have huge recube vig, and if you use my method, my take points and add the gammon values, you will be dropping cubes you should take. So we have developed an easy to use adjustment for those scores which varies depending on the score. At some you can reduce the GAT by a huge number (up to about 10 percent) if there are lots of gammons, and again, we have developed a formula for that which we are currently testing with rollouts, but it seems to get us pretty close to XG's decisions most of the time. (When I say "we" I mean two of my students and I have been developing and testing this and sharing it with John to see if he agrees that our numbers truly work, and again, they seem to work very well so far.) I am not prepared to provide those adjustments because they haven't been tested enough yet, and I'd like to have a year or two to use them to beat Carter more.
So, to save you a little time, our effective live cube take points are as follows, keeping in mind that we have rounded them all off to make them easier to use and remember over the board:
5A/5A 20 percent 5A/4A 19 (Leader) 24 (Trailer) 5A/3A 19(L) 21(T) 5A/2A 17 (L) 23 (T) 4A/4A 22 4A/3A 21(L) 25(T) 4A/2A 20(L) 19 (T) 3A/3A 25 3A/2A 26(L) 25(T) 2A/2A 32
Except where the live and dead cube TP's are the same or where you have a dead cube when you take the cube, the above numbers are pretty much a result of taking 68% of the difference between dead and live. Again, you will find, if you test these, they work perfectly for races over about 70 pips and you learn to raise the take points as your race shortens and you have less opportunity to get in a redouble. And you lean to adjust them somewhat if there are lots of gammons (although our new formula adjusts the GV's rather than the take points because it is an easier, more accurate adjustment that gets us a result closer to what XG says.)
I hope this is helpful. If not, I am $60 an hour for lessons and I will give yo the full lesson on using the above and the adjustments and how to do it quickly over the board. Any of my students who have taken lessons from me in the past 8 years can tell you this is relatively easy to do and it works.
I also have tables up to 9 point matches and some over the board shortcuts and tricks to remember them, along with over the board shortcuts and tricks to figure and remember gammon values, and I have a similar chart for Recube Take Points at all scores.

BGonline.org Forums is maintained by Stick with WebBBS 5.12.