[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Non-doublet Replies: Five rules for beginners

Posted By: Axel Reichert
Date: Saturday, 22 April 2017, at 3:26 p.m.

In Response To: Non-doublet Replies: Five rules for beginners (nack ballard)

Hi Nack,

for checking, I took the word straight from the horse's mouth, http://www.extremegammon.com/Openingbook.aspx. I assume that these are rollouts, not evaluations.

In nit-picking mode, your rule set according to #5 would advise 61P-62D, 61P-63D, and 61P-64D, all of which are blunders > 0.08. So #5 needs more clarification or, like it did it, more vagueness ("most similar legal alternative"). This also allows for some leeway, e.g., to use Z as a stand-in for illegal S and vice versa (where D might be slightly weaker). By the way, for your latest proposal the largest error is NOT 32Z-64P (-0.0546), because according to #2 you will correctly play 13/3*. (-:

However, I am not sure about my interpretation of the "R" in "CRMU", could you please clarify?

  • If "R" means to run beyond the bar point, CRMU results in 63R-63 being played 13/10* 13/7 (not sure about the Nactation), 64R-62 being played 13/11* 13/7, 64S-64 being played 13/9 13/7*, and 64S-63 being played 13/10 13/7*, since the leftover gets played from the mid. All these replies are blunders from 0.060 to 0.077.
  • If running just TO the bar point already qualifies as "R", CRMU results in 41$-42 and 51$-42 being played 24/20*/18. Both replies are blunders, 0.069 and 0.099 (GNU Backgammon rollout for the latter, since no XG data is available).

With the first interpretation your proposal results in a total equity loss (for the 15*27 = 405 non-doublet replies) of 1.9411. 91 replies are not the best ones. I did not check for the second interpretation, because it seemed less likely and I wanted to save the work and ask you first about CRMU.

When comparing your various proposals I noticed that you shuffled around the sequence of rules. I struggled with the same problem, e.g., is it better to use "Hit > 4, then P" or "P, then Hit > 2"? Then it occured to me that an elegant solution might be:

  1. Hit > Potential Point (or P)

Which means that if you have a point-making roll, you should prefer to hit up to that point. Hence 31P will not be abused to hit on 3, whereas 64 will happily be played 13/3*. If there are no blots to be scooped up, the point will be made as usual. This avoids compromise solutions such as "Hit > 4", exceptions like "x1$-31H", or having to treat the P rolls separately. Since I like the exceptions as sub-rules, separated from the general guidelines, we have:

  1. Hit > Potential Point (or P)

    1. D-64R

  2. Hit > 2
  3. If hit from mid: S (or D). If hit from non-mid: D (or S)
  4. 65R, 43Z, S (or most similar legal alternative)

    1. D-x1$

    2. 6xS-65K

This rule set results in a total equity loss of 1.8185 and 85 non-best replies. No blunders > 0.06. The largest errors are (again not sure about the Nactation details):

  • 63S-31 [P Z H-0.0595]
  • 52S-51 [S R X-0.0582]
  • 41$-64 [R Z-0.0567]
  • 64S-52 [X S-0.0557]
  • 64S-54 [X S-0.0557]
  • 62S-31 [P Z H-0.0534]
  • 51$-64 [R Z-0.0527]
  • 51S-41 [K D U S-0.0511]

Blitz afficionados could even skip exception 4b and use

  1. Hit > 2 (and kill connected points)

This is somewhat weaker, but surprisingly little, since e.g. it loses on 32S-32, but gains on 32Z-32.

By the way, my feeling is that getting below 0.04 requires too many additional rules, my four above get enough bang for my bucks.

Best regards

Axel

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

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