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

BGonline.org Forums

Excessive criticism?

Posted By: Matt Cohn-Geier
Date: Thursday, 15 April 2010, at 2:39 p.m.

In Response To: Excessive criticism? (Chase)

I'll try to reconstruct the process I went through OTB. Actually I think I came up with 16% for our "perfect" recube point rather than fully live which is rather high I guess and might explain the rest of the problems.

  • 14-away Neil's # is 4; 8-away is 6; 10-away is 5.5

    If we pass we are at -14 -6 = 8x 4 = 18%
    If we take and lose we are at -14 -4 = 10x 4 = 10%
    If we take and win we are at -10 -8 = 2x 5.5 = 39%
    If we take and win 8 we are at -6 -8 = 2x 6 = 62%
    If we lose 8 we lose the match.

  • Cubeless takepoint: risk = 18 - 10 = 8; gain = 39 - 18 = 21; 8/29 works out to 27.6%. I have no idea how I came up with 24% here but I was almost certain that our normal takepoint on a 4-cube couldn't be higher than money 25% so if I had come up with 27.6% I would have tossed it out the window and note that I know Neil's #s don't always work for lopsided scores so the problem is likely with -14 -4.

  • Auto recube takepoint: we risk 18; we gain 62 - 18; 18 / 62 = closer to 1/3 than 1/4. Since this is higher than our cubeless takepoint I can throw this out the window.

  • Perfect recube: suppose we risk only 8 but always win 8 points. In that case we risk 8 but gain 62 - 18 = 44; so 8/52 = something close to 15.5% (16/104)

I guess when I posted looking at this in retrospect I mixed up what I figured our 'perfect recube' TP was OTB with what GNU reports the fully live TP is. I figured our takepoint was about halfway in between our cubeless TP and perfect recube TP (i.e., in somewhat more than half of the games we would reach an efficient redouble). That works out to 22% with Neil's #s but since I had thrown out the 27.6% number and was working with 24-25% I figured about halfway between 15-16% and 24-25% was 20%. I threw out our automatic recube TP because it was higher than the cubeless TP.

Let's look at what the #s really are:

    On g11:

    -14 -6 is 16.2%
    -14 -4 is 8.8%
    -10 -8 is 39.5%
    -6 -8 is 61.8%

  • Cubeless:

    Risk is 7.4%, gain is 23.3%; TP is 24.1%

  • Automatic recube:

    Risk is 16.2%, gain is 45.6%; TP is 26.2%

  • Perfect recube:

    Risk is 7.4%, gain is 45.6%; TP is 13.9%

Halfway between 24.1 and 13.9 is 19%.

    On R-K:

    -14 -6 is 15.3%
    -14 -4 is 8.2%
    -10 -8 is 39.6%
    -6 -8 is 61.7%

  • Cubeless:

    Risk is 7.1, gain is 24.3, TP is 22.6%

  • Automatic recube:

    Risk is 15.3, gain is 46.4, TP is 24.8%

  • "Perfect" recube:

    Risk is 7.1, gain is 46.4, TP is 13.3%

Halfway between 13.3% and 22.6% is about 18%. In reality in this position we can take with something probably just under 17% (maybe even 16.5%).

Basically there were a large combination of problems here: Falafel quickly felt strongly that it was a pass; I don't have much experience with recubes in matches longer than 11 points; Neil's #s led to something way off the right #s at 14-away; possibly I did the math wrong but I used some judgment to push it as low as I could using Neil's #s and my methods, which was 20%, but Mochy said he thought it was even higher; Stick also argued for the pass and noted the fact that our board would crash which led me to believe I might have overestimated recube vig; as it turned out even if I had worked out everything right I had still underestimated recube vig.

And, of course, in spite of all this, I still felt we had a razor thin take, but I wasn't one to argue for it at that point.

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.