[BLML] implementing 12C1c (weighted adjustments)

Herman De Wael hermandw at skynet.be
Thu Mar 3 10:19:56 CET 2016


I knwe there would be some things I'm missing by not attending Prague.
Congratulations EBL (and WBF).
And I fully agree with the methods as descibed by Peter. That is not so 
strange - we have been known to be in agreement before (at least once, 
that I recall).

:)

Herman.

Herman De Wael schreef:
>
>
> Timothy N. Hill schreef:
>> The ACBL changed its method of assigning adjusted scores, effective
>> January 1, from 12C1e (most favorable/unfavorable) to 12C1c
>> (probability-weighted). We don’t yet have any software support for
>> 12C1c adjustments.
>>
>
> congratulations!
>
>> I have a few questions for old hands with 12C1c.
>>
>> Consider this example: A pair stops in game after receiving
>> misinformation and scores 680, but we estimate they had a 50% chance
>> of bidding slam if correctly informed. The other scores are -100,
>> 650, 680, 710, 1430, and 1460.
>>
>> The ACBL suggests we slightly favor the non-offending side in
>> assigning weights. Is this the practice elsewhere?
>>
>> Say we weight 680 at 40% and 1430 at 60%. Manually calculating the
>> adjusted matchpoints is easy enough: 40% x 2.5 + 60% x 4.5 = 3.7.
>>
>> How should the other scores be matchpointed?
>>
>> I think the other scores should also be weighted:
>>
>> -100                           0 650                           1 680
>> 40% x 2.5 + 60% x 2   = 2.2 710   40% x 4   + 60% x 3   = 3.4 40%
>> 680, 60% 1430   40% x 2.5 + 60% x 4.5 = 3.7 1430   40% x 5   + 60% x
>> 4.5 = 4.7 1460                           6
>>
>> Is this the preferred method elsewhere?
>>
>
> It is (equivalent to) my preferred method. But I know of no regulation
> that adopts this approach, nor of any program that supports it.
>
>> Unfortunately, the current ACBLscore won’t do this calculation. For a
>> small top, we could matchpoint the whole board manually, but that’s
>> impractical for a large top.
>>
>> If we enter the adjusted score as 3.7 matchpoints, the current
>> ACBLscore will Neuberg the other results:
>>
>> -100               0 +  1/12 = 0.08 650               1 +  3/12 =
>> 1.25 680               2 +  5/12 = 2.42 710               3 +  7/12 =
>> 3.58 40% 680, 60% 1430   40% x 2.5 + 60% x 4.5 = 3.7 1430
>> 4 +  9/12 = 4.75 1460               5 + 11/12 = 5.92
>>
>> How do other scoring programs support 12C1c adjustments?
>>
>
> Similarly.
>
>> Is there a syntax for entering “40% of the matchpoints for 680 plus
>> 60% of the matchpoints for 1430”?
>>
>
> Only the "Maastricht method" that David Stevenson and myself developed
> in 2000, and which states that the scores be listed in descending order
> of NS value, with percentages or fractions, and with a plus between all
> (I mean the word plus).
>
> Herman.
>
>> Thanks!
>>
>> Tim _______________________________________________ Blml mailing
>> list Blml at rtflb.org http://lists.rtflb.org/mailman/listinfo/blml
>>
> _______________________________________________
> Blml mailing list
> Blml at rtflb.org
> http://lists.rtflb.org/mailman/listinfo/blml
>


More information about the Blml mailing list