Hand evaluation (was Re: Losing trick count)
David
<38D6BD32...@home.com>...
>
>
> al...@my-deja.com wrote:
>
> > This is particularly important to avoid LTC overevaluation of
> > three-suited hands.
>
> Of course three-suiters are overvalued in most evaluation methods. You
> always need to make a negative adjustment for 4441 patterns, whether
> it's 2 points, half a loser, or whatever.
It is important to recognise that most mechanical methods of hand valuation
over-value all patterns with three suits the same length: 4333, 4441 and
7222. (I presume the same applies to 10111 hands but I don't have any
experience with that shape:-))
Note that CCCC (the Kaplan-Rubens count) subtracts 0.5 points for 4333
shapes but not for any of the others. I think it should, and I use such an
adjustment in my valuation metric for simulations.
I also differentiate between hands with two doubletons and those with a 3-1
distribution of side suits: CCCC, like many other methods, counts these as
equal but my studies and those of others such as Alex Martelli show that
there is a statistically significant difference (to support the experience
of experts -- see for example Brian Senior's book on Hand Evaluation).
David
--
David Morgan
dmo...@webone.com.au
Samuel Ieong
: I also differentiate between hands with two doubletons and those with a 3-1
: distribution of side suits: CCCC, like many other methods, counts these as
: equal but my studies and those of others such as Alex Martelli show that
: there is a statistically significant difference (to support the experience
: of experts -- see for example Brian Senior's book on Hand Evaluation).
:
: David
I believe that CCCC gives 3-1 2 points (for singleton), but gives 2-2 only
1 point (first doubleton is not counted). Of course I could be
mis-remembering ...
Regards,
- Samuel
jan kamras
David wrote:
> It is important to recognise that most mechanical methods of hand valuation
> over-value all patterns with three suits the same length: 4333, 4441 and
> 7222. (I presume the same applies to 10111 hands but I don't have any
> experience with that shape:-))
And what abt 13-0-0-0? :-)
> I also differentiate between hands with two doubletons and those with a 3-1
> distribution of side suits:
I've never done any simulations, but from practical experience I find
there is, a priori, a *huge* difference between, say, 5422 and 5431. Of
all "normal" patterns, I've found 5431 to give the best combination of
flexibility both in bidding and play.
Alex Martelli
news:8bf0gr$b78$2...@news.ycc.yale.edu...
> David (dmo...@webone.com.au) wrote:
> : I also differentiate between hands with two doubletons and those with a
3-1
> : distribution of side suits: CCCC, like many other methods, counts these
as
> : equal but my studies and those of others such as Alex Martelli show that
> : there is a statistically significant difference (to support the
experience
> : of experts -- see for example Brian Senior's book on Hand Evaluation).
> :
> : David
>
> I believe that CCCC gives 3-1 2 points (for singleton), but gives 2-2 only
> 1 point (first doubleton is not counted). Of course I could be
> mis-remembering ...
I believe a 3-1 is also meant to be evaluated at 1 point, and online
versions of CCCC seem to agree. The "discount first doubleton" tag
is apparently meant to discount 1 point for shape too, not to make
e.g. a 6-4-2-1 evaluate the same as a 5-4-3-1, etc.
It sure WOULD be nice if Jeff Rubens published a "reference"
version of a program to compute CCCC...!
Alex
Alex Martelli
news:38DB2B20...@home.com...
[snip]
> > I also differentiate between hands with two doubletons and those with a
3-1
> > distribution of side suits:
>
> there is, a priori, a *huge* difference between, say, 5422 and 5431. Of
> all "normal" patterns, I've found 5431 to give the best combination of
> flexibility both in bidding and play.
According to Thomas Andrews' statistics as measured on Ginsberg's
deal library, 5431 has partnership-expectation of 0.28 tricks better in
offense (at our best [double-dummy par] contract), 0.12 better in
defense (against their best contract). While not "huge", this is
certainly an important difference for shapes which popular hand
evaluation methods tend to equate; it's more, for example, than the
difference between 5422 and 5332 (0.27 in offense, 0.04 in defense).
6331 versus 6322 is similar -- 0.27 in offense, 0.13 in defense.
In "How Shape Influences Strength", with a very different statistical
method (based on long simulation runs for similar hands, rather
than averages on all sorts of hands), I get lesser differences --
0.10 tricks for 7321 compared to 7222, 0.12 for 6331 compared
to 6322. These are offensive-tricks only, in the single-hand
projection (measured partnership-expectation values are thus
just 2/3 of these), and for very specific hands -- one-suiters with
a very strong long-suit, which is assumed to become trumps.
The higher difference observed on Ginsberg's deal library most likely
are due to the higher possibilities, with the more shapely 3-1
rather than 2-2 remainders, that *another* strain (not our
longest suit) will provide a superior trump suit for the partnership.
5431 should then excel in this respect.
Alex
jan kamras
Alex Martelli wrote:
>
> According to Thomas Andrews' statistics as measured on Ginsberg's
> deal library, 5431 has partnership-expectation of 0.28 tricks better in
> offense (at our best [double-dummy par] contract), 0.12 better in
> defense (against their best contract). While not "huge", this is
> certainly an important difference for shapes which popular hand
> evaluation methods tend to equate
Don't forget that 5431's also tend to be easier to handle in the bidding
than 5422's.
Alex Martelli
That can cut both ways -- 5422 can sometimes bid NT in cases
where 5431 does not give that option. E.g., some players will
happily open 2NT with 2=2=5=4, 20 HCP, and stoppers, &tc.
Anyway, a delightful paean to 5431 shapes can be found in the
January 1954 Bridge World, "The Ubiquitous 5-4-3-1", by
Bernard Trippett; he quotes S. Garton Churchill about this
shape being "rich in partnership value", and recommends
(quoting Acol practice, but with no specific citations) highly
optimistic bidding with 5431's including majors -- a specific
example he gives is AJT73-A642-8-975, which he considers
a clear 1S opener (partner has 9852-KQ8-A9532-6, and the
hands made 4S at one table while being passed out at the
other; I'm not sure which event he's quoting this hand from,
or if it's an artificially constructed example).
Would YOU open with AJTxx-Axxx-x-xxx...?-)
Alex