|
This K&R algorithm file is from
http://www.gg.caltech.edu/~jeff/knr.txt
Also see the C source code at http://www.gg.caltech.edu/~jeff/knrsource.c
[Notes in brackets tie to BidBase HandVal table ID #'s
or are other comments by the author of BidBase.
This algorithm has some peculiarities. I've noted
them in the text below. Aside from that, K&R makes
no adjustments for bidding by other players, as
some other valuation methods do. This makes K&R
more suitable for evaluating opening bids than for
subsequent bidding.]
[One of the major features of K&R is the adjustment for
honor points for suit length. A-K-Q are valued 3-2-1
initially, then the 4-3-2-1 honor points are multiplied
by the suit length and divided by 10. So Kx is adjusted
up by 3/10*2=.6 while K-sixth gets 3/10*6 = 1.8. ]
Recapping the K&R algorithm from the
Protected high cards:
A = 3, K = 2, Q = 1 [K1000-1030]
Shortness:
void = 3, singleton = 2, doubleton = 1 [K1120-1140]
[K1150] Discount first doubleton
[Although this file, the C source file and the Perl
source file -- all from Jeff Goldsmith's web site --
say "Discount first doubleton", the C and Perl code
actually discount the first distribution point, whether
a doubleton or not. So a hand with 3 distribution points,
whether a void, a doubleton and singleton, or 3 doubletons,
would be worth 3 - 1 = 2. Jeff Goldsmith confirmed this
via email.]
[K1160] 4333 = -.5
[This is rather bizarre because if the hand is NOT 4333,
then it has a distribtion point, which is reduced by 1
at K1150, above, and if it is flat, it is reduced by .5,
so EVERY hand is reduced by either 1 or .5, for which
Goldsmith had no explanation.]
Suit quality:
4321 count divided by 10 * number of cards in suit
[K1170-1230 -- instead of calculating this adjustment
by suit, BidBase does it by card. E.g.: For AQ32,
instead of 6/10*4 = 2.4 as specified above, BB does:
4/10*4 + 2/10*4 = 1.6 + .8 = 2.4, etc. ]
Adjustments to the 4321 count (before dividing by 10 and
and * by #cards):
Seven card suits: missing Q or J (+1)
[K1070: -Q, -J = 1/10*7 = +.7 ]
Eight card suits: missing Q (+2 with or w/o J) or J (+1)
[K1270: -Q = 2/10*8 = +1.6]
[K1280: *Q & -J = 1/10*8 = +.8 --
If the Q is missing, we add 2 (actually, 1.6),
regardless of the J, so add +1 only if the
J is missing when the Q is NOT missing.
If we just had "-J", then when QJ were both
missing, the program would use both 15 & 16.
Also note that you must use the asterisk
because just "Q & -J" would only be true with
Q and no higher.]
Nine or longer: missing QJ (+3) or Q (+2) or J (+1)
[K1240: -Q & -J -- both Q & J are missing]
[K1250: -Q & *J -- only Q is missing]
[K1260: *Q & -J -- only J is missing]
[Note: +3/10 * suit len - 8, etc.]
For Intermediates in six-card suits or shorter:
Ten with two honors: 1
[K1210: ??J, ???J -- Although this file says "with
two honors, the C source shows 2 *or more*.]
Ten with the Jack: 1
[K1210: *JT -- any JT combo]
Other tens: .5
[K1220: T, ?T & -JT -- Either Ten and no higher
or T with 1 higher except J, which was
counted in 21, above.]
(modifier about six-card suits misplaced?)
[No idea what the line above means.]
[K1230: ??9, *T9, *98 = .5/10* suit len]
Short honors:
stiff king = .5
[K1050: K is counted for 2 pts in #2, so we deduct 1.5]
doubleton queen if with A or K = .5
[K1070: ?Q is either AQ or KQ. ]
stiff Q or Qx = 0
[K1060: stiff Q] [8: Qx]
Lower honors:
Queen if without ace or king and suit is 3 or longer = -.25
[K1090: Qxx+ ]
J with exactly two higher honors = .5
[K1010: ??J in 4-card suit (protected honor)]
J with exactly one higher honor = .25
[1110: ?J in 4-card suit (protected honor)]
[ Odd that in AKQJ or Jxxx, the J gets nothing.]
Doubleton queen = .25
[BB counts Q as 1 in K1030,
AQ/KQ deducts .5 in K1070,
Qx deducts 1 in K1060.
Don't understand why another .25 is added here.]
Ten under two higher or with 9 and one higher = .25
[K1040: ??T or ?T9 -- Under "Suit Quality", when it says
"Ten with two higher", the C source code shows
two *or more*, so you would think that would also
be the case here, especially since right above they
explicitly refer to a J with "exactly two higher",
yet in the C source, the calculation shows Ten and
*exactly* two higher, not two or more.]
Use of K&R Points:
1H, 1S, 1NT opening optional at 12.0
mandatory 12.5
1C, 1D opening optional at 13.0
mandatory 13.5
2C with major 22.0
2C with minor 24.0
|