Do people employ more-precise (non-integer) HCP-counting?

[JLilly]

I recall reading material that judged, in terms of trick-taking ability, the relative values of high cards to be:

A 4.5

K 3.0

Q 1.5

J 0.8

T 0.2

As you can see, these numbers are not integers, though over all 52 cards in the deck, they still sum to 40. If I recall correctly (which is unlikely) the context that produced these values was double-dummy NT hands.

 

Every good player should of course assess their strength in the context of the playing environment, partner's bidding, opps' bidding, etc. But I'm curious about people explicitly using non-integral HCP-evaluation methods.

[JLilly]

Me, I don't think explicitly in non-integer terms, but I make soft adjustment that in practice are non-integral. For example, playing 15-17 strong 1NT openers, I'd deem the following good enough for 1NT:

♠QJTx

♥KQx

♦AJxx

♣JT

The thinking would be "the (Q)JT sequences are nice; and I'm not aceless. Add that to {3,4} in the majors -- eminently a NT hand".

[ggwhiz]

Me, I don't think explicitly in non-integer terms, but I make soft adjustment that in practice are non-integral. For example, playing 15-17 strong 1NT openers, I'd deem the following good enough for 1NT:

♠QJTx

♥KQx

♦AJxx

♣JT

The thinking would be "the (Q)JT sequences are nice; and I'm not aceless. Add that to {3,4} in the majors -- eminently a NT hand".

 

As a rookie I once opened 1nt on

 

♠Qxx

♥Qxx

♦Qxxx

♣AKQ

 

and soon turned white as a sheet.

 

My "soft" adjustments these days relate specifically to shape, source of tricks and body cards(T98 type) as in 5-4-3-1 shape is my fave for the number of possible landing spots.

[spotlight7]

Reith(?) Count A=4.3 K=3.1 Q=1.7 J=.9 is useful for close decisions.

 

 

If you want to avoid adding up 'numbers' Ace balances a Queen and King balanced a Jack.

[TylerE]

Integer values... but adjustments aren't to the count directly, just basically "do you like your hand".

[silvr bull]

I do a first quick 4-3-2-1 count to triage the hand into groups (bad, OK, good, great, etc). Then I do soft adjustments to several factors (bidding, fit, suit lengths, supported tens and nines, etc.) more as plus or minus than numerical count adjustments.

[johnu]

I recall reading material that judged, in terms of trick-taking ability, the relative values of high cards to be:

A 4.5

K 3.0

Q 1.5

J 0.8

T 0.2

 

As you can see, these numbers are not integers, though over all 52 cards in the deck, they still sum to 40. If I recall correctly (which is unlikely) the context that produced these values was double-dummy NT hands.

 

Probably BUM-RAP points (I am the U in BUM although some just call me a bum) rounded to 1 decimal place.

 

A 4.5

K 3.0

Q 1.5

J 0.75

T 0.25

 

These were just normalizations of a very old 3-2-1-1/2 point count system created by multiplying by 1.5 and adding 1/4 point for tens so that the total number of points added up to 40. The goal was to allow players to use their normal Work (4-3-2-1) point count definitions that they previously used without having to make changes to the point ranges.

 

Of course, suit quality is not part of the point count.

 

A1098 is generally much more valuable than A432 with the 10 in a small doubleton.

[1eyedjack]

I think it is a bit of a blind alley. I reckon that a typical bridge player's development involves first, learn a simple evaluation system. Then another, and another. Then go down a spiral of "tweaking" your favoured method after having settled on one, in a seemingly never ending evolution in pursuit of some mathematical algorithm that will magically give you the right answer. But this all takes a long time during which you are gaining a lot of experience stored subconsciously, until eventually you say hang it and go with your gut. Which by then is more accurate than any formula and frees up your mental resources for other profitable analysis. That point is almost a quantum flip and the feeling of release when it comes is palpable.

[JLilly]

I think it is a bit of a blind alley. I reckon that a typical bridge player's development involves first, learn a simple evaluation system. Then another, and another. Then go down a spiral of "tweaking" your favoured method after having settled on one, in a seemingly never ending evolution in pursuit of some mathematical algorithm that will magically give you the right answer. But this all takes a long time during which you are gaining a lot of experience stored subconsciously, until eventually you say hang it and go with your gut. Which by then is more accurate than any formula and frees up your mental resources for other profitable analysis. That point is almost a quantum flip and the feeling of release when it comes is palpable.

 

1eyedjack, to what extent do you think that point depends on one's partners?

[1eyedjack]

I see your point. Quite a lot, I suppose, if the context is right. If you are making a limit bid over which partner is invited to act, it certainly helps if he knows how you arrived at your limit.

[Zelandakh]

I think people need to be a little careful with these value calculations. The value of an honour varies. The king of a long suit is much more valuable than as a singleton; the queen in Qxxx is less valuable than in AQxx. The same for johnu's example of AT98 - the ten is much less valuable without a supporting ace.

 

One reasonable approach is to take your normal Milton Work count and then apply pluses and minuses that are worth approximately a half a point. Many of these will offset, thus saving yourself the calculation. Where they do not, you probably have an upgrade (or downgrade) hand and will know what to do.

 

Getting into 1.7 territory is imho not the right approach. An approach based on honour and suit combinations, such as KNR, may be appropriate for bridge computers but is impractical for most players. In the same way that a chess GM does not look at a pawn structure in chess and think of it as being worth 0.17 of a pawn but rather just as giving a small advantage. For a chess computer it makes sense though.

 

people here will often talk about a good 15, or of a hand being worth 15+. This is just a shorthand for this process. You do not really need to know whether the "best" evaluation is 15.17 or 15.24 because these are treated equally in the bidding system. If there was more bidding space such that we could define hands more accurately then it might be worthwhile in some cases. As things stand it is pointless.

[Tramticket]

Don't forget that your hand evaluation should not be static, but should be continually updated as you get more information.

 

For example, you may judge a holding of KJXX to be worth 4 points when you first pick the hand up, but when LHO bids the suit it will surely be worth considerably less. If partner bids the suit it may be worth more than 4.

 

I also find it useful to try to keep in mind the defensive strength of a holding as well as the playing strength. For example you pick up a hand containing a singleton king - you initially down-grade this holding. Partner opens the bidding and RHO bids the suit containing your singleton. The auction develops into a competitive auction. I would now judge that the king may well take a trick in defence (when declarer finesses in trumps), but may be worthless if partner becomes declarer (the opponent holding the ace will play it on the first round of the suit and partner will be ruffing the second round in dummy).

[jogs]

Don't forget that your hand evaluation should not be static, but should be continually updated as you get more information.

 

 

That's why dynamic evaluation during the auction should be expected tricks by the partnership.

[yunling]

As Zel has pointed out, at the decimal level, card combination and suit length is quit important so working on values of individual high cards is far from enough.

I once created an evaluation table via some statistics, accurate to 0.25 points, and I tried it myself; but it is so complicated and rarely uesd that I would never recommend it to my partner. You must have a headache if you are told to remember something like

Kx=2.75 AK=6.75 KJTxx=4.25 AQTx=6.75 etc.

[jogs]

 

Kx=2.75 AK=6.75 KJTxx=4.25 AQTx=6.75 etc.

It's only necessary to know if standard pointcount has over or under estimated the value of the hand.

 

So instead of having a 14 HCP hand, the hand is 14+ or 14-.

[yunling]

So instead of having a 14 HCP hand, the hand is 14+ or 14-.

 

In rare occasions adjustments can be much greater.

e.g.

KQ

QJxx

KQxx

Jxx

is a 12- hand, not 14-

[jogs]

I remember holding an ugly 12 point hand in 2nd seat. No ace and only one king. I passed.

 

The auction went

 

(p) - p - (1NT) - p

(2NT) - all pass

 

We managed to win 11 tricks on defense. +300. Tied for bottom.

The other +300 was 3♠X-2. Most of the field were at 3NT+1 for 430.

 

LHO had 7 spades. Partner had 4 bare aces.

[SteveMoe]

Milton Work points are useful for NT and opening bids. Even there, adjustments for length and strength-within-length apply. After the first bid in a suit contract, the thinking goes to shape (working strength) and tricks pretty quickly. Rubens "In and Out" valuation applies. Generally use A=4.5 K=3, Q= 1.5, J=0.75, 10=0.5 when hand is skewed to AKs or QJs.