BBO Discussion Forums: Some questions about hand evaluation - BBO Discussion Forums

Jump to content

  • 2 Pages +
  • 1
  • 2

Some questions about hand evaluation

#1 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-22, 22:59

Dear all

There has been an interesting thread on the intermediate forum and use of Losing Trick Count (LTC), Winning Trick Count (WTC) and other methods for determining game and other contracts. There has also been occasional discussion on here regarding the Law of Total Tricks and/or Trumps (LOTT). We also have several simple point evaluation methods at our disposal which often require adjustment based on the auction - in fact all methods require adjustment - often assessed after the event :( I know obviously all methods are only approximate and nothing beats experience

People also regularly discuss the Kaplan-Rubens hand evaluation which we know is widely accepted as an excellent hand evaluation (mathematically based) but too complex for anyone to really use at the table. Recently I've also been reading several good sites on hand evaluation, especially using Richard Pavlicek's site (he is very statistically focussed in approach) who has his own methods of assessing hands. Those who play GIB on here of course know the basic approach used by GIB on HCPs + adjustments for total points. Ive also been referred to a great deal of literature (such as Marty Bergen) and have read much of Ron Klinger's advice and others over the years. I tend to rely on web sites.

Over recent months I too (along with others on other forums) have been becoming quite rigorous in trying to assess hands and contracts using a mixture of LTC, LOTT, HCPs, total points and Pavlicek Points (with some success I think in improving my bidding). I also extensively analyse hands afterwards using things like KR to try and understand (in my head) what it is that gives hands a good KR vs poor KR. Obviously we develop a bit of a feel over the years based on how many hands we play (without counts) and start to see what a good suit, hand, distribution looks like and what a poor one looks like.

But at the improving novice/beginner -> intermediate level I feel there is a bit of a gap. So I'm hoping, maybe with the assistance of a hand example that we could examine one in great detail and get the different perspectives. I know its hard with one example but the following hand seemed to me to have some interesting features as a competitive auction with a good balance of strength and a competition between part scores and game.



Note. Many of the numbers below are courtesy of Richard Pavlicek's hand calculator

North for example only has 4 hcps(5 total points), 10 losers, fairly ordinary distribution (except doubleton and ace) but with 4 trumps
South on the other hand has 16 hcps (17 total pointsK) (2 excellent suits), only 5 losers, 6 trumps, nice distribution despite possible wasted K singleton

That gives a total of only 20 HCPs (22 total), 15 losers, 10 trumps which does make 4 hearts

East has 10 HCPs (11 total points), 8 losers, 3 trumps
West has 10 HCPs (11 total points), 8 losers, 5 trumps

That gives a total of only 20 HCPs, 16 losers, 8 trumps which does make 3 spades

So there are 18 total trumps, 19 total tricks, points fairly evenly distributed - not a bad outcome for the Law - and worth competing
EW has one more loser than NS however on losers the contracts should theoretically be 3 hearts and 2 spades - not too far off but enough to cause problems missing a game bid. This suggests an adjustment based on number of tricks - maybe someone can advise on adjustments

I also looked at KR which upgrades S to 18 points based on the two excellent hearts and diamond suits
KR downgrades N from 5 total to 4.4
KR downgrades W by slmost 1 point to 9.15. Pavlicek downgrades distribution total from 11 to 10 due to unprotected K
E is downgradede by KR to 9.65 mainly from S qxx. Pavlicek gives E 11 total points

So Pavlicek total gives NS = 22 EW = 21
and KR gives NS = 22.4 and EW = 18.8

So all point count methods give NS roughly 22 points and EW somewhere between 19 and 21

Lets look at the auction

I open 1 heart with my excellent hand - North will assume approximately 7 losers in my hand (initially) and add to its 10 losers -> 17 losers which in theory is only a 1 level bid. However with heart support its definitely worth a 2 level bid. Alteranatively it has used the law and decided we can possibly go to 3H

At the 2 level from partner I would usually assume 9 losers plus my 5 losers gives 14 losers but i know we have 9+ trumps (3 level) so would either go straight to game (or bid 3) but I am forced up by the cue bid. That would seem to work quite well with losers, law of total tricks and also we are a bit light on points

What I'm not so sure about is how EW can use these methods with the overcall and the cue bid. They know they have 8 trumps and 19+ total points from the bidding. W had 8 losers, opening at the 1 level and E is assuming around 15-16 total losers going to 3???

Does anyone have any suggestions on all this and also the winning trick count (i dont use that yet) and whether there are other considerations in bidding that hand. Sorry for the long post

regards P
0

#2 User is offline   dsLawsd 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 299
  • Joined: 2017-September-15

Posted 2019-January-23, 03:47

I don't use LTC- here many would not overcall 1 with the stiff heart K. I have good former partners who would then pass N.
After this auction I would DOUBLE 3 showing a good hand since my normal game try of 3 is not available. Over 3 someone might hazard a double and avoid bidding game.
To me it is about judgment and not any point count evaluation, and the Law application can vary significantly depending upon the purity of "our suits" versus theirs.

A very good question and sample hand.....
0

#3 User is offline   The_Badger 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,125
  • Joined: 2013-January-25
  • Gender:Male
  • Location:England
  • Interests:Bridge, Chess, Film, Literature, Herbal Medicine, Nutrition

Posted 2019-January-23, 05:12

4 shouldn't make on a lead. There's no reason - I can think of - of playing West for a singleton K, except he might (and equally might not) have it for the vulnerable overcall.

On the K and R evaluation, North's hand seen in isolation comes out at 4.4, but what it doesn't do is take into consideration is the fit, the doubleton and the A as "a new evaluation based on partner's bidding."

The known 9 card fit adds a point, and the doubleton , too, so North's hand is now a respectable 6 count as a basic evaluation. Add that to the 18 point K and R evaluation of South's hand, and the total is 24.

3 as a contract on 24 "total points" sounds about right. Ok, so sometimes you miss a game, but as we have all discovered, any hand evaluation method has its plus points and minus points.
1

#4 User is offline   nudnikbp 

  • PipPipPip
  • Group: Full Members
  • Posts: 84
  • Joined: 2019-January-09

Posted 2019-January-23, 05:52

North's two heart bid seems automatic because he has four trumps, and ace, and a doubleton.
After the raise South has an easy four heart bid with his two suiter. North doesn't need much for four hearts to have a reasonable play, e.g., diamond king or diamond jack-ten along with the heart king or one other decent card.
0

#5 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 06:10

Thx for the responses. Would 4H not be a reasonable competitive sacrifice against 3S on this hand. That is why I went to 4H immediately rather than letting bidding go round one more round. I also thought there was a reasonable chance. As you say and I evaluated it was clearly a 3S vs 3H hand on most of the numbers but NS has a marginal advantage and 4H-1 would score ok?? However my estimate of total losers was 9+5 giving a chance of game, despite points and 9+ trumps in competition.

But you are correct, it was a lucky lead. I would have finessed the K and gone down 1. Pays not to believe all analysis afterwards in what is possible. 4H is only possible with the Trump or diamond lead and wrongly playing for a drop. :)

Regards P
0

#6 User is offline   nige1 

  • 5-level belongs to me
  • PipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 9,128
  • Joined: 2004-August-30
  • Gender:Male
  • Location:Glasgow Scotland
  • Interests:Poems Computers

Posted 2019-January-23, 06:30


Evaluation with my Dad's variant of LTC
(WTC: Winning Trick Count, where A=1.5, K=1, Q=0.5, void=3, singleton=2, doubleton=1, Decrement for duplication. Increment for trump-control).

North 2.5 winners (A=1.5, doubleton=1).
South 6 winners (doubleton=1 A=1.5 Q=0.5 A=1.5 Q=0.5 singleton=2 K not counted),.
Trump control=1 winner.
Total=9.5 winners.


The LTC (like the WTC) is a rule-of-thumb to quickly estimate trick-taking potential. Many refinements are possible (honour-concentration, honours in long suits, intermediates, and so on). Nevertheless, even in its crude form, it's a useful practical aid to judgement in the auction and post-mortem. It helps ordinary players. Many experts recommend it.

.
1

#7 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 06:35

Thx Nige. I must try using it.

It's definitely a borderline 3-4 on all the numbers

P
0

#8 User is offline   wuudturner 

  • PipPipPip
  • Group: Full Members
  • Posts: 80
  • Joined: 2011-November-07

Posted 2019-January-23, 14:04

Everybody is saying how hard it is to make 4♡ without the lead of the heart king. Can't be done they say. But, what if...

Assume West leads a spade spot. That pretty much denies a solid spade holding. So East has at least one spade honor. You know spades are 5-3. West overcalled 1♠. East raised. So I would bet a goodly sum on 5-3. (I've not seen the bots willing to overcall on a 4 card suit.)

Say you win the spade ace. Take an immediate diamond hook. This hand is an adventure of discovery, and drawing trumps is something that will need to wait.

East cue bid raised spades. So who has the diamond king? West had it, since your hook loses. Say they continue the attack on spades. East shows up with Q96 as you ruff the third round of spades. Now, exit with the club. East wins the ace, then tries the queen, ruffed by you.

This leaves you with a lot of information. East has shown up with Q96 in spades. AQ in clubs, and probably the jack there too. East originally passed in second seat though. The bidding has East with 11-12 total points, but 11- HCP.

West has KJ543 in spades, plus the diamond king. But a 1♠ overcall is listed as 8-17 HCP.

So you have two sources of information that tell you who has the heart king. You don't have it. Not dummy. Not East. And West probably needs it for the overcall.

The Rabbi lives again as the crowd roars! Cash the heart ace, with a glimmer of hope in your eyes. The king drops. Draw the last trump. Ruff the last diamond in dummy. You lose only a spade, a diamond, and a club.

The road to bridge hell is paved with the souls of those who drew trump too soon, OR too late.
0

#9 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 15:44

Thx Wuudturner :)
0

#10 User is offline   ncohen 

  • PipPip
  • Group: Members
  • Posts: 38
  • Joined: 2015-December-09

Posted 2019-January-23, 17:00

a cautionary note about LTC and LOTT.
i. LTC is mathematically equivalent to a count that assigns points for high cards and for shortness. Unlike the standard count, though, it assigns 3 each for A, K, and Q (each is a non-loser), and 3-6 and 9 for doubleton singleton void. So,
a. it counts A, K and Q the same. An A is worth more than twice a Q.
b. It is far more optimistic than the standard count about the value of distribution, that either each card over 3 in a suit is a winner OR that each card less than 3 can be used as a ruff. The first assumption fails if, for instance, dummy lacks the entries to establish a long suit OR declarer is short in dummy's long suit so not all of the extra cards are winners. For example, xxx xx xxxx xxxx with spades trumps. This hand is 11 losers, so it is assumed it contributes a winner. That may not be possible if the defense can draw trumps or declarer also has a doubleton heart or declarer has QJ10 of hearts and doesn't need a ruff. The optimism is more justified when there's a good trump fit.
I. There can be a lot of double counting. consider: AKQJ10987 65432 -- --
654 1098 KQJ KQJ10 There's lots of double counting in the minors. N has a 2 loser hand and S has an 8 loser hand. Under LTC, 14 tricks should be available.

ii. LOTT can be off by a zillion tricks.
a, It doesn't take into account long side suits:
AKQJxxx AKQjxx --- ----
---- x xxxxxx xxxxxx NS can make 13 tricks in spades unless the heart lead is ruffed. If EW have mirror hands, they can also
make 13 in Clubs or diamonds. That's 26 tricks with 14 total trumps.

b. It doesn't take into account one defender holding winning trump tricks after his partner is out:
AKxxxx AKx x xxx
xx xxxx QJ109 QJ10 Again, if mirror distribution and there's a heart loser (and club loser for EW playing in diamonds), then there are 16 total trumps and 12 total tricks.

So, if you have a good side suit, LOTT is too low. If you have strong cards in the opponent's suit, LOTT is too high.
0

#11 User is offline   nige1 

  • 5-level belongs to me
  • PipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 9,128
  • Joined: 2004-August-30
  • Gender:Male
  • Location:Glasgow Scotland
  • Interests:Poems Computers

Posted 2019-January-23, 17:33

You can construct deals to frustrate almost any mechanical method of estimating trick potential.

Duplication.
No declarer, in any contract, can take more than 5 tricks against best defence

Thomas Andrew's remarkable construction where
Each player can declare and make 3NT against any defence

South can make 7N but
so can West
It just depends on who bids it first.

South can make 6 on any lead

0

#12 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 17:34

View Postncohen, on 2019-January-23, 17:00, said:

a cautionary note about LTC and LOTT.


Thanks. I have just started adding them as an extra check on my normal bidding. Without any of these different methods, just using my old point counts, limit bids and trump counts I would have expected the auction to go (without interference) 1H - 2H - 4H or even in my old Acol days 1H-2H-? (I cant remember exactly)

I agree that all these methods can be seriously out with some distributions. Sometimes we rely too much on these numbers and not enough on understanding shape etc. I have found LTC quite useful recently. Not always such a big fan of LOTT except for competitive auctions

Your caution is well noted and I think all beginners should be wary of relying on any gadget as a strict rule. Better to learn the principles surrounding a rule rather than relying purely on numbers. They are a good guide though

regards P
0

#13 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 17:55

As a followup. I havent investigated Goren point counts yet. Its next on my list. What do people think of Goren points

regards P

PS Someone once mentioned Zar to me but it looked rather complex for easy application (not as hard as KR). One thing I find about KR as The Badger alluded to is that KR doesnt really look at adjustments for trumps and really just looks at each suit in isolation. That seems a bit strange that it doesnt have a distributional adjustment on top of the individual suits

My understanding of Goren is that my two AQ would add an additional 3 points to give me 19?? And after the raise of my opening my singleton is upgraded in value too? So my total Goren point count is getting close to 20. Im not a Goren user so am not confident of those numbers
0

#14 User is offline   nullve 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,228
  • Joined: 2014-April-08
  • Gender:Male
  • Location:Norway
  • Interests:partscores

Posted 2019-January-23, 18:05

View Postncohen, on 2019-January-23, 17:00, said:

i. LTC is mathematically equivalent to a count that assigns points for high cards and for shortness. Unlike the standard count, though, it assigns 3 each for A, K, and Q (each is a non-loser), and 3-6 and 9 for doubleton singleton void. So,
a. it counts A, K and Q the same. An A is worth more than twice a Q.
b. It is far more optimistic than the standard count about the value of distribution, that either each card over 3 in a suit is a winner OR that each card less than 3 can be used as a ruff. The first assumption fails if, for instance, dummy lacks the entries to establish a long suit OR declarer is short in dummy's long suit so not all of the extra cards are winners. For example, xxx xx xxxx xxxx with spades trumps. This hand is 11 losers, so it is assumed it contributes a winner. That may not be possible if the defense can draw trumps or declarer also has a doubleton heart or declarer has QJ10 of hearts and doesn't need a ruff. The optimism is more justified when there's a good trump fit.
I. There can be a lot of double counting. consider: AKQJ10987 65432 -- --
654 1098 KQJ KQJ10 There's lots of double counting in the minors. N has a 2 loser hand and S has an 8 loser hand. Under LTC, 14 tricks should be available.

I made essentially the same point about LTC recently:

View Postnullve, on 2018-September-23, 16:19, said:

Isn't the (simple) LTC method equivalent to counting distributional points using the (Goren) 3-2-1 scale and then adding 1 point (the value of a doubleton) for each "cover card", i.e. each ace, non-singleton king or non-singleton, non-doubleton queen? :unsure:

Zelandakh has often made the point that the (M)LTC method is a terrible point count method in disguise. For example, here.

View Postncohen, on 2019-January-23, 17:00, said:

ii. LOTT can be off by a zillion tricks.
a, It doesn't take into account long side suits:
AKQJxxx AKQjxx --- ----
---- x xxxxxx xxxxxx NS can make 13 tricks in spades unless the heart lead is ruffed. If EW have mirror hands, they can also
make 13 in Clubs or diamonds. That's 26 tricks with 14 total trumps.

b. It doesn't take into account one defender holding winning trump tricks after his partner is out:
AKxxxx AKx x xxx
xx xxxx QJ109 QJ10 Again, if mirror distribution and there's a heart loser (and club loser for EW playing in diamonds), then there are 16 total trumps and 12 total tricks.

So, if you have a good side suit, LOTT is too low. If you have strong cards in the opponent's suit, LOTT is too high.

Yes, LoTT can be off by many tricks, but is extremely unlikely to be off by more than 2 tricks even if no adjustments are made. And if adjustments of the kind Cohen himself mentions (e.g. in To Bid or Not to Bid) are made, then the total trick prediction will seldom be off by more than 1 when the full distribution around the table is known. I actually believe I'm then able to predict the exact number of total tricks more often than not, but I may be wrong. :)
0

#15 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 18:39

I disagree with the statement that LTC values AKQ the same. It takes account of combinations of honours and shortness too. So its more complex than simply valuing the same. On average a Q would be worth less than a King in LTC. It also upgrades supported honours and downgrades isolated honours. I would be interested in a full analysis of it but from limited experience so far am sceptical about writing it off yet :)
0

#16 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-23, 19:46

Following the above comments about LTC I've reevaaluated the above hand using the new losing trick count and also considered adjustments. Note that the following is (as far as I know) an ideal hand for use of LTC. Both hands have distributional hands with 8+ trump fits. It is not supposed to be used in other situations

However the NLTC which was proposed in 2003 by Koelman gives north the same number of losers and my hand an extra loser - half extra for king singleton, and half extra for doubleton
Nevertheless NLTC subtracts from 25(not 24) so results in the exact same estimate of bidding to 3H on the cards as shown. However given the 2-level bid I usually estimate that (unseen) as 9 losers giving me a possible game

If we make further adjustments as proposed by Klinger and others I believe my AQx and AQJ should be reduced slightly in number of losers giving me even closer to game

I know all methods are flawed but if you combine number of points, number of tricks, competition against 3S, losing trick count, winning tricks if you use that, Goren points if you use that, etc. it looks like a straightforward game bid however you look at it

P
0

#17 User is offline   thepossum 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,362
  • Joined: 2018-July-04
  • Gender:Male
  • Location:Australia

Posted 2019-January-24, 17:15

Following the criticisms of the LTC and NLTC I have just investigated the MLTC

Here are the comparisons for the above hand

LTC MLTC
N 10 10
S 5 6
EstTricks 9 9

W 8 8.5
E 8 9
EstTricks 8 7.5

So all losing trick count methods (assuming I can add up) give NS the same number of tricks but MLTC reduces EW by half a trick. This is just one hand so who knows how they work on average. I think sometimes we can obsess over modifications and precision when simple often works just as well :)

But the way I learned to use it, not assessing number of losers after the event, is to estimate losers from partners bid once a fit is found. North make the first estimate of mine as 7 losers + 10 -> 17 losers which doesnt suggest a 2-level bid. However they have enough trump support to go to 2-level. I then estimate north's 2 level bid as 9 losers + 5 losers -> 10 estimated losers and possible game. So the precision in the method appears to come from a combination of simple counts plus adjustments based on the level of the auction. I'm obviously not an expert and just starting out using it over the last month but it does seem to work quite well in conjunction with other methods in deciding what level I should consider bidding to

P
0

#18 User is offline   ali345 

  • Pip
  • Group: Members
  • Posts: 2
  • Joined: 2019-February-17

Posted 2019-February-24, 04:51

Thanks. Interesting to read and learn
0

#19 User is offline   ali345 

  • Pip
  • Group: Members
  • Posts: 2
  • Joined: 2019-February-17

Posted 2019-February-24, 04:51

Thanks. Interesting to read and learn
0

#20 User is offline   HardVector 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 479
  • Joined: 2018-May-28

Posted 2019-February-24, 17:07

If you are looking for a way to make good hand evaluations that don't involve complex mathematical exercises, try "Slam Bidding Made Easier", by Marty Bergen.
0

  • 2 Pages +
  • 1
  • 2


Fast Reply

  

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users