Join Bridge Winners
All comments by Srdjan Katusic
You are ignoring the author of this comment. Click to temporarily show the comment.
Thats promising line, but there is a catch: for red squeeze to operate against RHO, you have to cash king of clubs at some point, and still keep ace of diamonds as communication with dummy. Defense can arrange that this is not possible if RHO wins first trump trick and returns a heart. Actually thats what happened to me at the table :(
Oct. 17, 2017
You are ignoring the author of this comment. Click to temporarily show the comment.
I wish I have supported partner on previous round, why didn't I?
June 30, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx to Han Peters, who took trouble to develop a script, and to Thomas Andrews who developer Deal program, I have results of simulation that throws some further light on this bidding problem.

Parameters of simulation are:
-south hand is fixed to AJT762 32 J942 5
-north hand will have between 11 and 13 HCP
-it will contain exactly 6 hearts, 0-3 spades, and 0-4 diamonds and clubs

All hands satisfying this criteria will be processed by double dummy solver, and difference between score that would be achieved in 2h by north and 2s by south will be compared via IMP scale, NS being vulnerable. Such calculation is done for 1000 deals satisfying this criteria.

Results:

Winning action is to pass 2h. On average you earn 0.34 IMps per board if you always pass in such situation, compared to bidding 2s.

If you play system in which opener will prefer to support spades with 3, rather than rebid hearts with 6, it's even more clear to pass, expected average earning is 1.05 IMPs per deal.

However if our spade holding is altered from AJT762 to KQT762, than winning action is to bid 2s, you earn about 0.7 IMPs on average if correcting to spades rather than passing. Quite good to be aware of different evaluation of these two holding in this and similar situations.

Of course, simulation and double dummy analysis is not exactly covering all stochastic elements on table, including if bidding will die at 2h or 2s, but still give us good feeling about winning action, I believe.
May 23, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Probably most important criteria to judge which partial is better is how likely it is that partner heart suit is equal or better in quality to our AJTxxx spades. AS he has most likely exactly 6 hearts out of 11 missing, and he will have around 1/3 or outstanding 34 HCP's we can assume that he will have on average 1.5 top honor and either jack or 10, so average holding might be KQTxxx or AJTxxx, i.e. something similar to our spades. AS we have 2 hearts, and partner will on average have 1.5 spades, and we have possible ruffing value for hearts, playing in hearts looks better. But i will try to verify that via simulation and post results soon.
May 21, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx, I just did.
May 21, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx, Han, I'm trying to do this myself , but perhaps you can help me bit more to run simu myself:

I have developed simple script that create deals, something like:
south is “AJT762 32 J942 5”

main {
if {>=11 && <=13 && >=0 && <=2 && >=6 && <=7 && >=0 && <=4 && >=0 && <=4 } {accept}
reject
}

I'm now trying for some time to run it via dd solver, command should be something like:
deal::tricks south spades
deal::tricks north hearts

…but i'm failing to integrate this into script and get some output, i tried switch format/ddeline, but that's cryptic, can you perhaps help, i can't really find in docu how to do it..
May 20, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
THx, I checked deal by T. Andrews. It' seems easy enough to create set of hands under given criteria, but for me issue is how to easily push them thru some double dummy solver. Do you really have some ‘integrated workflow’ to create deals, push them to d.d. analyzer and analyze output of it?
May 20, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx all for discussing this case. I have used Richard Pavlicek tool http://www.rpbridge.net/cgi-bin/xch1.pl to calculate expectancy of partner length in spades:

void 9%
1 33%
2 39%
3 19%
average number of spades: 1.69

If we assume that partner will prefer to raise spades if he has three:

void 11%
1 41%
2 48%
average number of spades: 1.38

Perhaps this helps to judge if passing or rebidding spades is winning action on the long run.
May 19, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx for simulation! Can you share some numerical results, like expected number of tricks in heart or spade contract?

And, can you provide info about tool-set you use or recommend for such simulation?
May 19, 2016
You are ignoring the author of this comment. Click to temporarily show the comment.
Interesting analogy to gambling. I think its not perfect analogy, because we as bridge players are not like guys that are playing occasionally lottery and accept to loose small money in vain hope that we will earn big one. Instead, we are making so many bets, even in one single hand, and for each one we hope that it has positive earned value, to best of our knowledge, experience or, if there is nothing better available, on pure feeling. One session of bridge consist of probably more than 100 bets, many small and some bigger. If we are good, we are choosing what to bet on with one criteria: whats earning us best result on long run. Only in some special cases, like towards end of knockout matches, maybe it pays off to alter this strategy of ‘long-term earning’ to ‘insurance-policy’ strategy or ‘swing-generation’ strategy.
Dec. 16, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Maybe its easier to see it looking to combinations of split in key suit. In this example, safe-line shields us from one split, for 14 IMPs gain. As there are only 15 other splits , max-line would have to yield over-trick (and safe-line not) in 14 out this 15, just to break even.
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx Doug for comment. You have put good argument why short knockout matches with IMP scoring are doubtful idea. When team matches are scored on some VP scale instead of being knockout, state of match becomes less of concern in borderline decision if to go for safety play.
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
This does not look like safety play example (your contract is secure), but same type of calculation is possible. We calculate earned value of line (60% 12 14% 10) versus line (100% 11). Results is 0.6*1IMP + 0.4*(-1 IMP) = +0.2 IMP, more than zero, so line is justified, in this case, as correct max-line.
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Michal,

I did just first part of calculation with assumption that another table is also playing small slam. Next step of calculation takes into account possibility that another table plays in game or in grand, according to frequency sheet obtained from 20 tables playing same board. Calculated value proves your point, earned value for safety play can only increase by the fact that its not sure that another table plays optimal contract we are playing.

And of course you are right spotting inconsistency in my comment. Safe-line that fails when max-line succeeds is highly suspicious:) But there are some borderline cases (not for this board), for instance when you try to guard against 4-1 split in one suit, and you fell victim to ,say, 6-1 split in another suit, when key suit was in fact 3-2, so max-line succeeds…
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx Alan for adding additional complexity to safety issue :)

Yes, sometimes we should not look to decision if to play safe in vacuum. In case of this particular deal, lets just imagine that calculation has evaluated negative earned value for safety play. Still, if we feel that we are winning in knockout-type of match, we should go for it as insurance policy. Good example for opposite case (positive earned value for safety play, but stil we should not use it) is not so easy to find, perhaps short or very close knockout match?

Btw, did you know that all insurance policies are having negative earned value for one who is taking it? Otherwise, who would sell it :)
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Ideally, similar kind of calculation would be called for in each case where we consider if to go for safety line. If it really looks that overall earned value of safety play is negative, it's probably best to forget it. Except perhaps in special cases, discussed by Alan Frank in his comment.
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx Harry! I have googled for meanining of phrase ‘layman’s terms', find it out, and this one instantly becomes my favourite :)
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Good point. If you drop overtrick by technical mistake in normal contract, its quite simple to calculate that your line is earning -1 IMP :) Missed undertricks are probably more expensive. Goes to show that there is not so much space for beeing lazy also in team play :)
Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx Kit for comment.
Yes, this psychological factor adds at least 0.25 IMP of advantage for playing safe, so we can round it to full IMP :)

Devil is that partner and teammate will correctly guess why you didn't play safe. Because you didn't spot safe line, and not because you have evaluated it wrongly.

For your morale, there is probably only one worser thing from missing safe line, and that is playing safe and going down in case when max-line makes :)


Dec. 13, 2015
You are ignoring the author of this comment. Click to temporarily show the comment.
Thx for a comment. I like your method for ad hoc calculation at the table! It's fast and it works splendid if for comparing safe and max line, its all about a split in key suit. If its also about position of key cards within a split, then considering and weighting winning and loosing combinations of key suit is more precise, but of course bit harder, for ad hoc calculation.
Dec. 13, 2015
.

Bottom Home Top