reflections on some jcreech hands of yesterday.

[kenberg]

Repeating a mantra (repetition is what mantras are for) I think lesson hands are most useful if we look back at them later. I have a couple of examples from yesterday from my volunteer effort at Jim's session.

Red against white, the deal is on my right and I hold:

: KQ4
: J97
: Q
: JT8532


The auction begins:
Pass   Pass   Pass   1NT
Pass     ?

So: or NT and how high?  These being lesson/discussion hands it is reasonable to say I get to ask pard whether 2NT is a trf or an invit. Let's assume invit, so I can, and did, bid 2NT. There was some discussion of this so I will give my thoughts.


It's possible that we have 9 tricks in NT. In fact we do, but pard has a 15 count so a lot of luck is involved. It's also possible that 3 is a lousy contract. It is, and can be beaten one trick. Of course it is also possible that 3 is a great contract and 2NT is the lousy one. But as I see it, there is a reasonable chance that we can make game, so I kept it simple and issued an invit.

Of course comments are welcome, that's what I have in mind  with my mantra.

I gave this to the bots and I must say I do not understand their auction at all. It started the same, then after 1NT my hand transferred and bid 3 to show the stiff, this being of course forcing to 3NT. I regard this as very optimistic. As mentioned, 3NT can be made but the bot, perhaps reasonably, did not make it. 2NT is a much more reasonable contract and I have some thoughts about that but for the moment I have to run.

More later, but here is the bot action/play.

http://tinyurl.com/y5otr9mt


I'm back. Suppose the contract is 2NT on the auction 1NT-2NT(invit)-Pass and let's suppose, as happened at the table, the opening lead is a small to the !, a small back to the Q, and a third . The reason that there are 9 tricks is that the are 3-3 and the T falls. Great, but unlikely, and when we are in 2NT we only need that one of these two things happen.  So the chances for 8 tricks, needing on of two possible lucky breaks, are a lot better than  your chances when you need both lucky breaks. And there are other chances as well. Lead a . If N rises and cashes his (he didn't but might have and then gets out with a , win on the board and duck a . The count is now rectified and maybe a black suit squeeze is available if the other chances fail. As long as we only want 8 tricks, might as well try. Anyway, the do split and there are 8 tricks. Playing in clubs, we assume the split 3-2 in which case it is likely that 2 !c tricks must be lost. Assume the same lead of a small . The opponents should get 2 , 2 , and the A. That's 5 tricks for down 1.

Summary: I bid 2NT invit, partner naturally passed, there is still a lucky play for 9 ricks but a reasonable play for 8 tricks while a contract of 3 is likely to fail.But really the reason I bid 2NT was that if partner has a 17 count there is likely to be a good play for 3NT.

Back to the bot auction for a moment: I don't understand using 2NT as a trf to but that's not my main point. To my mind, the hand is not strong enough to transfer to and then bid 3 . It shows the shape, but I view it as an overbid. Also I think it is very unlikely, just looking at the E hand, that the hand could belong in 5 . Possible but unlikely. Ad surely I would not be thinking of a !c hand. So if I wanted to play in game, I would just raise 1NT to 3NT. It works here, but it's very lucky. I think invit is the way to go.

Now to a different hand .

 The bots passed it out:

http://tinyurl.com/y2ztdf5v


At our table the auction went:


Pass   Pass   1     1
Pass   2    Pass    2
All Pass 

I led the 8 and was asked why I led the 8 rather than a fourth best deuce. Of course pard and I had not discussed carding but there are lots of agreements, lots and lots and then some. I like 3/5 against suits, so then I would lead the 6, but I don't think 3/5 aganst suits is so common anymore. 2/4 leads are now common, with the understanding that when the lead is low from 4 the hand contains at least Jxxx. From a weaker holding, such as Txxx, the lead is second highest. With T9xx maybe the T, but with T8xx the 8. I am not all that fond of this approach,but from what I have seen it is the one that is most often played when no discussion has occurred.

These 4/2 leads have their meits or they would not be so opular, but I think the merits are a bit oversold. The idea is that leading small with J or better and second with T or less  will help pard in judging whether the lead was from an honor. But does it? With this agreement I lead the 5 from, say, Q85 because I have an honor. But I also lead the 5 from 85, this time because it is second highest. Against a NT contract I am not all that likely to lead the suit at all holding Q85 so I think the 4/2 convention has more merit against NT contracts. Against suit contracts I still like 3/5.
Incidentally, on that 2NT hand that I started with, the bot led the 2 from 3 small so maybe my robotic buddy is not so fond of the 2/4 agreement even against NT contracts.

[wackojack]

My experience of the bots is that they lead low from ALL 3 card suits that do not have honour sequences.  The down side of this approach is that partner is none the wiser the upside is that declarer does not know either.  Also the bots are adherents to "Bird Anthias" research.  This has studied thousands of 1NT-3NT hands and the double dummy stats say that you should NOT lead from a 4 card suit with unsupported honours.  In other words with no 5 card suit prefer to make a passive lead not giving away a trick.  If partner has a king or queen that is chopped, then it will always be finessed anyway.  That is the argument. The 10 is officially an honour and most players in UK do lead low from a 10.  MUD is popular but usually you can only find out if the lead is MUD or 4th when the 2nd round is played and then the damage might be done.  GIB gets over this by always leading low.  So you have to work out whether or not it is 4th highest from a 5 card suit or low from 3. 

[kenberg]

I have thought a little about this idea that leading from xxx might be better than leading frm Qxx, assuming we lead small from either holding. Perhaps so. But there are some aspects to it that the Gib double dummy analysis might overlook. Suppose we get a spot card lead and we have all the hig cards in that suit, so after we see that it was not a short suit lead we kow it was from xxx. Now suppose we have to guess where the Q is in another suit. There is a restricted choice aspect to this. If opening leader had xxx in the other suit as well, then he might have chosen that suit to lead. This increases the odds that the Q that we are trying to find lies with the hand that led from xxx.

I have seen this reasoning applied in the literature but often I think it is applied incorrectly, because if the guy on lead had xxx in one suit and Qxx in another, many would lead from the Q, at least in some situations. If the bots can be depended upon to prefer xxx to Qxx  then I think the Restricted Choice argument has merit.


Also, it seems to me that with some hand on some auctions we need to be aggressive, on others we might well look for a lead that gives nothing away.

Anyway, yes the bots do like to lead from xxx. But the other day a bot led small from QJxx. And with the KT8 on the board I did indeed play the 8, covered by the 9. Oh well.

[infidel]

I believe the bots play minor-suit Stayman, so 2N is a straight transfer to clubs. That said, the GF 3d was an overstatement, to say the least. However, I do think the bot blew the hand at trick one: If he runs the lead to the AS, then plays 2 rounds of clubs, he has 2 board entries in spades to set up and utilize the club tricks.

I sorta like mSS, but it does leave a hole where the invitational sequence belongs. It seems to me the hand is worth an invite, which I'd show via 2c-any-2N;  same sequence if playing 4-suit transfers