Join Bridge Winners
All comments by Craig Zastera
You are ignoring the author of this comment. Click to temporarily show the comment.
In my view, a “limit raise” typically delivers 10+ HCPs.

Perhaps there are some 9 HCP hands (e.g. 4441 with small singleton and 3 kings) that justify “upgrade” to LR status.
But that's about as low as I go.

But there are some shapely hands with, say, 6-8 HCPs that have “loser count” (or whatever other quantitative hand evaluation aids you like for hands with good support for partner's suit) that might be similar to that of (some) LR hands.

Sometimes, it might work out “OK” (in that the correct final contract is reached) to pretend that these hands are Limit Raises too.

But sometimes it won't because a flattish hand with 11 HCPs and four card support will not always be functionally equivalent to a much more shapely hand with 7 HCPs and 4 card support.

Thus, it can be useful if an appropriate system can be designed that can distinguish these two hand types.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Gary,
Glad you liked the simulation.

But here is an interesting and somewhat disturbing fact.

When I originally posted the results, I showed 6 outperforming 5 by some 1811 IMPs (VUL) over the 1000 deals.
That was the figure reported by the “DealMaster Pro” built-in Excel spreadsheet analysis by summing the column labeled “vul_imps”.

However, as I later reflected on that figure, it seemed like quite a lot considering that 6 made only 52% of the time.

Each of those deals should be a 13 IMP gain for 6 vs. 5.
But when 6 is -1, that should be a 12 IMP loss per deal.
Then there were the few (41) deals where only 10 tricks were available in s–a 3 IMP loss for 6 vs. 5 on each.

When I used the actual number of deals where 10, 11, 12, or 13 tricks made in s to calculate the net VUL IMP gain for 6 vs. 5, I got only 1294 IMPs (I edited my previous post to reflect that result).

But what had gone wrong? Had I simply mis-copied the sum from the Excel spreadsheet?

Unfortunately I had already deleted the original data, but since I was perplexed I did a new 1000 deal simulation.

Fortunately (but not surprisingly), the results were almost the same (just a few deals different for each number of tricks).

This time, I looked more carefully at the Excel spreadsheet data.
I noticed that it was crediting a *14* IMP gain for 6 for each deal where 12 or 13 tricks were available.

What??

Closer inspection of the spreadsheet (in particular, the “b_vul_scor” column) revealed that N/S were being credited with 1620 points for 6 each time 12 tricks were available and 1640 when 13 tricks were there,
instead of the correct figures of 1370 and 1390, resulting in a 14 IMP gain for each instead of the correct 13.

So apparently there is some sort of “bug” in DealMaster Pro that is causing it to miscalculate the scores for making vulnerable slams.
The NV case (“b_nv_scor”) is correctly reporting 920 or 940 for 6 making 12 or 13 tricks.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
I do not believe the 3 “pre-emptive” raise should be based on “nothing” (same with 4 pre-emptive raise).

Thus, I think this hand is fine for 3.

A mixed raise should deliver some real HCP values (i.e. an ace and or a king anyway) as well as some trump value.

To me, a mixed raise is more like a “shape” limit raise–fewer than 10 HCPs (8 is typical), but some good shape and around 8 losers.

I do acknowledge that this hand is a maximum for a NV 3 raise (but vul it would be avg-).
Oct. 21
Craig Zastera edited this comment Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
I do not think this is close.

Of course, one can construct lay-outs where acting is wrong, but I will bet that doubling here will score significantly better than passing over the long run.

If partner has four s and nothing better to do, he can leave it in.

Otherwise, LOTT suggests that we should be competing over their 2,
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Actually, I find the bidding on this deal at least as interesting as the play.

The play point about delaying the finesse in favor of first pursuing a 9th trick elsewhere is a good one but relatively obvious I think.
As long as we are careful not to risk 5 losers or stranding a winner, we might as well postpone the finesse as long as possible since if it is “on” we will always come to 9 tricks.

But the bidding is very interesting.
When I first read the commentary, I thought that the opening hand is so suit and slam oriented (in the context of a strong system), that I was thinking of 3 in reply to responder's GF 2.

But then I read about Kit's systemic minor suit shape-showing continuations over 2, so was OK with the systemic 3.

When partner continued with 3, I was all over Kit's actual choice of 3.

However, my rosy view of North's hand is so strong that over responder's 3NT, I would have persisted with 4.

On the actual deal, since responder has already “stretched” slightly with his GF 2, he would no doubt have back-pedaled with 5 which would have ended the auction (although, his hand, minimal though it is, does have very slammish cards with only the J likely not working hard).

A Deep Finesse analysis of the full lay-out here reveals that 6 makes easily (“easily” meaning that just about any reasonable line will succeed on the actual friendly lie of the cards). It is also possible for declarer to come to 10 tricks in NT, although that is “lucky” and would not be achieved at IMPs in real life (I think).

But this prompted me to do a 1000 deal simulation of these actual N/S cards (with random lay-outs of E/W hands) to compare 5, 6, and 3NT.

In s, the # of tricks makeable (double dummy) was:
13: 32 deals
12: 485 deals (so 6 is 52% double dummy)
11: 442 deals (so 5 is 96% double dummy)
10: 41 deals
—-
1000

In NT, the # of tricks makeable (double dummy) was:
13: 0
12: 0
11: 5
10: 244
9: 423 (so 3NT is 67% double dummy)
8: 328
—-
1000

Doing an IMP comparison of various contracts revealed:
(a) 3NT vs. 5
5 wins by 3600 IMPs ! (3.6 IMPs / deal average)

(b) 3NT vs. 6
6 wins by 4311 IMPs (4.3 IMPs/ deal average)

© 5 vs. 6
6 wins by 1294 IMPs (1.3 IMPs / deal average

Now I am not necessarily arguing that 6 should be reached on these two hands since real-world play will almost certainly not succeed 52% of the time as the double-dummy analysis shows because declarer may have to choose among alternative lines, e.g. setting up s vs. setting up s vs. cross-ruffing. Or possibly dropping a stiff offside K which DD will do but RW declarer won't.

However, I do believe that it is clear from this simulation that 5 is a much preferable contract to 3NT.

Why even at *matchpoints* 3NT beats 5 on only 244 of the deals, while 5 beats 3NT on 548 deals (the remaining 208 deals being ties).
Oct. 21
Craig Zastera edited this comment Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Aren't Kit's systemic rebids after 2:
2N: with 5 s and 4 s
3: with 4 s and 5 s

??
It is true that that scheme might wrong-side an eventual 3NT contract when opener has all his values in the minors with weak holdings in both majors (e.g. xx-xx-AQxxx-KQxx)
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Steve,
That is an excellent example. I would not double 2 with that hand either.

Not so much (this time) because I fear they might XX and play it there (but that is certainly not impossible), but because a lead from :Kxx(x) might lead to two fast tricks and a ruff.

For me, to double the Stayman 2, I would require a hand such that:
(a) chance of their playing 2XX successfully is almost
nill (meaning, probably, 5+ strong s)
and
(b) it must be almost certain that a lead will be
best. Hence, nearly all my strengh in the suit.

Something like :AKQxx would be ideal. As I said, I *might* risk it with :AKQJ.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Marshall,
Personally, I doubt I would ever double 2 with only four s (although I did consider adding :AKQJ to the mix).

Also, the porous 5 card suit you suggest (KJ876) seems risky, particularly without any specifications about the rest of the hand, because some other lead might be better.

Nothing worse than putting partner off of his normal (winning) lead by directing another only to find that the directed lead was wrong.

Keep in mind that the opponents could have as few as 23-24 HCPs between them on this auction. Therefore, South might have as many as 9-10 HCPs (particularly on the relevent deals where 4 is in fact beatable).
Thus, even if his s are in fact KJ876, he might have sufficient high cards in other places to make a non-club lead superior.

Nevertheless, I would be glad to repeat this simulation using whatever “aggressive” criteria for South's holdings that you consider appropriate for doubling 2.

Just give me a list of appropriate holdings for South (length and honor structure) for which you would consider doubling 2 to be “clear”, and I will do a simulation with those constraints.

But, in cases where South's holding has few enough HCPs that he can have significant HCP strength elsewhere, you should also specify what outside honor structure (and shape) would cause you to forgo doubling 2.
Oct. 21
Craig Zastera edited this comment Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Jeff,
You are probably right that simulation should reject deals where South would have doubled 2. Of course, to do that I have to decide the criteria for such deals.

I think there is quite a bit of variation amongst players as to how aggressively they double a Stayman 2.
Nevertheless, I attempted to do a new simulation with added conditions to eliminate South hands that would “obviously” double the Stayman 2.

The constraints I added on South were:
reject deals where South's s are:
(a) 5 headed by AKQJ, AKQ, AKJ, KQJ, AQJ
(b) 6 headed by above plus AK, AQ, KQ
© 7+ headed by above plus AJ, A, KJ, QJ

I generated 1000 deals with these added constraints.
4 was beatable on 325 of those deals.

Here are the # of deals where each possible lead can lead
to defeating 4:
6/3/2: 235
A/5/3: 196
9: 188
Q/J: 238
7: 197
2: 199
T/9: 270

As you can see, the main result (that a lead is best) is unchanged. The only significant difference I see is that the Q/J lead has now slightly surpassed the leads for second place honors.
Oct. 21
Craig Zastera edited this comment Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Abstain, as this hand is easily worth 3 the first time.

Once you misbid, no scientific recovery is possible as partner will never believe you hold this hand no matter what you do later. So you are flying solo now.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Passing is insane. We must have a play for at least (some) game when partner comes in vul at the 3 level over their strong auction.

I see the issue as whether I should just bid 4, trusting partner to have an excellent suit for his bid (which can't be based on too much HCP strength unless an opponent has psyched or they are having a bidding mix-up), or explore for alternative strain(s).

I really think that slam ambitions are excessive here (not impossible, I suppose, but I'm not going to try for one).

Given my pass over (1), I think 4 by me now would likely be interpreted as “fit showing” in order to help partner decide what to do if they compete beyond 4.

This seems like the wrong message to send (particularly since I will be on lead vs. a () contract, hence cannot possibly be making a lead-directing call).

Thus, I think (paradoxically?), I will just raise to 4 with this hand, trusting partner to have a very long/strong suit (I really think he has 7).
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Shows a desire to attempt to take 9 tricks without the benefit of a trump suit.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
I too chose 3NT at least partly on the basis that I do not have a natural 4 available (that would show black suits, at least 5-5 in our methods).

If OP had explicitly stated that 4 is natural for purposes of this problem, that would be one thing.

But since nothing was said about 4, I thought it would be wrong for me to choose that call here since in my methods it would not be available (to show a single suiter).

So for me, part of the interest of this problem is to consider what alternative(s) one might choose on hands where a natural 4 might be appealing if available.

We also use 4 here to show a very strong minor 2-suiter (with 4NT showing a somewhat less strong one).

But the comment about using one of these calls to show a very good or single suiter instead of having two ways to show minor 2-suiters in order to help with problems caused by loss of natural 4m overcall seems like it might be a good idea. Perhaps I will adopt it.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
Must abstain as this hand is an obvious 2 opener.
Once you misbid the first time, there is really no recovery for a scientific auction since partner can never include this hand among the possibilities.
Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
This seems like a pretty good deal for computer simulation (because the hands are narrowly defined), so I tried one (1000 deals). I later repeated this with 5000 deals, so I include those results too (which do not reveal anything different from the original simulation).

Constraints I used were:
North: given hand

West: balanced hand with 4+ s and < 4 s with:
(a) 4 s, 16-17 HCPs
(b) 4 s, 15 HCPs and a weak doubleton somewhere
© 5 s, 15-17 HCPs

East: (a) balanced with 4 s, 8-9 HCPs, not 4=3=3=3
(b) 4 s, other suits <= 5 cards, 8-9 HCPs

Since this is an IMP problem, only concerned with deals where 4 is beatable. There were 347 of those (1644 on the 5000 deal simulation)

Here are the number of deals of those 347 (1644) for which each possible lead will allow 4 to be beaten (with best defense thereafter, of course):
6: 263 (1210)
3/2: 263 (1211)
A: 232 (1034)
9: 222 (1028)
5/3: 233 (1069)
Q/J: 248 (1192)
7: 196 (1037)
2: 197 (1039)
T/9: 292 (1353)

So, looks like is the best lead based on double-dummy simulation.

Note that I did not attempt to factor in possibility of leader's partner doubling the Stayman (2) as a lead director. No doubt, there are some deals where his holding is sufficient that most would have done so.
Oct. 20
Craig Zastera edited this comment Oct. 21
You are ignoring the author of this comment. Click to temporarily show the comment.
I held this hand in the finals of our District 19 flight A NAP.

I did in fact overcall 2.
Partner later criticized this saying that my hand is insufficient for that action.

His hand was: A98-87-KQJ43-QT2
Opening bidder held: JT764-AK52-T98-K
Leaving responder with: Q3-T-A7652-J8643

Although 4 is in fact makeable on this lay-out, it is not really a percentage contract. 3 pairs did bid 4, but only one made it (for a cold top).
Playing in a partscore making 170 was good for 73% of the matchpoints. +140 was still above average (55%).

After my 2, we had a bit of a bidding misunderstanding.
We play “transfer advances” of overcalls.
We had agreed that these would apply after “sandwich position” overcalls (as here), provided that responder had either raised opener's suit or bid NT (i.e. their side has bid only one suit).

Further, we had discussed the following somewhat different auction in detail:
(1)-2-(2)
For that specific auction we had agreed:
Double: responsive, showing minors
2N: transfer to s
3: transfer to s
3: strong raise (LR+, 3+ trump)
3: lesser raise
It is unclear (to me) exactly what our agreements are
if the auction had been (1)-2-(P).
I supposed that here, analagous to above agreements:
2 would be minors
2NT: s
3: s
3: strong raise
3: weaker raise

However, *normally* playing “transfer advances”, NT advances are considered natural, so perhaps *that* should apply on this second auction.

But the actual auction was different still.
(and yet another interesting variation would have been had *opener* rebid 2–would that now put us in the same situation as after (1)-2-(2) ?? Unclear).

Anyway, over my 2 (and pass by opener), my partner bid 2NT intending it as natural and game invitational (a reasonable choice with his hand I think if that is what the call means).
I, however, took it as a transfer to s, and duly removed to 3 (after alerting).
Partner removed that to 3NT, ending the auction.
This was -2 (best he can do) for a very poor result.

Instead of 2NT, he might have chosen 3 (clearly a transfer to s), to be followed by 3 / my 3. This would unambiguously show a decent (10+ or so) hand with good s and doubleton (usually Hx) support.

True, his actual hand has significant black suit values and poorish s for that choice of sequence.

He also might have tried 3 over my 3 on the actual auction, but of course by then UI issues arise.

Anyway, the moral is that agreeing “transfer advances of overcalls” opens up new opportunities for bidding disasters.
My experience is that this convention is quite a bit more complex than it appears on the surface, so partnerships considering adopting it had best discuss it in extreme detail.
Oct. 20
Craig Zastera edited this comment Oct. 20
You are ignoring the author of this comment. Click to temporarily show the comment.
Thanks for all your votes.

My partner held this hand and rebid only 2.

My hand was: A64-JT54-A754-A6
A decent minimum, but certainly no thought of moving over partner's 2.

It turns out that 4 is ice-cold as (my) RHO held:
98-AK83-KQ62-Q83

I think partner's hand is borderline between 3 and 4 as it has only 7 losers. At IMPs, 4 for sure, but at matchpoints I think one could justify a *conservative* (IMO) 3.

Had he bid 3, I think *my* hand is close between PASS and boosting to 4 (at matchpoints). Probably it should risk 4 based on the aces, possibly useful T, and the doubleton (and I think I would have done so, but we'll never know).

I did some simulations of this responding hand opposite various hands opener might have held for this auction.

I think opener's most likely shape for this bidding is
3=4=5=1 (they likely have at least an 8 card fit, although in practice they did not).
If opener has this shape, then:
if he has 11 HCPs ( not K or Q): 4 is 71%
if he has 12 HCPs (any ): 4 is 75%

If opener is only 3=4=4=2 (actual shape), then:
if he has 12 HCPs: 4 is 40%
if he has 13 HCPs: 4 is 63%

The above simulations were only 200 deals each (took a long time to generate deals with appropriate specs to match the actual bidding), so percentages are rough estimates.
Still, I think they strongly support my conclusion that this responding hand is probably a 4 bid, and should certainly have at least invited with 3 (and even that puts a lot of pressure on opener to accept with near minimum hands which will still offer a good play for game).
Oct. 20
You are ignoring the author of this comment. Click to temporarily show the comment.
Technically at this vulnerability, 3 would be prudent (only 7 tricks).
But I think it is worth a little optimism to stretch to 4 because 3 might miss a cold game when partner has as little as, say Ax-Kxxx-xxxx-xxx
Oct. 18
You are ignoring the author of this comment. Click to temporarily show the comment.
I think both North and South bid bizarrely, so I voted “equal blame” and mean “a lot of blame for each.”

First, why would North with a minimum opening hand, nothing in s, and fair tolerance for partner choose to make a penalty double of (3)?

To me, North has an obvious PASS–nothing particularly interesting to say over (3) beyond his 1 opening.

Second, if playing 2/1 GF (OP doesn't say), I do not think the South hand is worth 2. I like 3 (invitational), but perhaps this N/S don't have that method available.

Anyway, after starting with a “light” 2 response based on extra shape and deficient HCPs, passing the double of (3) does not seem right.
With two more s than promised and minimal defense, pulling to 3 seems right.
Oct. 17
Craig Zastera edited this comment Oct. 17
You are ignoring the author of this comment. Click to temporarily show the comment.
That might be a good rule of thumb, but I don't think it is correct here.

First, West has made a VUL 3 overcall with a known poor suit. Therefore, he must have at least 7 s, and 8 is a real possibility.

Second, East failed to return his partner's suit at trick 2.
One possible explanation for this is that he doesn't have any.
Oct. 17
.

Bottom Home Top