Monday, July 30, 2007

Inference from Redundancy, Part IV(B)

My Response Structure to Jacoby 2NT:
For unbalanced hands, Opener can either jump to the four-level, bid Three Spades, or bid Three Clubs.
3S = Minimum, stiff or void somewhere. Responder can next bid 3NT to ask for the stiff. Opener bids a stiff minor. With a stiff in the other major, Opener bids 4H. If spades are trumps, 4S shows a void in hearts. Last train, if available, might be used. Thus, a "minimum" with a club stiff is more flexible than with a stiff in the other major. A diamond stiff is more flexible if spades are trumps.
4-bid = trick source, with a stiff or void in the lower of the two remaining suits. Thus, 4C shows a 5-5 hand, concentration of values in clubs and the major, and a stiff or void in diamonds; 4D shows 5-5, clubs and the major, and a stiff or void in diamonds. 4H, if spades are trumps, shows 5-5 with a stiff or void in clubs.
3C = other shortness. 3D asks. If Opener has a stiff or void, with extra's, he shows the stiff or void. The version I play needs fixed here. I use 3H for the other major, 3S for clubs, and 3NT for diamonds. However, I think this should be 3NT for clubs, 3S for diamonds, and 3H for the other major, because this maximizes the ability for Responder to use 3NT as Serious 3NT when Opener cannot himself distinguish "minimum" from "slightly better than minimum" through 3S and a positive reaction to Last Train.
If Opener has the 5-5 trick source with a stiff or void in the higher suit, he bids 3C and then jumps to the four-level after 3D. So, for example, 1S-P-2NT-P-3C-P-3D-P-4C shows a club-spade two-suiter with shortness in hearts, ...4D a diamond-spade two-suiter with shortness in hearts, and ...4H a heart-spade two-suiter with shortness in diamonds.
With a balanced hand, Opener can bid Three Diamonds, Three Hearts, Three Notrump, or Four-of-the-Major.
Four-of-the-Major is weakest. Three Diamonds shows most non-minimums. 3H shows 5422 and just above a minimum. 3NT shows a sixth trump and just above a minimum. Cuebidding follows all.
Summary:
The structure helps to gain more definition to each call, but the approach is still quite similar to standard, in a sense. With balanced hands, Opener is making primarily a quantitative bash. With an unbalanced hand, Opener is making a more defined shape bash. With the trick-source hand, Opener is still making a Picture Jump, but with more definition to his call (as to the shortness).
Because of this structure, we have some improvement, but it seems apparent that Responder will want to have a hand that can handle both shape bash and quantitative bash bidding. That analysis will follow next.

Sunday, July 29, 2007

Inference from Redundancy, Part IV(A)

The last issue is the inference from the redundancy of a Jacoby 2NT auction alternative. This part has been the most difficult to assess, for several reasons. First, I have a gut dislike for Jacoby 2NT, even though I play it. I have a tendency to avoid Jacoby 2NT, whereas it seems like many have a tendency to jump right to Jacoby 2NT. I'm not convinced that Jacoby 2NT solves the world's problems.
The second problem in describing the redundancy from Jacoby 2NT is in describing when you would use Jacoby 2NT, which in turn turns on the systemic responses to Jacoby 2NT.
From a theory perspective, it should be relatively easy to understand the issue with Jacoby 2NT. The normal responses force a two-way bash, to a degree. In the old style, there were two "bash" bids induced. First, one could bid a stiff or a void, jumping with the void. This is a "shape bash" bid. Second, one could use a quantitative bash of 3M, 3NT, or 4M, all showing HCP ranges without shortness. The newer school has combined the stiff and void bids into the non-jump, with a trick-source jump, akin to a Picture Jump. That's getting closer to my tastes.
The assessment of the impact of Jacoby 2NT will require several steps. First, I will describe the response structure that I currently use after a Jacoby 2NT raise. This response structure is not critical, as many similar structures do exist and are useful. However, it may be of interest to some, who perhaps are playing a much less involved structure and, accordingly, are finding that Jacoby 2NT auctions leave a lot to be desired.
The second step will then be to describe what hand patterns for Responder are best handled with a Jacoby 2NT response, considering the response structure.
The third step will then be to assess what impact this tool has on alternative auctions. For, it seems clear that a cuebidding sequence will never be selected when Jacoby 2NT is ideal. Thus, when a cuebidding sequence does occur, and when Responder's hand starts to look somewhat like a hand suitable for a Jacoby 2NT auction, we will then have an idea of what "Jacoby 2NT flaws" might have justified this alternative route.
As an aside, I had some difficulty deciding which came first, the chicken or the egg. One could decide to look at this problem in the same way that I analyzed "shape bash" splinters and delayed splinters. In other words, just as shape bash splinters handle problem cuebidding hands, Jacoby 2NT could also handle problem cuebidding hands. However, for some reason that I cannot quite articulate it seems that the reverse should occur when analyzing Jacoby 2NT. This may be thinking based upon years of being in with the "straight to Jacoby 2NT" school. However, it seems to me that Jacoby 2NT caters to bash bidding by Opener, and that therefore the reverse is true. One partner bashes, the other caters. So, because the structure for Opener is "bash," Responder should generally only bid Jacoby 2NT with a hand that is appropriate for decision-making after the bash.
It also seems apparent that Jacoby 2NT is a captaincy-grabber. I generally dislike bids that grab captaincy as your first call. There ought to be a very strong reason to make that move.

Sunday, July 22, 2007

Inference from Redundancy, Part III

So far, we have discussed the nuances of two forms of "shape bash" bids. The first was a "12-14" delayed splinter, through 3C. The second was a direct splinter in immediate response to a major opening. The latter (direct splinter) could be used for voids only. Alternatively, the delayed splinter (3C...stiff) and the direct splinter each show three of four holdings (or more):

1. Stiff is not a stiff honor
2. Minimal trump honors, top three (delayed = one top honor, direct = two top trump honors)
3. Suit for normal 2/1 (minor) has only one of the top three honors
4. Fourth suit has no 1st/2nd-round control

The 3NT Direct Raise -- Another Shape Bash?

The next issue is the 1M-P-3NT auction.

This auction has been troubling for me for some time. The typical definition of an auction 1M-P-3NT has a HCP range for 3NT, sometimes the light variety (maybe 13-15) and sometimes the strong variety (maybe 16-18), but no definition of hand type is provided other than "balanced." No discussion of the honors expected is done.

Consider, for instance, the wild difference between these two "13-15" hands, in response to a One Spade opening:
.
1. Axx QJx QJxx QJx
2. Jxx Axx Axxx Axx
.
These two hands seem to be wildly different, of course. Imagine a partner holding KQxxx Kxxxx Kx x. Opposite the first, partner cannot make more than 4S. Opposite the second, the slam is very strong odds to make. How, precisely, does Opener explore his options, if 3NT says nothing about strength?

I have forced partners to pick one or the other -- Aces and spaces OR quacks -- as the "honor expectancy" of this call. Either might make sense, but playing that you might have one, the other, or a blend of these seems unplayable.

In any event, whatever you select tailors cuebidding that does not elect that option. Discuss partner's preference, and then elect a style.

If 3NT shows Aces-and-spaces, then cuebidding sequences will not ever show a 4333 3-card range with Aces-and-spaces. If 3NT shows Quacks, then cuebidding sequences will not show a balanced quack hand.

Saturday, July 21, 2007

Inference from Redundancy, Part II

Direct Splinters
The question raised has been an analysis of why one would use a direct raise convention rather than starting a 2/1 auction, and what inferences can be gleaned therefrom. The first issue discussed was the 12-14 splinter through a 3C raise, showing three or more of (a.) only one top trump, (b.) only one top card in the suit that would otherwise start the 2/1 auction, (c.) no honor in the stiff suit, and/or (d.) no control in the fourth suit. A 2/1 auction with a stiff, therefore, will show more than 14 HCP or that two or fewer of these four statements can be made.
The next question concerns the direct splinter, e.g., 1H-P-4C. This could be played in at least one of two ways.
One method that I have used is for this direct "shape bash" bid is used exclusively to handle voids on the side. This does have some merits, as hands with voids are often tough. If the range is still 12-14 (which makes sense), then the parameters of three of the four criteria for 3C...Splinter auctions also makes sense. You just will always know that one of the four (not an honor in the short suit) will always be present, of course.
Another method is for this call to show extra strength, perhaps roughly 15-16. I hate HCP ranges, because these make little sense to me. I often overlap for disclosure reasons. The hand is "15-16" if it looks like that playing strength.
In any event, the parameters of the "three of four" from above can be most easily tweaked for the direct splinter by increasing the strength of the trump contribution to two honors. Thus, whereas 3C...Splinter would typically show one top honor, plus the rest of the criteria being met, the direct splinter would show the same criteria but two top honors in trumps.
That appeals to me because these hands often start with a cuebid of three of the major, showing good trumps but nothing else to cuebid. It wastes a cue, as otherwise three of the major would also be a cue. An example. 1H-P-2C-P-2D-P-2H. Opener bids 2S. Responder, with something like QJx-AQxx-x-AJxxx, could not cuebid 2NT (poor trumps), 3C(two top trumps), or 3D(diamond A/K/A). He cannot Picture Splinter. So, he cuebids 3H, which becomes a redundancy because merely bypassing 2NT showed the two top hearts.
Direct splinters, which are "shape bash" bids, should be reserved from problem hands (voids) or for problem cuebidding hands. That serves two purposes, again. First, partner can better read your likely hand if you splinter. Second, partner can better read your necessary hand parameters if you do not splinter.

Thursday, July 19, 2007

Inference from Redundancy?

A few troubling thoughts have kept me up at night over the years, and I believe to be approaching somewhat of a resolution of my angst. I thought I'd share this.
Angst #1: What is 1S-P-3NT? I'm not satisfied with a simple "13-15 balanced with three-card support." This begs the question of whether this is quacks or primes, or a blend. If Opener is balanced and interested in a notrump quantitative bash, this is not that relevant. If Opener is shapely, this is critical. Few answer this question.
Angst #2: When would you elect to Splinter immediately instead of bidding a 2/1 GF and then later cuebidding (or Picture Splintering)?
Angst #3: When is Jacoby 2NT best?
Angst #4: When is 3C, as a route to show a 12-14 Splinter, a better start than a 2/1 and cuebidding, or Picture Cuebidding?
The "definitions" of these calls (direct Splinter, Jacoby 2NT, 3NT, and 3C) do not seem to be tied to theory or strategy, or prepared auctions. The failure to elect these options is rarely, if ever, discussed as tailoring and further refining cuebidding sequences where these options are not elected. I'm not satisfied with that, so I plan to explore this a bit, in a series of posts, where I vent rambling thoughts as I myself think this through.
Delayed Splinters through 3C
The issue popped up recently as a result of my analyzing a specific bidding problem, which relates to the 3C call with an included 12-14 splinter option. So, I'll start with that.
As a recap, my technique uses 1M-P-3C as a multi-faceted bid, showing either 9-12 with four trumps or a 12-14 splinter. 3D asks. 3M shows a minimum limit (9-10 if you will). 3NT shows a maximum limit raise with good controls (primes), meaning 4+ controls. 4M shows a maximum limit raise with poor controls. Four of a new suit is a mini Picture Jump, with typically 5422 shape, a concentration of values in trumps and the side suit, and no controls in the other two suits, maximum limit raise ("maximum" in the sense of not good enough to GF). Three of the other major, the missing bid, shows a 12-14 HCP splinter, which Opener can ask about.
So, why splinter through 3C rather than cuebid?
This call should be a prepared call. Imagine a typical hand. You have QJx AJxx x KJxxx, and partner opens 1H. The alternative option is to bid 2C. You expect a likely 2D call from Opener, which would be nice because you can then set trumps at the two-level by bidding 2H. Life is good.
But, what will happen next? All times that it matters, Opener will bid 2S. You will then cuebid 2NT, showing poor trumps (not two of the top three honors), because you are not right for a Picture Splinter. Let's now assume that partner bids 3C, the cheapest cue possible, which could be based upon the Queen of clubs. You are still not right for a Picture Jump, because although you have clarified the poor trumps, your clubs are also poor. You cannot cue 3D, because you lack a diamond honor. You could cue 3H, though, and show that you have one top heart honor. Great! A cuebid!
Now, suppose Opener makes the cheapest cue available again, 3S. What next? You are too weak for 3NT. You cannot cuebid 4C, because you lack two top clubs. The best that you can do is to cuebid 4D, showing a diamond control that is not an honor -- shortness. Thankfully in this one auction you could at least show this! In any other auction, you could show nothing to partner except the one top heart. Not very impressive, eh?
So, why splinter through 3C rather than cuebid? This hand provides the solution. Responder will "shape bash" a 3C...Splinter auction when he lacks ability to make meaningful cuebids in a predicted auction. The features leaning one toward a "shape bash" 3C call might include:
1. Not two of the top three cards in the suit you would pick for a 2/1 start.
2. Poor trumps (not two of the top three honors)
3. No control in the side suit
On the other hand, you might tend to start with a cuebid with the stiff being the Ace, King, or Queen, because you might elect to show that card as the honor in support of a suit he bids rather than as a stiff.
This list is not a checklist, of course. The idea is to think through plausible auctions that will occur after a 2/1 and to elect the "shape bash" of 3C when cuebidding will be a mess. This will typically occur when perhaps three, or all four, of the criteria above are present (poor trumps, no side control, not two top in the suit you would initiate, and perhaps a stiff honor).
Thinking this through, and discussing this with partner, leads to some inferences. First, a "shape bash" of 3C...Splinter implies three or more of the four criteria above. Opener can better evaluate his holdings this way. Conversely, a 2/1 auction that suggests a likely stiff or void will be done with a hand that does not have three of the four criteria.
Consider the example, again. QJx-AJxx-x-KJxxx. This is a great hand for a "shape bash" auction. You start with 3C in response to a 1H opening. Partner bids 3D, asking. You bid 3S, partner asks (3NT), and you show (4D). Partner will now expect that you have only one top heart, that your diamond stiff is not an honor, and that you only have one of the top three honors in each of the back suits, one of which is the Queen. He will expect that one of these assumptions, but no more, might be inaccurate. Very nice, eh?
What about the alternative? Suppose something about your hand is different. So, you bid 2C. Partner bids the expected 2D, and you set trumps with 2H. Partner cuebids 2S, and you show poor trumps by bidding 2NT. Partner shows a club card by bidding 3C. You deny a diamond card by bypassing 3D to bid 3H, showing one top heart. Partner cuebids 3S. You bypass 4C because you lack a second top club and bid 4D to show the stiff diamond. What is different about your hand? Why did you not splinter through 3C initially?
We know that you have a stiff diamond and that this stiff is not an honor. That favors a "shape bash." We know that you have precisely one top heart honor, also favoring a "shape bash." We know that you lack two top clubs, again favoring a "shape bash." So, you must have more than 14 HCP's. The most you can have outside of spades would be AJ in hearts, J in diamonds, and AJ in clubs, for 11 HCP's. So, you must have 4 HCP's in spades. QJ does not do it. So, you must have, at a minimum, the King. Because Opener cuebid spades twice, he seems to have the stiff Ace or a void, but he can read your spade King by the "redundancy" inference.
What if, instead, the auction were more strained. Suppose that partner's rebid had been 2H. Now, you would have to bid 3H to support hearts. If Opener bid 3S, you would bypass 3NT (non-serious) and 4C (not two top clubs) to cuebid 4D. Now, Opener would have a somewhat vague picture of your hand, but he would be able to assess that you did not have a "shape bash" suitable holding, meaning something different. Whatever his hand is, and whatever his needs are, he would be better placed.
So, the 3C...Splinter should be reserved for "shape bash" hands, meaning three or more of the criteria from above. Not using a "shape bash" 3C approach suggests a stronger hand (15+) or that two or fewer of these criteria are present.

Friday, July 6, 2007

Denial Cuebids at 5-Level

The logic of an auction can call for a "denial cuebid" even though that has never been discussed. A case in point is a slight tweaking of a prior post here.

Suppose that Responder bids 4NT as 1430 RKCB for spades, in an auction where the partnership has definitely ruled out the possibility of a grand. Suppose that Opener's response is 5C, contextually showing three key cards. Suppose that contextually this must include the Ace or King of trumps but that Opener has already denied two top trumps (perhaps a prior 2NT cuebid, for example).
Now, as mentioned in the prior post, 5D cannot ask about the Queen of trumps, as that is already known -- Opener has already denied the spade Queen. If Opener has already denied a diamond control, then, as mentioned earlier, 5D must be asking for a tertiary control, the Queen, to protect his King-empty.
If you think of Last Train as a "denial" cuebid of sorts, this becomes fairly easy. A Last Train call, in a sense, is a "denial" of sufficient extras to embark on the next course. It seeks something from partner.
Well, a similar concept occurs here. There are three ways to bid after 5D. Opener could deny the diamond Queen by bidding 5S, a sign-off. Or, he might bid 6S, accepting (or perhaps some esoteric other call that offers a choice of slams or suggests 6NT). Third, and key, Opener could bid 5H.
Now, as Responder, in this auction, is asking the questions, it makes little sense for 5H to be taken literally as a "Last Train" call, because that implies a vantage point that is errant. Opener is responding with Responder as captain; Last Train is used by the captain.
Rather, it seems to me that 5H should be a denial cue, showing the diamond Queen (a positive response) but "denying" ability to accept a hypothetical Last Train call. Accepting the slam try (or making an esoteric choice-of-slams call) implies ability to accept the hypothetical Last Train call.
Look that this another way. After 5D in the example, Opener would bid 5S (sign off) without the diamond Queen. If Opener has the diamond Queen, he will make a positive answer. Opener's task is then to imagine that he could make an insufficient bid of 5D to answer yes, allowing Responder to bid 5H as Last Train. If Opener would accept that Last Train call, Opener bids the slam. If Opener would decline, he bids 5H, below 5S and indicating that he would decline a Last Train call. Responder may have enough anyway and continue.
What this does is to allow Responder to ask Opener is he has the diamond Queen and that "something else" that would make the slam worth bidding, all with the one 5D call. If all Responder needs is the diamond Queen, he will not care about that "something else."
If you recall the problem (this month), Opener held Qxxxxx in spades and the diamond Queen. After the 5D asking bid, Opener could bid 5H if he had held the diamond Queen but only Qxxxx in spades. With the diamond Queen and a sixth spade, however, he has that "something else" to justify bidding slam.
Once accepting, however, Opener might suggest 6NT by bidding 6C (if he had AJ in clubs), 6D (if he held QJ in diamonds), or 6H (if he held AJ in hearts). There is no cost to this, and this probably should be considered. Except, I suppose, that partner's head might explode.

Tuesday, July 3, 2007

An interesting problem hand posted on BBF leads to an interesting auction:

Opener: ♠QJ10xxx ♥Axxx DQx ♣A

Responder: ♠A9xxxx ♥K DKx ♣KQxx

The auction seems rather straight-forward for the start:

1♠ (easy opening)
2♣ (I hate 2NT with 6-4 pattern)
2♥ (natural)
2♠ (sets trumps -- start cuebidding)
2NT (not two of the top three spades)
3♣ (two of the top three clubs, and Opener's weakness in trumps is not fatal)
3♠ (no diamond control, not two top hearts, but one top spade honor; note that this promises two or more diamonds, limiting pattern to 5422/5431/5440/6421/6430/7420/6520/5530/5521)
3NT (serious interest)
4♣ (one of the top three clubs; pattern is now 5422/5431/6421/5521)
4D (diamond control)
4♥ (heart control -- Ace or King)

Note that 4♥ is not Last Train because only the Serious 3NT bidder can use Last Train.

Responder now knows that Opener has the club Ace, the heart Ace, and one of the top spades. He also knows that Opener does not have the diamond Ace or a diamond stiff. Opener, therefore, has: (K/Q)xxxx(x)-Axxx(x)-(Q/x)x(x)-A(x)

So, Responder wants to know (a.) if Opener has the spade King or a sixth spade and (b.) if diamonds are protected.The five-level should be relatively safe, although admittedly we might lose a spade and two diamonds with a diamond lead.

A better course, however, may be a simple quantitative 5♠ call. As Opener cannot have another club card, spade card, or heart card of any interest, and cannot have the Ace or King of diamonds, or diamond shortness for that matter, the only interesting unknown card is the diamond Queen. As 4NT would clarify the spade situation as far as honors and would allow Opener to show the diamond Queen (see above), the inference is that Responder has play for slam opposite less than this. It is a tight inference, but there nonetheless. In the end, the sixth spade and probable useful diamond Queen should be enough to accept, whether on general principles or because Opener actually figures out the (esoteric?) nuance.

Of course, one might say that Opener needs Kxxxxx and the diamond Queen to accept. But, if Responder needs that hand, he would sign off at 4♠. Remember that a cuebid (4D) that yields the only possible cuebid (4♥) cannot result in a "sign off." 4♠ after 4♥ would still be ever-so-slightly forward going.

That last point should perhaps be restated. If the style is conservative, Opener being concerned about diamond-diamond-trump as three losers and not entering the five-level, Opener should probably bid on after a 4♠ call from Responder. A 5D cue, showing the diamond Queen, might work well. Then, Responder could last train if he needed Kxxxxx or simply bid the slam otherwise. Opener would surely not cue 5D with Qxxxx-Axxx(x)-Qx(x)-A(x).

Sunday, July 1, 2007

Three New Possible Ideas for Saving Space

Part One -- Serious Three Spades?

Borrowed from the ideas of others, there are a few ways to tweak your cuebidding techniques that may save some space when it matters. These are not ideas that I would strongly endorse, as I do not use them myself, but they were interesting suggestions that I have received from others.

The first is to use a call of 3♠ as a “Serious 3NT” substitute. A simple auction might be 1♥-P-2♣-P-2♥-P-3♥-P-?

Using this approach, Opener’s meanings for 3♠ and 3NT would be reversed. Thus, 3♠ would not show a spade control but would show generally serious slam interest, with or without a spade control. 3NT, then, would be a non-serious cuebid and would artificially show a spade control.

If Opener bids 3♠ as a general serious slam try, then Responder can bid 3NT as an artificial technique to show a spade control.

This accomplishes a few changes. Without this change, 3♠ by Opener would show a spade control but would not clarify his slam interest. 3NT, in contrast, would show serious slam interest without a spade control. Under the new approach, 3NT would clarify that Opener has a spade control, but without serious slam interest, whereas 3♠ would clarify serious slam interest without clarifying whether Opener has a spade control.

This does nothing to assist Opener in checking on a spade control when he is serious but lacks a spade control, for he could just as easily bid 3NT, denying a spade control, and hear a sign-off that closes the auction. It might help find a second spade control, however, such as when Opener has a spade control but wants to know if Responder has one himself. It also may help to more rapidly end auctions where both partners have minimal interest; Opener’s cue of 3NT (spade control, not serious interest) might be enough because of the non-serious interest for Responder to simply bow out and not continue with cooperative cues.

Personally, I’m not sure if these mild possible gains outweigh the artificiality, but it is something to consider.

Part Two -- Poor Trumps Two Spades?

Another idea to save space would be of a similar type, at the two-level. Consider the auction 1♥-P-2♣-P-2D-P-2♥. One could reverse the meanings of 2♠ and 2NT for Opener (and change the continuation by Responder after the new 2♠). 2♠ would then deny two top heart honors, 2NT showing two top heart honors and a spade control. After 2♠ by Opener, 2NT by Responder would be a spade cue (control).

This idea is much more interesting to me, primarily because knowledge of poor trumps may justify an immediate signoff after this new 2♠ call, with no information exchange useful to the opponents.

However, I am still concerned about adding more artificiality. Also, I like the fact that Opener’s bypass of 2♠ to bid 2NT will shut down a number of auctions where Responder otherwise would make a fruitless Picture Jump (because of no spade stopper). This new idea would result in a larger number of Picture Jumps after 2♠.

But, again, something to consider.

The third new idea may also be of interest, perhaps.

There are some auctions where the partnership might want one or the other partner to ask the questions at the RKCB level. Suppose, for example, that hearts are trumps. Suppose, as well, that one partner bids 4D, eventually, as a Last Train call. Suppose, finally, that his partner may now want to answer RKCB rather than ask, as his answer to RKCB will fully describe his hand for his partner.

In that scenario, some folks bid 4♠, either as a cue or simply as a request that partner ask the questions. This has some merit. It preserves 5♣ and 5D as meaningful calls (perhaps Exclusion, perhaps a cuebid, or perhaps my suggested variation on RKCB – this suit focused for key cards.

However, if you are in an auction where none of the alternative meanings for 5♣ or 5D could possibly make sense, or if those meanings are not your preference, then you can save space by the “ask-or-answer” technique I recently heard about. The idea is for 4♠ to be Kickback RKCB and for all other calls to answer a hypothetical RKCB.

This yields the following structure, again used after 4D as Last Train:

4D(Last Train)-P-?

4♥ = declines
4♠ = I accept, but I’m asking the questions here
4NT = I accept, 0 or 3 (or reverse if 1430)
5♣ = I accept, 1 or 4 (or reverse if 1430)
5D = I accept, two without
5♥ = I accept, two with

You could do the same basic thing after a 4♥ Last Train call, with spades trumps, or possibly in some minor-suit auctions. You could use this technique in other non-Last-Train auctions, as well.

Personally, I see the most merit to this approach in one specific auction. That would be a Serious 3NT, followed by a Last Train call by that person. However, the person who provided the example of a “good auction for this technique” provided an auction where nothing was really gained.