1NT and 2NT are not the same

After 1NT-P-2♦ and 1NT-P-2♥, "super-accepting," opener forcing the auction to the 3-level, has become common. Some people do it only with 4-card support and a maximum; some do it every time they have 4-card support; a few people do it with a tip-top maximum and only 3-card support. Some people always jump to 3 of the agreed major (I don't think they should -- see the sidebar), some people assign five different means to the bids from 2♠ through 3♥.

The common thread is that of "Law of Total Tricks Protection": being willing to reach the 3-level with a known 9-card fit, even if it doesn't make, because the opponents presumably could make a partscore their way, and might find it if we subside at 2 of a major.

Reasonable people can disagree about how likely the opponents are to come in at the 3-level after 1NT-Pass-2♥-Pass-2♠. But I think almost everyone will agree the opponents are almost certainly not coming in at the 4-level after 2NT-Pass-3♥-Pass-3♠.

That leads us naturally to our key question:

What hands make game opposite a minimum responder?

By "minimum responder", I mean a 5-card suit and 0-3 HCP. Any hand with a 6-card major, even ♠xxxxxx ♥xxx ♦xxx ♣x, might try Texas rather than Jacoby. Similarly, any responder with a 4-count might try transferring and then rebidding 3NT. Only really awful responding hands pass out 3 of a major. Responder isn't even planning to make 3M, just expecting it to cost less than a hopeless 2NT.

If someone told you to make up a hand of 20 or 21 HCP that would take 10 tricks opposite ♠xxxxx ♥x ♦xxxx ♣xxx, what would you pick? Well, ♠Axxx ♥xxxx ♦AKQ ♣AK fills the bill (and might even take 11.) So does ♠AKxx ♥Axx ♦AKQx ♣xx. With ♠Axxx ♥AKxx ♦AKQx ♣xx you'll need 2-2 spades but have chances. With ♠QJxx ♥KQJx ♦Axx ♣AK you'll do well to win eight tricks.

In short, you want a hand very heavy in quick tricks — aces and A-K and A-K-Q combinations — not a hand with stray jacks and queens, not a hand with a lot of finesse positions where you might not have entries to take those finesses. And, maybe most surprisingly of all, you don't necessary want trump honors. Normally Q-x-x-x is a fine holding opposite partner's suit. A queen in partner's suit is better than Q-x-x in a side suit. But even x-x-x-x-x opposite x-x-x-x plays for 2 losers 40% of the time; for the specific purpose of making a 20-HCP game with a known 9-card fit, you want quick tricks in the side suits even more than you want trump control, and the trump queen and jack are only insurance against bad breaks, not prime values.

I did a series of double-dummy simulations to try to identify the most desirable features in the 2NT bidder's hand. I dealt 2000 2NT openings with 4 spades, and then for each one, dealt a thousand responding hands with 0-3 HCP and five spades, and checked to see how many times 4♠ made. On 731 of those 2000 hands, 4♠ made more than half the time, justifying a superaccept at matchpoints. Here were the 10 best hands:

• ♠A963 ♥A3 ♦AKQ54 ♣A7: 89.8%
• ♠AT87 ♥A9 ♦AKQ43 ♣A7: 89.8%
• ♠AQ82 ♥AT964 ♦A3 ♣AK: 87.1%
• ♠AQT7 ♥AT975 ♦A5 ♣A3: 86.5%
• ♠A962 ♥A2 ♦AKT32 ♣AQ: 86.3%
• ♠AKT9 ♥AK ♦84 ♣AK765: 86.0%
• ♠AKJ3 ♥AKQ85 ♦42 ♣A6: 85.9%
• ♠AK64 ♥AKT73 ♦A3 ♣K5: 85.4%
• ♠AKJ4 ♥T7543 ♦AK ♣AQ: 84.3%
• ♠AK32 ♥AT ♦92 ♣AKQJ4: 84.3%

What do all those hands have in common? They have 21 HCP and no singleton or void, but they are hardly typical 2NT openings, with 5-4-2-2 shape, pure suits (there is only one Kx and one AQ between all ten hands), and either A-K or A-K-Q sequences or doubletons (possible ruffing values) in every side suit. Only two jacks (both in trumps, and probably still making 4♠ even if you replaced the ♠J with the ♠2 on those two hands.) The top five hands all have four aces. If you judge most of these hands too strong for a 2NT opening, you're probably right. You might upgrade them to 22 or 23 and open them 2♣-then-2NT, or perhaps 2♣-then-2♥ on the 4-loser hands with 5-card heart suits.

These aren't "typical" superaccepts; they are extreme examples, even more extreme than my constructed examples three paragraphs above, but they possess all the features we expected a good super-accepting hand would.

So, what does a more realistic super-accept look like?

Here are then hands with a 70% chance of making 4♠ (about 8% of my sample of 2NT openers were this good or better):

• ♠AJT3 ♥AT95 ♦A4 ♣AK8
• ♠AK83 ♥AK3 ♦K3 ♣AT62
• ♠AKJ3 ♥T3 ♦AKJ83 ♣AT
• ♠AQT4 ♥AK96 ♦J5 ♣AK7
• ♠AK94 ♥K85 ♦AK97 ♣A4
• ♠AT94 ♥KJ9 ♦AK ♣AQ54
• ♠A952 ♥AQ74 ♦AK ♣A62
• ♠KJT7 ♥AQ93 ♦AK8 ♣A6
• ♠AQT4 ♥AK ♦73 ♣AK952
• ♠KQJ2 ♥T6 ♦AK ♣AKJ83

These show the same features as we expect them to: lots of 5-4-2-2 patterns, no 4-3-3-3s. Nine out of ten have three aces and the the tenth has a K-Q-J sequence. Lots of A-K combinations; no unsupported queens, only one J-x (serving mostly as a ruffing value in a 21-HCP hand, you'd bid the same with x-x.) Only one hand has two finessing positions (♥K-J-9 and ♣A-Q in row 6).

Here are ten hands with a 60% chance of making 4♠ (22% of my sample hands were this good or better):

• ♠AK42 ♥AK32 ♦K2 ♣A73
• ♠AK76 ♥KQ ♦KJ ♣AJT52
• ♠A943 ♥A64 ♦AK ♣AJ95
• ♠K653 ♥AQT8 ♦AK5 ♣A8
• ♠AK54 ♥AKJ8 ♦Q7 ♣A97
• ♠AQ75 ♥A5 ♦K8 ♣AK874
• ♠AKQ5 ♥92 ♦AQJ72 ♣AT
• ♠KQ74 ♥AKT ♦KT ♣AQ76
• ♠A984 ♥K84 ♦AQ ♣AKT2
• ♠AQ54 ♥AQ ♦83 ♣AKJ73

What do these hands have in common? None of them is 4-3-3-3. Eight of them have at least 3 aces. Only one has an unsupported queen and none has an unsupported jack. Compared to the previous group we see more A-Q and A-J-T type holdings in the side suits.

Here are ten hands with a 50% chance of making 4♠ — the worst hands you would superaccept with at matchpoints:

• ♠AJ92 ♥K62 ♦AQ8 ♣AK7
• ♠AKJ4 ♥KQT52 ♦A2 ♣K9
• ♠AKT7 ♥A74 ♦AJ63 ♣KQ
• ♠KQ98 ♥AQ ♦KQJ53 ♣A6
• ♠AK42 ♥K952 ♦AQ6 ♣A2
• ♠AQ84 ♥AQ ♦A73 ♣KQ94
• ♠AKQ6 ♥653 ♦AQ83 ♣AQ
• ♠AJ85 ♥A64 ♦AK ♣KJ32
• ♠AK85 ♥KT43 ♦AQ ♣A84
• ♠AK52 ♥AJ ♦A73 ♣KQ95

Again we see no 4-3-3-3 hands, most hands having at least 3 aces (both exceptions have 5-card side suits headed by K-Q-J or K-Q-T), and little or no wastage in the doubletons, though side-suit finesse positions have become even more common; most of the hands in this set feature two side-suit tenaces.

At IMPs you can go slightly worse still — down to about a 42% chance of success non-vulnerable, and a 34% chance of success vulnerable — and still show a small positive IMP expectation. Two typical hands with a 35% chance of making 4♠ that still show a small profit vulnerable at IMPs: ♠AKQ9 ♥AKJ ♦J3 ♣K852 and ♠AQJ3 ♥J6 ♦AKJT ♣KJ8. These examples have three flaws each (wasted honors in the doubletons or finessing positions in the side suits,) vs. the two in the 50% hands above.

I did a series of logistic regression analyses, to pinpoint desirable features in your hand. Feel free to write me and ask about the details. They didn't uncover anything too amazing. If you really like point-counting, you can try this: count 4.8 for each non-trump ace; 3 for each non-trump king; 1.5 for each non-trump queen; 0.7 for each non-trump jack; nothing for trump honors; add 1 for 5-4-2-2 shape, deduct 1 for 4-3-3-3 shape. Only superaccept if that adds up to more than 22. (But, seriously, don't blindly count points, in any system. Know that A-K-Q in the same suit is better than A-x-x in one suit and K-Q in another.)

Recommendations

From examining the above hands, your partnership might be inspired to form an agreement such as "our superaccepts will never be 4-3-3-3, will always have three aces, and will never have two flaws (wasted Qxs or side-suit finesse positions)."

If your partnership style is to open a lot of 5-4-2-2 patterns with 2NT, you might consider defining 2NT-P-3♥-P-4m as promising HHxxx in the minor. If that isn't your partnership style, you might consider defining those higher steps as promising a small doubleton, so that partner can evaluate your ruffing potential.

In any case, I recommend (as I do over 1NT) having your "normal" superaccept be 3M+1, and using 4M-1 specifically for a very pure absolutely-no-tenaces superaccept, and using 3M+2, 3M+3 and 4M to show the special feature of your choice in the side suits.

Continuations

One reason I like the idea of promising 3 aces with a superaccept is that it diminishes the need for ace-asking and for 4-level cuebidding. If responder uses Blackwood, he's probably interested in trump quality. If responder prefers to cuebid, he's probably fishing for a particular side king, and isn't afraid of going to the 5-level.

If responder has no slam aspirations, he will sign off, via the retransfer. If responder does have slam ambitions, he can bid a new suit immediately, he can go past 4M immediately, or he can retransfer and then take a third call. I propose the following simple set of agreements, but you can choose a different set if you like:

After 2NT-P-3♥-P-3NT (generic superaccept) —

• 4♣, 4♦: splinters. Opener is expected to sign off with a wasted king or queen in this suit, continue otherwise.
• 4♥: retransfer. Opener must bid 4♠.
• Retransfer followed by 4NT: RKC.
• Retransfer followed by 5 of a new suit: your choice. Control-asking? Exclusion RKC? Idle?
• 4♠: heart splinter --- swapped with 4♥ so that retransfers are possible.
• 4NT: asking opener to cuebid his cheapest king.
• 5♣,5♦,5♥: cuebid (and likely asking opener to cue the next step up if he has 2nd round control in it)

After a small-doubleton-showing superaccept, responder still retransfers if he wants to use RKC, bids on if he wants to cuebid. After the no-tenaces superaccept or the leap to 4♠ (to show a small doubleton heart) there is no retransfer, so we're forced to make the immediate 4NT by RKC here.

Examples

This was Part I: Superaccepts after 2NT-3♦ or 2NT-3♥

Onward to Part II: The third round after 2NT-3♥-3♠-3NT

Onward to Part III: Responder bids two suits after 2NT