What strength should a 1NT open be? Do you really need it? Comparing sequences that start with 1♣ with sequences that start with 1NT, you find that there is nothing you can show after 1NT that you can't show after 1♣, and indeed the 1♣ gives more fit-finding options at a lower level. So maybe you don't need a 1NT open?
To answer this, I thought it would be worthwhile to start from first principles to work out the logical strength for NT openings or rebids. In this analysis, I am assuming typical flat hands, ignoring long suits that run for more tricks. Of course you have to use judgement to vary the numbers depending on the hand.
There are different ways to bid NT. You can open 1NT, open 1 of a suit and rebid 1NT, open 2NT, open 1 of a suit and rebid 2NT. Without any preconceptions, what are the most efficient ranges for these? To make 3NT with balanced flat hands you would expect to have 25 points, to make 2NT 22.5 points, to make 1NT 20 points. (Either side can typically make 1NT with 20, the advantage being with declarer.)
An obvious point to note is that there is room for game inquiry over a 1NT open or rebid, but no room over a 2NT open or rebid. Responder here bids game or passes. For this reason the opener range for a 2NT bid must be minimal, and 2 points is practical and realistic (like the Benjamin 2NT bids).
A starting point has to be "what strength is a minimum open?" This analysis adopts the common agreement of 12 points to open. The NT ranges would be different if it was not 12. If an opening is 12 points, for "safety" responder needs 8 to reply, but 6 is better because going 1 off is often better than the opposition making their 2-level contract. Take 6 hcp as the minimum reply. Again, this is the standard.
As we could have the lowest range of NT shown by 1♣(say) then 1NT, and a higher range by a 1NT open, or the other way round, let's start by looking at the next higher range which is 1♣ followed by 2NT. How strong should this be? Responder will bid with 6 hcp, and to make this useable for 2NT, opener must have minimum of 17 hcp. A 2-point range therefore means 1♣ followed by 2NT is 17/18.
This means 12-16 needs to be covered by our two lowest ranges. Splitting this into a 3-point range and a 2-point range, this again can be either way round. The 3-point range will need an invitation, but a 2-point range does not really need one : is the presence or absence of a Jack in a random suit going to make a significant difference? Not really; with a 2-point range you can just decide on game or not. Alternatively, the whole 12-16 may be covered by just a suit open and a 1NT rebid, but how wide can a range be to have effective invitations? Responder will invite if a combined 25 hcp is possible, but should not do so unless there will be a minimum of 23, as you must be able to play in 2NT. This means a 3-point range for opener is needed for an effective invitation. We therefore do need two bids for this range.
( Note therefore that the Acol wide-ranging 1NT rebid of 12-16 is wrong. If opener can have 16, responder has to make a forward move beyond 1NT with 9, and if opener has only 12 then you are playing 2NT with a combined 21 count. Very bad indeed! Moreover, if opener then bids game with 15, you are playing 3NT on a flat 24 count. This is often wrong, too. So let's try responder making a forward move with 10 points, and passing with 9. Now the 2NT contract is better, but still a risky 22. And you are now missing 25 point 3NT games. Who invented this silly idea?)
This answers one of the initial questions - you do need a 1NT open.
If the 1NT open has a 3-point range - what range is best?
If you open 1NT and partner has a hand that can bid or invite game, then you have no problem. Everybody has methods for this. The problem comes when partner has a weaker hand, so has to pass. With a weak hand you cannot determine the best contract and stay below 2NT.
Yes, the 1NT open is pre-emptive, but when you open 1NT you already expect the hand to be yours rather than the opponents (statistically your side has more points than theirs), and you are pre-empting partner more than you are pre-empting the other side. For this reason, you want partner to have at least an invitational hand the majority of the time. Let's look at this with regard to the strength needed to open 1NT.
If you have an average of a 13 count [Acol 12-14], partner is statistically expected to have 9 points, so 11+ for a game invitation is comparatively rare. This is a waste, as all he is going to do 70% of the time is pass or make a guesswork takeout. You don't find your major fits. Compared with people who are playing a strong NT on these hands, you will see on the travellers that your 1NT tick scores badly against their 2♥ tick.
If you have an average of 16 [standard strong NT 15-17], partner typically has 8 hcp, so he is already at the invitational level. You can use Stayman. More than half the time you will have a sensible auction.
As is apparent, a weak NT is waste of space in a constructional sense, and as the opening range gets stronger the better it is. When constructing a system, this has an influence. 1NT should be strong.
Our 12-16 range that needs covering could therefore be split as 12/13 opening 1♣ and rebidding 1NT, with a 1NT open of 14-16.
How about a 1NT open with a 2-point range?
This is better still, as you do not need an invitation. Responder can use Stayman and rebid 2NT to play, and the immediate response of 2NT is freed to have some other use. This is more effective, so our 12-16 coverage is best covered by 12-14 opening 1♣ and rebidding 1NT, while a 1NT open is 15/16.
Ranges higher than a 2NT rebid?
We found above that a 2NT rebid should be 17/18. If you can have a 3NT rebid (see below, but assume you can find a sensible method for this) how strong should it be? As responder will pass with 5 hcp, opener must be a maximum 19 when opening 1♣ to avoid missing the 25 point game. As responder can be 6, opener rebidding 3NT must be a minimum of 19 to have that 25. So 1♣ then 3NT = 19 exactly.
This makes the 2NT open = 20/21. This is NOT guaranteed safe, as responder may have nothing. But a few points may enable it to make, and the chances are partner has something, even your normal partner.
Above 21 there are various methods. The Benjamin style gives two different ways of rebidding 2NT after a 2-open, and Kokish after a 2♣ open can achieve this. These can show 22/23, and 24/25.
Summary of all bids
1♣ then 1NT = 12-14. Responder needs 11/12 to invite, but fits can be explored at the 1-level.
1NT = 15-16. No invitation needed.
1♣ then 2NT = 17/18
1♣ then 3NT * = 19
2NT = 20/21
Higher ranges beneath 3NT split into 2-point ranges according to method chosen
* Jumping to 3NT is not usually a good idea, as it misses the oportunity to find fits when responder has additional length in one suit, or is 2-suited, so your methods will normally have some forcing bid, followed by a possible 3NT rebid.
Compare this NT ladder with standard bidding (ie in the rest of the world, not Acol)
Very close, perhaps not surprisingly.
Standard has 1NT open = 15-17, and 1NT rebid = 12-14. Now 1♣ followed by 2NT would be 18-19, and no special forcing method is therefore needed to show 19.
Simple and practical.
Compare this NT ladder with classical Acol (1NT open = 12-14 and rebid is 15/16)
The openings of 1NT, and 1♣ then rebid 1NT, are reversed. That's a matter of choice, Acol's choice now being poor, as opening 1♣ on weaker hands gives much better fit finding as shown above.
Compare this with alternative Acol (1NT rebid is a wide-ranging 12-16)
That extended range can push you into a failing 2NT if opener is minimum and responder is minimum for the invitation, or can lead to missed games, as explained above.
Poor, but compared with classical Acol this has the advantage of being able to open 1♣ rather than 1NT.
Handling the 19 count
This can be awkward. Standard 1NT = 15-17 avoids the problem. The normal way is to make a forcing rebid then follow with 3NT. Some people play 2NT = 19/20, but this is ineffective as responder needs to be stronger to be able to look for a fit. It is bad when responder is very weak. Also with a 19 count you are not going to miss game by opening at the 1-level if responder is bidding on 6 hcp.
Alternatively, if you are playing 5 card majors, transfer walsh enables you to have two ways of rebidding 2NT which therefore can show 17/18 and 19/20. This is the best option as it shifts the remainder of the NT ladder up, so a 2NT open = 21/22, while 2♣ then 2NT is 23/24, and 2♣ then 3NT is 25+, keeping 2-point ranges all the way and avoiding the restrictions imposed by Kokish methods.
Of course in the right range you open 1NT on 2335, 2344 and 3334 shapes, but what about 2245 or singletons? If these are "forbidden" by style then you will probably open 1 something and rebid 1NT if responder plugs one of your gaps. This means you suffer the same problems of the wide ranging 1NT rebid : ergo you may decide one or both of these should be opened 1NT, or maybe only use a 1NT open for a 2245 without a 4 card major.
Ray Green (fromagegb at gmail.com) revised Feb 2016