Counting points for a 1NT or 2NT opening

The question

BidMaestro opens 1NT with a balanced 15 to 17 high-card points and 2NT with 20 or 21. A hand of 14 high-card points with a good five-card suit is often said to be worth 1NT. A hand such as ♠KJ2 ♥QJ ♦KJ853 ♣QJ4 has 14 high-card points and a five-card suit, but its queen and jack of hearts are a doubleton and may take no trick. Should the count go up for the long suit, down for the short honours, or neither?

What the sources say

Audrey Grant’s Modern Competitive Bidding recommends upgrading: in her view, almost any reason to move a hand up into the notrump range is worth taking. She counts ♠A95 ♥Q10 ♦K96 ♣KQ1082 as 14 high-card points plus 1 for the five-card suit, and opens 1NT. She also treats tens and nines as reasons to upgrade, and counts 19 high-card points with a five-card suit as a 2NT opening.

Bridge World Standard counts high-card points only: 15 to 17 for 1NT. The ACBL Standard American Yellow Card and the WBF Standard Card give the ranges and say nothing about upgrades.

Dubois’ Bidding Notes describes 1NT both as “15 to 17 points” (points that may include length) and as 15 to 17 high-card points, and warns that opening 1NT on 14 erodes partner’s confidence and should be done only rarely. Timm’s The Two-Over-One Game Force System adds a point for a five-card suit with an honour, within a full count that also takes points away. Kokish and Kraft let long suits count towards their 2NT range, and advise being conservative.

Two sources agree on the other half of the question. Dubois and Timm both take a point away for a doubleton queen or jack. BidMaestro tested a stricter version: a queen or jack in a doubleton counts nothing, and a jack in a three-card suit counts nothing. Ax, Kx and AK keep their full value.

The deals and the candidates

BidMaestro bid 150,000 random deals with each candidate rule. Only the decision to open 1NT or 2NT changed; everything else stayed the same. The first 50,000 deals were used to compare three replacement counts with the present one:

high-card points plus 1 for each card over four in a suit;
high-card points with short queens and jacks counted as nothing;
both changes together.

How the auctions were scored

Each deal whose final contract changed was scored by Monte Carlo Double Dummy analysis. The cards of the side whose bidding changed were kept, the other two hands were re-dealt 200 times, each layout was played Double Dummy (all four hands visible), and the contract was scored against the average. Results are given in points per changed deal, with a margin of error; a difference smaller than about twice its margin of error can be chance.

What we found

All three replacement counts lost.: Adding a length point: about 23 points per changed deal worse (1,872 deals). Counting short queens and jacks as nothing: about 24 worse (3,146 deals). Both together: about 23 worse (3,190 deals). Each loss is five to eight times its margin of error.
The losses came from hands that no longer opened 1NT.: With the length point, a 17-count with a five-card suit counted 18 and opened one of a suit instead: about 119 points worse on those deals. With short honours counted as nothing, 15 to 17 counts fell below the range and opened a suit: about 40 points worse.
Hands lifted into the range gained little.: So a second test let a hand count up into the range, but never moved a hand already in range out of it. Because this rule was chosen after looking at the first results, it was tested on the other 100,000 deals only. Upgrading with the length point: about 3 points per changed deal better, within its margin of error of 3 (2,483 deals). Upgrading with the length point but counting short queens and jacks as nothing: about 9 better, with a margin of error of 5 (962 deals).
A light 1NT that is passed out loses.: When the upgraded 1NT was passed out, the upgrade lost about 34 points per deal (216 deals); a passed-out upgraded 2NT lost about 102 (77 deals). When partner bid on, the upgrade gained: about 8 per deal after 1NT and about 52 after 2NT. The two roughly cancel. With every optional convention switched on, the upgrade results were the same within their margins of error (about 4 and 7 points better).

What we decided

BidMaestro keeps counting high-card points only for the 1NT and 2NT openings: 15 to 17 for 1NT, 20 to 21 for 2NT, with no point added for length and none taken away for short honours. No alternative was clearly better, and the simplest count is the easiest to learn. The example hand at the top, with 14 high-card points, opens 1♦.

Limits of the study

The study measures each count together with BidMaestro’s own later bidding, with all four players using the same system. Double Dummy play is perfect play, which tends to favour the declarer slightly. Tens and nines, which Grant also counts, were not tested.

← Back to research