Which conventions to learn first

The question

BidMaestro starts every new player with the Standard group only: five-card majors, a 15–17 1NT, Stayman, Jacoby transfers, weak twos, takeout doubles and the rest of that group. Everything else is optional. A player who wants to add conventions has to choose where to start. This study measures which conventions make the most difference to the results, so that the ones worth most can be learned first.

How it was measured

The study used 150,000 deals. Each deal was played as a team match at two tables. At one table a partnership playing one set of conventions sat North–South against a partnership playing another set; at the other table the two sets swapped seats. Each table’s contract was scored with all four hands visible (Double Dummy), and the difference between the two tables was converted to IMPs, the usual team-match scoring units. A convention’s value is the IMPs per 100 deals that a set of conventions gains by including it.

The ranking was built one step at a time, in two directions:

• Adding. Starting with the Standard group alone, each optional convention was tried in turn and the one that gained most was added. This was repeated until every convention was in.
• Removing. Starting with every convention on, each was tried switched off and the one whose loss mattered least was removed. This was repeated until one was left.

The two directions gave exactly the same order, in both studies. That is a good sign that the order does not depend on where you start.

Each convention’s value depends on what is already played. So the tables below give two figures: the value at the step where the convention was added (with the conventions above it already in), and its value when added on its own to the Standard group. The “±” figure is the margin of error. Where two neighbours in a table differ by less than their margins of error, treat them as tied.

The figures are averaged over all deals, including the many where a convention never comes up. A gain of 4 IMPs per 100 deals is about one IMP in a 24-board session.

All bidding, including slams

† Not reliably measured yet; see What the study cannot show.

Bidding to game only

Many players want to bid their games well before they think about slams. For this second study the five slam conventions (RKC Blackwood, Control-Showing Cue-Bids, Splinter Bids, Serious 3NT and Gerber) were left off at both tables. Any contract above game that was not doubled was scored as the game in the same suit by the same player, so bidding a slam or missing one made no difference to the score.

† Not reliably measured yet; see What the study cannot show.

What the results suggest

• RKC Blackwood comes first, by a wide margin. Without it the app still bids slams on its point count, but it cannot first ask partner how many keycards they hold. No other convention comes close.
• For bidding to game, learn the doubles first. Negative doubles and support doubles head the game-only list, followed by New Minor Forcing and 2/1 Game Force. They are among the first five in the full list too.
• 2/1 Game Force is worth more for slams than for games. Its value falls from +2.69 to +1.14 when slams stop counting. Knowing early that the hand will reach game leaves room to look for slam.
• Some conventions only pay alongside another. Control-Showing Cue-Bids lose on their own (−0.59) but gain +1.90 once RKC Blackwood is played: they are how a partnership decides whether to ask. Splinter Bids behave the same way. Bergen Raises lose on their own (−0.99) and gain a little (+0.25) once the conventions above them are in place, so learn them later rather than early.
• Some conventions cover the same ground. Jacoby 2NT is worth +0.58 on its own but only +0.11 when added after 2/1 Game Force and the slam conventions, which already handle most of its hands.
• Below about +0.3 the order is not meaningful. The conventions there each add a little, and their margins of error overlap. Learn them for the hands they handle, in any order.
• Gerber almost never comes up. The app bids slams in notrump on the points the two hands hold, so it rarely asks for aces over partner’s 1NT or 2NT. Gerber changed no contract in this study, so it could not be ranked. Serious 3NT, Minimax and Inverted Minors showed no difference beyond their margins of error.

What the study cannot show

Five conventions are not reliably measured yet: Multi 2♦, Cappelletti / Hamilton, Michaels Cue-Bid, Unusual 2NT and Inverted Minors. In the study, each partnership read its opponents’ bids by its own convention card. So when only one side played a convention, the other side read that convention’s bids as if they had their natural meaning. In a real game the bids are alerted and the opponents are told what they mean. These five are the conventions most affected, so their figures (including Multi 2♦’s loss) will be measured again once the study can give each side the other side’s card.

Being told what a bid means is not the same as knowing how to bid against it. Against opponents who have not met Multi 2♦, Cappelletti or Inverted Minors before, these conventions may be worth more than any study between two well-prepared partnerships can show.

The study also has some limits that apply to every convention:

• Both tables were bid by BidMaestro. The values are for BidMaestro’s versions of the conventions, played without mistakes. A convention that is hard to remember is worth less at the table if it is sometimes forgotten.
• Contracts were scored with all four hands visible (Double Dummy), on the actual cards. Real declarers and defenders do not see all the cards, so some contracts would score differently in play.
• The deals were dealt at random. In real play, some conventions come up more or less often depending on how the opponents bid.
• The study ranks conventions by results only. It does not measure how much effort each one takes to learn. A convention that gains a little but is easy to learn may still be worth learning early.

← Back to research