Main     Article 7A45 by Richard Pavlicek    

Broken Scoring Fix

International Matchpoint (IMP) scoring is widely accepted, and properly so, as the best form of scoring for serious bridge. I won’t go into its many virtues, but will instead bring out its worst flaw and propose a change to fix it.

On the following hands North-South bid accurately to the best contract. After a routine Stayman inquiry revealed the wrong major, North felt that his hand was too strong to invite slam with 4 NT. Therefore, he bid 5 NT (forcing) to allow for the possibility of 6 C if South had a second four-card suit in clubs. And so it came to be.

6 C North S 4 3
H A Q 10 4
D K Q 5
C A Q 10 4
None Vul


All Pass


2 C
5 NT



1 NT
2 S
6 C


S A J 10 5
H K 6 5
D A 2
C K J 9 7

Six clubs is not a claimer but pretty close to it. Even a 5-0 trump break can be survived if three diamonds and two hearts cash (give up a spade and crossruff the remainder). Call it 99 percent. Well bid!

At the other table North was less diligent, jumping directly to 6 NT after Stayman. Six notrump is certainly a good contract, basically needing the H J to drop or at least one spade honor onside — about 91 percent — but clearly inferior to 6 C.

Suppose the above scenario occurred 100 times. On average, 6 C would make 99 times, and 6 NT would make 91 times. At total points 6 C nets 99 × 920 - 50 = 91,030 points, while 6 NT nets 91 × 990 - 9 × 50 = 89,640 points. Consequently, bidding the superior slam in clubs gains 1390 total points, or an average of about 14 points per board. This is the way it should be.

Now consider the picture at IMPs. On the 91 occasions when both slams make, 6 NT gains 91 × 2 = 182 IMPs. On the eight occasions when only 6 C makes, 6 NT loses 8 × 14 = 112 IMPs. Thus, we have a complete turnaround! Bidding the inferior notrump slam now gains 70 IMPs, or an average of 0.70 IMPs per board. This is obviously wrong.

The primary objective in bidding is to reach the best contract, yet doing so here gets kicked in the face by the current scoring method. The problem lies in the assessment of 2 IMPs for high scores that differ by little. With low scores, say 180 versus 120, the 2-IMP differential is on the mark; but to gain 2 IMPs for 990 versus 920 is disproportionate.

To fix this, I propose a change:

If both tables make an undoubled slam, a difference of 20-80 is 1 IMP, and 90-100 is 2 IMPs.

In this case, making 6 NT (990) versus 6 C (920) would gain only 1 IMP, which neatly corrects the scoring flaw: In 100 deals 6 NT gains 91 while losing 112, properly favoring the club slam. Note that if both slams make an overtrick (1020 vs. 940) it is still 1 IMP, but if only 6 NT makes an overtrick it is 2 IMPs.

The adjustment works equally well when vulnerable, and/or for major- versus minor-suit slams. For instance, 6 H (1430) versus 6 D (1370) gains only 1 IMP, while 1460 (overtrick) versus 1370 gains 2 IMPs. It also works for grand slams (e.g., 2210 vs. 2140) but in that case it is always 1 IMP since overtricks do not exist.

This simple change will restore the traditional strategy to reach the best slam, as opposed to the current winning strategy to steal the 2 IMPs. Those who prefer the latter can always play matchpoints — though “matchpoint bridge” is arguably an oxymoron.

Is anyone listening?


© 2015 Richard Pavlicek