Can I use simulation to show that one bidding point count is more accurate than another?

Question

I've created a new bridge point count system with aces worth 4.5 points, kings worth 3.0 points, queens worth 1.5 points, jacks worth 0.75 points, and tens worth 0.25. There are still 40 points in the deck, in line with the Work 4-3-2-1 system.

Apparently, a compilation of hands played in tournaments was run, and the following tables were created decades ago for the Work point count. Assuming "double dummy play, they showed, among other things, that a partnership with 25 hcps and a reasonably balanced hand was a (slight) favorite to make 3NT.

Supposed I had the computer recalculate points under the new system and sort the percentages of successful and failed contracts at each point level under double dummy play. Such an exercise might show higher ratios of success for 3NT and higher contracts for 25+ points than the original simulation, and higher rates of failure for 24- point contracts.*

Is this an appropriate test? Would the results described above demonstrate the advantages of the new point count system?

*The simulation could, of course, show the reverse, which I would take to mean that the new system was inferior.

Answer 1

Absolutely!

I believe Deep Finesse (100% free according to the web site) is the double dummy analyzer used by the A.C.B.L. for modern tournament to determine par scores on hand records. That would likely be your best choice for play analysis.

Yes, a double dummy analyzer is, to the best of my knowledge and understanding, the best technology available for hand analysis. It is imperfect, but world-class players rarely miss a line of play that can be deduced from the available information.

However, before you jump in, you are not the first, by a long shot, in this pool. I would recommend researching the attempts already made along this line. One other such is:

Two thirds the total obtained from:

• A = 5;
• K = 4;
• Q = 3;
• J = 2;
• T = 1;

with fractions of 1/3 and 2/3 rounded down and up respectively.

Other factors to consider including:

• assess distribution in both NT (hand shape) and trump (fit & ruffing potential) contracts differently;

• adjusting for suit quality;

• traditional (uniform) approaches miss that Aces and Kings are overvalued in NT contracts and off-suit quacks and tens are overvalued in suit contracts.