[BillPatch]

Auction S E W N
Pass Pass 3♦ Pass
Pass ??

East Hand
AKQ96
9762
97
JT

My simulation will recommend all 3 choices as vulnerability varies.
(IMPS--modified 1300edt 5/30)

[BillPatch]

Both vulnerable (the original problem) was the only one of the set where defending the contract by
passing was indicated. Pass was leading 3♠ by .32 imps/bd, pass/double by .74 imps/bd, 3♠/double
by .28 imps/bd. Sample size 54. A minimum sample size of 200/treatment(pass, double, 3♠) would be required to reach the .95 level of
significance that pass was better than double. (Tech Stat Theory Two-tailed ANOVA experimental design)

Neither vulnerable Double was best. It beat pass by .96 imps/bd. Double/3♠ by .56 imps/bd.
3♠/pass by .76 imps/bd. Doubling the sample size from 50 to a minimum of 100 could determine at
the .95 level of significance that double was better than pass.

With only the defensive bidders vulnerable Double was best. It beat pass by 1.56 imps/bd.
Double/3♠ by .98 imps/bd, pass led 3♠ by .04 imps/bd.

As reported on my other reopening poll post, 3♠ was best 2.9 imps/bd better than passing.

[PhilKing]

I might be being a bit thick, but how does a simulation prove anything here?

[mike777]

PhilKing, on 2014-June-01, 16:23, said:

I might be being a bit thick, but how does a simulation prove anything here?

great question, I always assumed simulations prove nothing but only hint at a direction?

granted bad models=bad directions

[Zelandakh]

PhilKing, on 2014-June-01, 16:23, said:

I might be being a bit thick, but how does a simulation prove anything here?

Indeed with a 95% confidence. Maybe the null hypothesis is not what we think it should be though.