[jdonn]

I don't like the idea of a database for several reasons:
- How do you determine what is a "similar" hand? Even slight changes to this actual hand, such as changing the 10 of hearts into a low spot, would potentially have a significant impact on the bids people choose.
- It would take a massive amount of data. You would not only have to find lots of "similar" hands (to reiterate, I don't think it's feasible to come up with a satisfactory definition of what that means). You would need it to be in 4th seat, at this vul, and have the auction go 3!S P P.
- I don't think objections Frances stated are invalid. They may be invalid to the theory of a database, but not to actually implementing one. How would someone determine such things as whether or not each pair is playing Leaping Michaels?

It sounds to me like an idea that makes some sense in theory, but is completely impossible to implement in a fashion that is useful.

[Fluffy]

Weighting choices by the "distance" from actual hand and the one compared would solve many of the issues I think.

[Zelandakh]

Fluffy, on 2015-July-14, 06:34, said:

Weighting choices by the "distance" from actual hand and the one compared would solve many of the issues I think.

Or just define the boundaries between calls according to criteria provided by experts. It is arguably easier to define a system by these boundaries than the meanings of the calls themselves and their definition then makes it easy to see which hands fall into which partitions.