Hand Evaluation: Conclusions, and Warnings
While some conclusions can be drawn from these experiments, it is important to take care about the nature of the conclusions.
The primary caveat in relation to this data is that it is based on double-dummy results, which do not always reflect real-world trick-taking value. You can see examples in Problems with using double-dummy data.
So, without further ado, lets look at some possible conclusions, and some caveats.
Card values
Normal high card points (HCP) evaluate an ace=4, king=3, queen=2, jack=1. The double-dummy data, however suggests that this undervalues aces and overvalues queens.
Alex Martelli has proposed what he calls "BUM-RAP:" ace=4.5, king=3, queen=1.5, jack=0.75, ten=0.25. This does appear to be more effective, at least for suit contracts, but you have to be careful. Like high cards, BUM-RAP gives a total of 40 for the whole deck, but you can't just switch BUM-RAP in as a replacement.
Why is that? Well, let's assume you are playing 2NT openings as 20-21 balanced. If you switch from 20-21 HCP to 20-21 BUM-RAP, does this mean 19.5 to 21.25? 19.75 to 21.50? 20.00 to 21.75? It turns that the average 20-21 HCP hand evaluates to 20.9 in BUM-RAP, and that the 20-21 HCP range is closer to 20 to 21.75 in BUM-RAP.
This adjustment then requires you to change the ranges of your 2NT rebids (18.00-19.75), your 1NT openings (15.00-17.75.) Essentially, you keep your old ranges, but always round down the BUM-RAP valuation.
The Card Values for Three Notrump article shows that when targetting three notrump specifically, standard Work Point Count does better than BUM-RAP, and does significantly better if we add a value of 1/2 for tens.
Cards in Combination
Based on my data and his own research, Alex Martelli has noted that cards in combination are worth slightless less than cards in seperate suits, and that cards in long suits are worth less than cards in shorter suits. This appears to be true.
It is a wild leap, though, to discard decades of expert practice regarding cards in combination.
Suppose you open 1 ♥ holding ♠ 6-4; ♥ A-Q-J-T-6; ♦ 8-7; ♣ K-J-9-5. Partner responds with 1 ♠, and you rebid 2 ♣. With this auction and this hand, partner is perfectly placed to evaluate his hand - he will value club and heart honors higher and spade and diamond honors lower. In other words, having cards in your long suits makes it more likely that partner will be able to evaluate his hand in combination with yours. Indeed, most of the time when experts recommend to value honors together, it appears the advantage is that partner has the ability to gauge the fit.
Another possible reason the data might seem to disagree with expert practice might be similar to that seen in my article on 4333 in notrump. In this case, it might be that, while on average you score more tricks in a suit with fewer values in your longer suit, it might be that the "real" cases when you score more tricks in a suit due to this effect are cases when you belong in notrump.