[awm]

We all know that high card points don't completely capture playing strength. There are two main ways to deal with this, as well as some "in-between" strategies:

(1) Upgrade hands with positive features, rarely to never downgrade hands.
(2) Downgrade hands with negative features, rarely to never upgrade hands.

This makes a difference of almost a full point in your 1NT openings. Supposing a 14-16 notrump, someone playing approach (1) might open as many as a third of balanced 13s with 1NT, while upgrading as many as a third of balanced 16s out of the range. Someone playing approach (2) will virtually never have a 13-count and might downgrade as many as a third of 17-counts into the range.

My experience has been that most good players are a lot closer to approach (1) than approach (2). This somewhat invalidates Han's simulation, since he is comparing this 18-count against all balanced 16-counts, many of which we'd be upgrading out of the 14-16 range in any case. It also might be a poor match for Nigel's simulation, since his assumptions about which hands invite or bid game are probably wrong opposite an opener who upgrades good hands (for example, inviting with a normal 6 opposite 17-19 is probably wrong if opener would upgrade "good" 19s; inviting with a normal 9 opposite 14-16 is probably wrong if opener would upgrade "good" 16s, etc). This is somewhat reflected in Josh's post and 655321's post.

This hand is a "very bad 18" and is probably comparable to a "slightly below average 17." Since I play roughly style (1) and almost never downgrade, I'd show a minimum 17-19 on this hand. Certainly there exist some 16-counts which are arguably better than this hand, but I'm not opening a 14-16 notrump on those 16-counts anyway.

I guess the conclusion is "if you virtually never downgrade anything (and presumably upgrade pretty frequently), then downgrade this hand by one point" but "if you fairly frequently downgrade by a point when you don't like the hand (and presumably rarely upgrade), then downgrade this hand by two points." Hopefully partner knows what to expect.

[hanp]

My simulation is being invalidated, help!

[Jlall]

FWIW my partner and I do not upgrade very much, especially relative to most experts. I would say I pretty much never downgrade though.

[MickyB]

awm, on Jan 9 2010, 11:46 PM, said:

This somewhat invalidates Han's simulation, since he is comparing this 18-count against all balanced 16-counts, many of which we'd be upgrading out of the 14-16 range in any case.

Han's simulation shows it to be pretty much an average 16-count. Surely this wouldn't fit into the top-third for upgrading purposes?

It's somewhat system dependent too - I suspect that, of the pairs not playing strong club, 14-16ers upgrade much less frequently than 15-17ers, because they don't want to have to take another action in competition (or, if applicable, jump to 2NT) on a "good 16 count".

[awm]

The simulation results seem to indicate that:

(1) This hand is better than the average 16-count, but worse than the average 17-count.
(2) This hand makes game good opposite 9 hcp, so-so opposite 8 hcp, and bad opposite 6-7 hcp.

So if your style is that virtually all 16-counts open 1NT, and partner normally invites with all 9s opposite, then you should open 1NT with this hand.

However, if your style is that a lot of the "good" 16-counts upgrade out of the 1NT range, and partner passes with most 9s opposite a 1NT opening (and similarly passes most 6s opposite 17-19) then you should upgrade this hand to 17-19.

The latter is a lot more my style.

[rogerclee]

awm, on Jan 9 2010, 06:08 PM, said:

However, if your style is that a lot of the "good" 16-counts upgrade out of the 1NT range, and partner passes with most 9s opposite a 1NT opening (and similarly passes most 6s opposite 17-19) then you should upgrade this hand to 17-19.

MickyB said:

Han's simulation shows it to be pretty much an average 16-count.

[helene_t]

Sorry but I don't think it's very interesting to hear people saying that they don't open this hand 1NT because their notrump range is not really 14-16 but rather some shade of 13+ - 16- or whatever. Yeah with some people I play 12-14 so I obviously wouldn't open 1NT and with others I play 15-17 so I obviously would blah blah blah.

Han's simulation show that this hand is an average 16-count. Whether is makes game opposite an average 9-count we don't know as it depends how big declarer's advantage relative to DD is. But assuming that declarer's advantage is roughly the same for this hand as it is for an average 16-count, Han's simulation give us valuable insight nonetheless.

[hanp]

I imagine you all know the joke about the biologist, the physicist and the mathematician who see a bunch of white sheep.

I think it goes too far to say that my simulation shows that this hand is pretty much an average 16-count (as Mickyb and Helene write) or an above average 16-count/ slightly below average 17-count (as awm has mysteriously written).

Keep in mind that I only simulated the expected number of tricks opposite a balanced 9-count. To properly evaluate the hand, one would need to check how well the hand performs opposite all kinds of hands, including, for example, 3-5-4-1 hands, where I expect this hand to play extremely well.

Also, this simulation is still double dummy, and 500 hands may be too few.

[gnasher]

If I were doing a simulation like Han's, I'd compare this hand only to 4432 shapes, so that we could say "This hand is as good as a 4432 x-count."

In general, a 4432 shape has more playing strength than a 4333, and less playing strength than a 5332. A 4432 shape is also the most common of these three shapes, and is probably what most of us would regard as typical for a balanced hand. It's rare to upgrade or downgrade a 4432 shape, but some 5332s are upgraded and some 4333s are downgraded.

I think that if you agree a range of 14-16 but also agree with using judgement, it means that 1NT covers the range of playing strengths from a 4432 14-count to a 4432 16-count. That means that 1NT includes some (but not all) 5332 16-counts, and some 4333 17-counts.

[hanp]

If you were doing the simulation you might get 4,279 for the 16-counts and 4,502 for the 17-counts Andy.

[pooltuna]

hanp, on Jan 11 2010, 05:35 AM, said:

If you were doing the simulation you might get 4,279 for the 16-counts and 4,502 for the 17-counts Andy.

Is there something unique to the simulation where you get less 16 point hands than 17 point hands which a priori strikes me as being against the odds?

[helene_t]

pooltuna, on Jan 11 2010, 02:47 PM, said:

hanp, on Jan 11 2010, 05:35 AM, said:

If you were doing the simulation you might get 4,279 for the 16-counts and 4,502 for the 17-counts Andy.

Is there something unique to the simulation where you get less 16 point hands than 17 point hands which a priori strikes me as being against the odds?

Han wanted to compare 16-counts to 17-counts so he selected 500 of each. The probabilities of getting 16 rsp. 17 are irrelevant.