BBO Discussion Forums: How low do you go? - BBO Discussion Forums

I found these two forum discussions convincing.

"I don't feel strongly between 1♦ and 1♣, but 1NT is a gross misevaluation of this hand."

"Definitely 1NT for me, not even close with 4333."

mike777, on Jun 11 2008, 09:39 PM, said:

edited my post. Meant definitely not 1NT.

1♦

Since I seem to be alone in believing that 1NT is the right opening bid and opening 1D and rebidding notrump a clear underbid, I did a simulation (1000 deals).
I specified for partner a balanced hand with exactly 10 HCP. This should be a fair test.
If you open 1NT you will play 3NT while if you open 1D and rebid 1NT you are likely to miss game.
I also specified that partner will have at most 4 cards in a major.

Result:

Double dummy 3NT would make in 592 cases and fail in 408 cases. The average number of tricks available for declarer in a notrump contract was 8.7 tricks
Since double dummy analysis clearly favors the defense the result at the table would be even more in favor of 3NT making.

rhm, on Jun 14 2008, 11:42 AM, said:

Thank you rhm Chastening but interesting that most of us can be so wrong

rhm, on Jun 14 2008, 11:42 AM, said:

Intuitively that conjecture seem reasonable. Any experimental data to back it?

rhm, on Jun 14 2008, 05:42 PM, said:

Being in agreement only with NickRW and me is, I suppose, much the same as being alone.

Quote

I'm surprised that you got such a high success rate. Did your balanced 10-counts include 5332 shapes? If so, presumably some of them would be upgraded to 11-counts.

Quote

Is this clear? Obviously, the defence don't always make the right lead or find the best defence thereafter, but there will also be occasions where declarer has a two-way finesse, or has to guess which suit to play first, or can make only by playing against the odds.

gnasher, on Jun 14 2008, 12:49 PM, said:

My simulation allowed for a 5 card minor (but not a 5 card major). Double dummy analysis at least does not support the idea of adding a point for a 5 card suit and it also does not see 4333 much of a disadvantage at a notrump contract compared to other distributions.
Anyway partner's hand opposite this hand consisted most of the time of kings and queens, which are slightly overrated at notrump and it would not occur to me to upgrade such hands from 10 to 11 points.

Of course double dummy analysis works both way. With regard to double dummy analysis favoring the defense this depends a bit on the contract and level.

For an analysis you can look at

http://crystalwebsit...my_accurate.htm

The general conclusion from this website is that double dummy analysis is quite close to average results at table, but in 3NT declarer tends to make about 0.19 tricks more at the table than according to double dummy analysis.

rhm, on Jun 14 2008, 11:42 AM, said:

Did you consider getting too high if partner has 9 or 8? Most of us believe very strongly in being quite aggressive opposite a 1NT opener regarding inviting and bidding game.

Also it's quite wrong for your simulation to include 5 cards in the minor we open, partner would certainly invite opposite a minor suit opening with 5 card support and a 10 count.

That double dummy analysis suggests upgrading this hand does not surprise me - Thomas Andrew's work suggested that shape was almost irrelevant in determining whether 3NT would make, in fact, 4333s performed slightly better than other shapes.

What I haven't worked out is why the DD results conflict with most players' beliefs. Two possible reasons -

We remember the times that a long suit gave us nine tricks, and attribute this to having a long suit. We don't remember the times that having a doubleton rather than a trebleton allowed the opponents to cash one more trick in the suit, or gave us one fewer stopper.

Playing double dummy is of greater benefit when our hands are flatter. With a five-card suit, it is usually right to play on it; With no long suit, we might have to guess which of our suits break 3-3, or which finesse to take.

jdonn, on Jun 14 2008, 02:15 PM, said:

Of course I considered that.

But your argument is like staying in bed, then your chance of being overrun by a bus is minimized.
If you underbid of course your chance of going down is less likely.

My point was that this hand is worth 1NT, but should partner invite I will decline and may go down if it turns out that 7 tricks are the limit.
This can happen to any minimum notrump opening bid and the under bidders will have a field day. I just ran a simulation and changed partner's hand to 8 or 9 HCP and 2NT made in 704 out of 1000 cases. (Average number of tricks per deal were 8; 311 times 3NT would still have come home)

With regard to 5 card suits:
Large double dummy results do not support the notion that you should add a point for a five card suit
If you have a fit in a suit, then opponents frequently will have a good fit too and opponents have the opening lead. They are ahead in the race of developing a suit in a notrump contract.
I would certainly not add anything to a hand with few aces and few intermediates where my 5 card suit is only so so.

rhm, on Jun 14 2008, 02:51 PM, said:

WTH I didn't even make an argument! It went like this.

You: If partner has 10 you will make 3NT 6 out of 10 times, therefore you should open 1NT.

Me: The first part of what you said doesn't support the conclusion you drew, since there are other factors. For example, you didn't consider how well or badly 1NT will work when partner has amounts other than 10!

You: Something I don't understand, then admitting that if partner has 8 or 9 you will go minus 30% of the time and gain nothing the other 70%, and chalking it up to essentially 'sometimes doing what is wrong works' instead of actually factoring it into your conclusion.

This is the discussion? You claim essentially that 10 should never invite opposite a 1NT rebid, even with a fit for partner. You completely ignore that it's often right to bid 3NT with 9, yes with 9! But heck let's ignore all that and just take what you have claimed so far. We will only consider if partner has 8 9 or 10. Let's say he has 8 33% of the time, 9 35%, and 10 32% of the time that he has one of those amounts (feel free to calculate exactly but I bet that's about true when you hold 14 since just under 9 is his average, actually a bit lower since he didn't open). Let's take your assumption that 10 will bid 3NT and 8 or 9 will invite. Say if you open 1♦ you get to 1NT.

You lose .3*.68 + .4*.32 = about 42% of the time. You gain .6*.32 = about 19%. So by your own simulations opening 1NT is horrible. But instead of acknowledging that, you say about the losses that sometimes "the under bidders will have a field day" and you use the gains to claim that "1NT is the right opening bid".

If you argue that there are tons of other factors, such as partner might have less than 8, or 1NT has preemptive effect, you are right! Which is why you can't draw any conclusions based on your first simulation alone.

And people wonder why the forum regulars and statisticians are cranky lately...

The original question was, in essence, whether this hand is equivalent to a balanced 15-count. I think that the right way to test this by simulation is this:

For several values of responder's strength, measure how often 1NT, 2NT and 3NT make (1) when opener has a 4432 or 4333 15-count, (2) when opener has this hand, and (3) when opener has a 4432 or 4333 14-count. Comparing the results of these would tell us whether this is as good as a 15-count, and if not how close it is.

I realise, of course, that a suggestion of this sort is easier to make when you're not planning to do any of the work.

Edit: Whether to include 4432 shapes depends on whether our question is "Is this as good as a balanced 15-count?" or "Is this as good as a 4333 15-count?"

This post has been edited by gnasher: 2008-June-14, 15:08

jdonn, on Jun 14 2008, 08:19 PM, said:

Well, I am one of the (would be) 1NT bidders on this hand (not that I actually play 15-17). And I would be one of those who would take a direct punt at 3N opposite such a bid with 9.

So I ran my own sim. Partner has exactly 9hcp with a 4333 and at least one ten and one nine or two tens - or - 4432 with one ten. This is min I personally would punt 3N with opposite a 15-17 in the abscence of some extra shape. 3N made on 11 cases of 20.

And no I am not going to post any hands and nor am I going to do a bigger sample. This forum is way to ungrateful about sim results and I have done a great many similar sims in the past so am bored with this question. If you don't want to take such results into account - fine by me.

Nick

I suspect that if you give opener a random balanced 14, you will do okay in 3NT opposite a random balanced 10. Does this mean we should upgrade random balanced 14s to 15?

For any three-point range there will be hands where opener has the best possible 1NT bid and partner has the best possible pass, and game is okay (usually not great, but above the threshold to bid it at IMPs). The reason things are this way is that the bottom of the range is substantially more frequent and opposite this even 2NT is often too high, and also that you expect opener to accept an invite with "above average" hands and not just true maximums. So even though balanced 10 opposite good 14 makes game often, you don't want to invite opposite 12-14 with a balanced 10 because you will also get to the lousy contract of 2NT in the frequent event that partner has only 12 or a bad 13, and you will reach a bunch of bad 3NTs opposite opener's "good 13" or "bad 14" hand types.

I don't think this really calls for "upgrading" more often -- if anything it suggests playing two point notrump ranges instead of three but it's generally hard to manage that without a lot of other systemic issues.

Another point is that no one's bidding methods are good enough to bid every single game that is expected positive imps while avoiding every single game that is expected negative imps. I wouldn't be too bothered by missing the occasional 50% game, or by bidding the occasional 25% game -- this kind of stuff happens. Now if game was making 90% of the time opposite 10 hcp I would feel dumb for not opening 1NT, but these numbers like 55% and 60% are not (by themselves) very convincing to me.

Suppose you were asked what you would open but you were not allowed to count your points...

I suspect I would think "this hand sure looks like a strong notrump to me - I will open 1NT!". Wouldn't you think that too?

Don't you think this hand rates to be worth more tricks in the long run than a substantial % of 4333 15-counts that you open 1NT without thinking twice (but probably after having counted your points)?

If you have confidence in your own judgment you should be willing to follow it when it tells you that something is quacking like a duck. You should not have to count exactly how many feathers the duck-candidate has - you can see it is a duck regardless of its feather-count.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

fred, on Jun 14 2008, 06:24 PM, said:

Impossible to say for sure since counting points on balanced hands is pretty much a subconscious thing, but no, I doubt I would.

rhm assumes the truth of Victor Mollo's famous pronouncement "Offered the opportunity to play with open cards (rather than at normal singe-dummy), declarer should reject but defenders accept"

I guess that, as you increase the high card strength of the declaring partnership, there is a steady decrease in the defenders' advantage at double dummy -- as compared with single-dummy.

For example, in a slam contract that is dodgy at single-dummy, you would usually back declarer rather than the defence at double-dummy.

Assuming the correctness of the latter conjecture, what is the HCP difference break-even point?

The results of a simulation (published on RGB?) confirmed another plausible theory: "You need fewer HCP to make a contract when the points are split between the 2 partnership hands."

To take an extreme illustration, I vaguely remember that you can expect to make 3N with 12 opposite 12, but not with 24 opposite 0.

These observations may have some bearing on the current argument.

(Don't take this too seriously)

Today (Round 6, Board 2 of European Championships) Italy's Duboin held:



Presumably having not read this thread, he failed to downgrade his hand and opened a 15-17 1NT.

Hopefully your reaction to that was "how silly - this is an obvious 15-17 1NT".

If so consider: do you think this 15-count is better or worse than the 14-count that this thread is about?

If you answered, as I suspect most people would, "about the same" then I submit that either you should open the 14-count 1NT or Duboin should have opened 1D with his 15-count.

And given how silly that sounds, hopefully you are now convinced that 1NT with the 14-count is right

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

Fred,

In my experience, AJx is a poor holding relative to its high card count. Do you agree with this? If so, do the tens and nines in this hand really not only make up for that, but push the hand's total playing value into the 15-17 range? Is this the sort of hand you would only open 1NT if playing 14+-17? Would you contend that most people (mostly referring to non experts) that play '15-17' are really playing 14-17 value wise?