Weak Notrump

Dealer North, NS non-Vul, EW Vul, IMP scoring.

N E S
1NT(12-14) (P) ?

South holds:

QJ
J6
KJ95
KJ752

Co Wiersma

2017-07-12 22:30:36 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

I guess 2NT would be about right
Though at imps, I would not criticize 3NT

Co Wiersma

t***@att.net

2017-07-12 23:53:20 UTC

3NT, hoping for scattered points between the hands and a running minor.

judyorcarl@verizon.net

2017-07-13 12:11:07 UTC

Post by t***@att.net
3NT, hoping for scattered points between the hands and a running minor.

That's what you do with Ax xx Axxx KJxxx.

As it is, partner would need 2 aces and two entries to this mess. Also heart stop, and partial spade stop.

Carl

Sandy Barnes

2017-07-13 01:11:43 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

At IMPS, it is a coin toss. I would shoot out 3NT, but it could be a loser. How will partner be able to judge his hand correctly now looking at this hand? This makes 2NT slightly problematic.

Player

2017-07-13 10:40:33 UTC

Thrunt.

Douglas Newlands

2017-07-13 11:06:22 UTC

Post by Player
Thrunt.

yep, it does say imps!

doug

judyorcarl@verizon.net

2017-07-13 12:07:45 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

CCCC is 10.6: Edgar Kaplan would pass in tempo.

Carl

Will in New Haven

2017-07-13 17:04:10 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

Not vul? I pass. I would bid 3N vul at IMP.

--
Will in New Haven

p***@infi.net

2017-07-13 19:38:46 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

At notrump, I rate this about 11.4 (no Aces, no tens, one nine, five card suit) so it's a maximum pass.

Player

2017-07-14 00:53:06 UTC

"At notrump, I rate this about 11.4 (no Aces, no tens, one nine, five card suit) so it's a maximum pass."

"CCCC is 10.6: Edgar Kaplan would pass in tempo."

Actually I think it is 11.3174 recurring HCP. Lol Victor Mollo would find a number of Walruses amongst you lot.

Lorne

2017-07-13 23:13:39 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

Looks pretty close to me, probably go for an invite (2N or whatever you
play). Might go for 3N if vul.

Fred.

2017-07-14 13:27:21 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

I saw my choice as being between pass and 2NT.
At match points I think I would have passed,
but here I bid 2NT.

The whole hand was

973
AKQ7
QT84
Q4

T84 AK652
9832 T54
A73 62
T96 A83

QJ
J6
KJ95
KJ752

Partner raised to 3NT. East started with the spade king
giving us -150. Unlucky, but just looking at our cards
I would expect the opponents to most often take at least 3
spades and 2 aces to set it.

I'm not saying Ron is wrong, but it's Walrus 1, Player 0,
Fred. 0 :-)

Fred.

KWSchneider

2017-07-15 15:30:38 UTC

3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

Travis Crump

2017-07-16 22:31:25 UTC

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

My simulation[error around .01-.02]:

3N makes 0.3816 out of 10000
3N makes with 12 0.2606848287928018 out of 4001
2N makes with 12 0.5168707823044238 out of 4001
3N makes with 13 0.36973569561428987 out of 3443
3N makes with 14 0.5868544600938967 out of 2556

Blasting 3N usually wins 6 when right and loses 5 when wrong so you'd
need about 45% to break even. Inviting loses 2 or 4 IMPs often enough
when partner declines that I think pass is right. You can sort of hand
wave bidding being right if you think you'll have a significant single
dummy edge[but then aren't you a favorite in the match anyway?], or your
decisions over an invite are much better than random. But at best it is
still going to be close to even for bidding or passing.

Travis

judyorcarl@verizon.net

2017-07-17 17:46:56 UTC

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

3N makes 0.3816 out of 10000
3N makes with 12 0.2606848287928018 out of 4001
2N makes with 12 0.5168707823044238 out of 4001
3N makes with 13 0.36973569561428987 out of 3443
3N makes with 14 0.5868544600938967 out of 2556
Blasting 3N usually wins 6 when right and loses 5 when wrong so you'd
need about 45% to break even. Inviting loses 2 or 4 IMPs often enough
when partner declines that I think pass is right. You can sort of hand
wave bidding being right if you think you'll have a significant single
dummy edge[but then aren't you a favorite in the match anyway?], or your
decisions over an invite are much better than random. But at best it is
still going to be close to even for bidding or passing.
Travis

That agrees with my presumption.

But.

With deals like these, at-the-table declarer play figures to be much closer to double dummy than the defense.

Often, the defenses inflexibly accurate timing.

So I think the att results would be more favorable than the simulation.

Carl

Fred.

2017-07-17 20:56:43 UTC

Post by ***@verizon.net

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

That agrees with my presumption.
But.
With deals like these, at-the-table declarer play figures to be much closer to double dummy than the defense.
Often, the defenses inflexibly accurate timing.
So I think the att results would be more favorable than the simulation.
Carl

The other side of it is that the more controls your side is missing
the less accurate the opposition timing has to be to beat you. Note
than on the actual hand that less than perfect defense would allow
us to get within two tricks of the contract.

Fred.

p***@infi.net

2017-07-18 16:57:17 UTC

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

Interesting. We know that 3NT is a contract declarers make more often at the table than double dummy. My usual assumption is to add 10% of DD failures to the success percentage; so blasting 3NT would make about 44% rather than the DD 38% odds. That's still a bit of a loss at the 6 to 5 IMP odds, about -.13 IMPs per deal. The 10% advantage is based on some data from Richard Pavlicek, though I am not at all sure his 10% and mine are quite the same.

Inviting with the given hand, with opener accepting on 13 or 14 and passing on 12, gains .005 IMPs per deal, too close to call.

Inviting, opener accepts with 14, gains .025 IMPs per deal. If you had to use a "range finder" 2S bid for that purpose, you would have todeduct for the possibility of a lead-directing double or negative inference. So I doubt it would be worth it.

So pass or invite are both close, while blasting loses.

But suppose Player makes 15% of DD fails: now blasting gains .203, invite/accept 13 .204, a toss-up but surely there is some loss in the slower sequence. Invite/accept 14 gains only .109, worse than either of the other approaches.

So everyone who answered this may be right, depending on what declarer's actual advantage is. Blasting breaks even slightly above a 12% edge for declarer, but invite is still better at that point. I do not believe, however, that invite/accept 14 ever becomes the optimal approach, so I have no desire to adopt the 2S "range-finder" gadget. The risk chasing a game only opposite a max does not justify forcing partner to at least 2NT.

judyorcarl@verizon.net

2017-07-18 19:41:04 UTC

Post by p***@infi.net

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

Interesting. We know that 3NT is a contract declarers make more often at the table than double dummy.

Yes, but the general situation usually has more aces than may be here. This time, you will often have to lose the lead twice or more to set up 9 winners. short-suit aces are platinum for 3NT.

Carl

p***@infi.net

2017-07-19 02:40:34 UTC

Post by ***@verizon.net

Post by p***@infi.net

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

Interesting. We know that 3NT is a contract declarers make more often at the table than double dummy.

Yes, but the general situation usually has more aces than may be here. This time, you will often have to lose the lead twice or more to set up 9 winners. short-suit aces are platinum for 3NT.
Carl

Well, double dummy has to lose the lead just the same. I don't know whether Aces are relatively more valuable at the table than double-dummy. If they are, I am astonished at how many players fail to downgrade aceless hands. Player seems to be one.

KWSchneider

2017-07-19 18:16:17 UTC

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

I decided to answer my own question with a simulation, running two (2) 1000 hand sims (1000 hands meeting the criteria) where:
1) Dummy has exactly QJ.J6.KJ95.KJ752
2) Opener has 12-14 HCP - no constraints were placed on tens.
3) Opener is balanced with no 5+cM

Note that the DD and SD results were run concurrently on the same hands.
Results:
Sim1 - DD = 41.7% (8.04 trick avg); SD = 57.0% (8.62)
Sim2 - DD = 40.6% (7.98 trick avg); SD = 57.1% (8.62)

I found it interesting that the SD results were remarkably consistant across the two sims. Since these take 10 hours to run, I intend to repeat the Sims 3 more times to yield 5000 results and hopefully report tomorrow night (EST).

And in view of the 57% SD outcome, I feel vindicated in my position that the result in the posting was more bad luck than poor decision making.

On the other hand, I'm concerned that the DD results were not closer to Travis' 38%. I'd be interested in seeing what the differences in simulation constraints were. Perhaps over 10,000 sims mine will average out to 38%, although they are not trending in that direction.

If anyone would like more detail, please let me know. I have a spreadsheet available for each sim, with the 1000 deals, each hand, and the DD/SD result.

Kurt

--
Posted by Mimo Usenet Browser v0.2.5
http://www.mimousenet.com/mimo/post

Travis Crump

2017-07-19 22:03:16 UTC

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

1) Dummy has exactly QJ.J6.KJ95.KJ752
2) Opener has 12-14 HCP - no constraints were placed on tens.
3) Opener is balanced with no 5+cM
Note that the DD and SD results were run concurrently on the same hands.
Sim1 - DD = 41.7% (8.04 trick avg); SD = 57.0% (8.62)
Sim2 - DD = 40.6% (7.98 trick avg); SD = 57.1% (8.62)
I found it interesting that the SD results were remarkably consistant across the two sims. Since these take 10 hours to run, I intend to repeat the Sims 3 more times to yield 5000 results and hopefully report tomorrow night (EST).
And in view of the 57% SD outcome, I feel vindicated in my position that the result in the posting was more bad luck than poor decision making.
On the other hand, I'm concerned that the DD results were not closer to Travis' 38%. I'd be interested in seeing what the differences in simulation constraints were. Perhaps over 10,000 sims mine will average out to 38%, although they are not trending in that direction.
If anyone would like more detail, please let me know. I have a spreadsheet available for each sim, with the 1000 deals, each hand, and the DD/SD result.
Kurt

I just used 12-14 HCP, and Deal's balanced function which is balanced
returns true if the hand is some 4333, 4432, or 5332 without a 5-card major.

p***@infi.net

2017-07-20 15:22:15 UTC

Post by KWSchneider
3N is the bid at IMPS. Simply bad luck that it doesn't make. A simulation probably supports this.

At first I misread "SD" as "Standard Deviation", but I assume you meant "Single Dummy." Wow --- that's a huge difference. By my methodology that means declarer makes 28% of contracts that should fail double-dummy. (Actually, declarer makes more "DD fails" than that, but goes down on some that would make double-dummy.) Any way you look at it, that's much higher than Richard Pavlicek's 10% edge suggests. And this is single-dummy for both declarer and defenders, correct? Perhaps you are using an unrealistic setting for the AI. It would be lovely to have a real-world comparison of, say, Meckwell's results on 1NT-3NT auctions vs. double-dummy expectations, but few of us mortals play that well. Also, of course, their strategy of wearing down the opposition in extended knockout matches by bidding thin games skews the results. It would be interesting to see their results early and late in a match.

You might compare a variety of base hands to see if the SD vs. DD gap is unusually high for this layout, but I'd be surprised if it were. Anyway, thanks for your efforts!

KWSchneider

2017-07-20 19:44:10 UTC

At first I misread "SD" as "Standard Deviation", but I assume you meant "Si=
ngle Dummy." Wow --- that's a huge difference. By my methodology that means=
declarer makes 28% of contracts that should fail double-dummy. (Actually, =
declarer makes more "DD fails" than that, but goes down on some that would =
make double-dummy.) Any way you look at it, that's much higher than Richard=
Pavlicek's 10% edge suggests. And this is single-dummy for both declarer a=
nd defenders, correct? Perhaps you are using an unrealistic setting for the=
AI. It would be lovely to have a real-world comparison of, say, Meckwell's=
results on 1NT-3NT auctions vs. double-dummy expectations, but few of us m=
ortals play that well. Also, of course, their strategy of wearing down the =
opposition in extended knockout matches by bidding thin games skews the res=
ults. It would be interesting to see their results early and late in a matc=
h.
You might compare a variety of base hands to see if the SD vs. DD gap is un=
usually high for this layout, but I'd be surprised if it were. Anyway, than=
ks for your efforts!

All players are playing SD in the SD Sims. Defense is much harder than declaring, so DD improves defense significantly, especially leads and switches in notrump, where declarer does not have a trump suit to help control the situation. DD helps declarer less so, finding all singleton K and doubleton Q, hence the overall shift to the declarer in SD.

In the Sims, some DD declarers make more tricks than SD, but much more often the converse. AFAIK Pavlicek pulled his 10% number out of the air - I have no idea how he supports it.

Now, SD has a couple of minor issues, which will impact easily less than 5% of the hands. These are the hands where 'expert' defense (considering the implications of 'non-bids' and detailed signal reading) will produce more defensive tricks than my SD results (although I have defense level ratchetted up higher than declarer level). This is offset by the SD defender's abilities to instantly recognize impending through-ins or squeeze plays many tricks in advance, and be pro-active in avoiding them. If I were to hazard a guess it might drop the percentage advantage from 57% to maybe 55%.

One more Sim result to report:

Sim3 - DD = 39.4% (8.00 trick avg); SD = 58.0% (8.66)

Kurt

--
Posted by Mimo Usenet Browser v0.2.5
http://www.mimousenet.com/mimo/post

KWSchneider

2017-07-21 13:32:54 UTC

Post by KWSchneider
1) Dummy has exactly QJ.J6.KJ95.KJ752
2) Opener has 12-14 HCP - no constraints were placed on tens.
3) Opener is balanced with no 5+cM
Note that the DD and SD results were run concurrently on the same hands.
Sim1 - DD = 41.7% (8.04 trick avg); SD = 57.0% (8.62)
Sim2 - DD = 40.6% (7.98 trick avg); SD = 57.1% (8.62)
I found it interesting that the SD results were remarkably consistant across the two sims. Since these take 10 hours to run, I intend to repeat the Sims 3 more times to yield 5000 results and hopefully report tomorrow night (EST).
And in view of the 57% SD outcome, I feel vindicated in my position that the result in the posting was more bad luck than poor decision making.
On the other hand, I'm concerned that the DD results were not closer to Travis' 38%. I'd be interested in seeing what the differences in simulation constraints were. Perhaps over 10,000 sims mine will average out to 38%, although they are not trending in that direction.
If anyone would like more detail, please let me know. I have a spreadsheet available for each sim, with the 1000 deals, each hand, and the DD/SD result.

Summary of Results (1000 deals each)
Sim1 - DD = 41.7% (8.04 trick avg); SD = 57.0% (8.62)
Sim2 - DD = 40.6% (7.98 trick avg); SD = 57.1% (8.62)
Sim3 - DD = 39.4% (8.00 trick avg); SD = 58.0% (8.66)
Sim4 - DD = 39.2% (8.03 trick avg); SD = 56.9% (8.62)
Sim5 - DD = 37.6% (7.94 trick avg); SD = 56.0% (8.58)

Over 5000 deals
DD = 39.7% (8.00); SD = 57.0% (8.62)

Kurt

--
Posted by Mimo Usenet Browser v0.2.5
http://www.mimousenet.com/mimo/post

Lorne

2017-07-21 15:37:44 UTC

Summary of Results (1000 deals each)
Sim1 - DD = 41.7% (8.04 trick avg); SD = 57.0% (8.62)
Sim2 - DD = 40.6% (7.98 trick avg); SD = 57.1% (8.62)
Sim3 - DD = 39.4% (8.00 trick avg); SD = 58.0% (8.66)
Sim4 - DD = 39.2% (8.03 trick avg); SD = 56.9% (8.62)
Sim5 - DD = 37.6% (7.94 trick avg); SD = 56.0% (8.58)
Over 5000 deals
DD = 39.7% (8.00); SD = 57.0% (8.62)
Kurt

How confident are you with the SD code? My understanding is that all SD
calcs are done using monte carlo sims with DD calcs for each sim. This
can lead to odd results in certain circumstances eg it is well known
that most SD software will play a trump suit of AKQ10xx opposite xx by
cashing the ace, crossing to dummy, and leading to the K taking an
unecessary risk of a ruff when crossing to dummy.

My concern in this sim is that real defenders look at partners lead and
make assumptions - eg the lead of a K vs NT suggests KQ10xx most of the
time, lead of a low card denies such a holding plus some other touching
honour situations and promises at least one high card for most, but does
the software take this into account when picking deals to use in the sim
for SD plays for a defender ? I can see a situation where failure to
make assumptions available from a lead or lack of a certain lead will
cause the SD result to be inferior to real play for the defenders.

ps if you are wondering why SD software misplays AKQ10xx opposite xx it
is because each individual sim uses DD play to pick the best card(s),
and DD you can pick up all 3-2 breaks + Jxxx onside. Since DD you can
make more contracts if you cross to dummy after the ace (the risk of a
ruff is smaller than Jxxx onside) the sim will say that is the best play
but then having got to dummy the next sim says low to the K is better
than a finesse. It is all because the DD test for each monte carlo sim
will always know when to finesse and when not to after crossing to dummy.

KWSchneider

2017-07-21 21:20:19 UTC

Post by Lorne

Summary of Results (1000 deals each)
Sim1 - DD = 41.7% (8.04 trick avg); SD = 57.0% (8.62)
Sim2 - DD = 40.6% (7.98 trick avg); SD = 57.1% (8.62)
Sim3 - DD = 39.4% (8.00 trick avg); SD = 58.0% (8.66)
Sim4 - DD = 39.2% (8.03 trick avg); SD = 56.9% (8.62)
Sim5 - DD = 37.6% (7.94 trick avg); SD = 56.0% (8.58)
Over 5000 deals
DD = 39.7% (8.00); SD = 57.0% (8.62)
Kurt

How confident are you with the SD code? My understanding is that all SD
calcs are done using monte carlo sims with DD calcs for each sim. This
can lead to odd results in certain circumstances eg it is well known
that most SD software will play a trump suit of AKQ10xx opposite xx by
cashing the ace, crossing to dummy, and leading to the K taking an
unecessary risk of a ruff when crossing to dummy.
My concern in this sim is that real defenders look at partners lead and
make assumptions - eg the lead of a K vs NT suggests KQ10xx most of the
time, lead of a low card denies such a holding plus some other touching
honour situations and promises at least one high card for most, but does
the software take this into account when picking deals to use in the sim
for SD plays for a defender ? I can see a situation where failure to
make assumptions available from a lead or lack of a certain lead will
cause the SD result to be inferior to real play for the defenders.
ps if you are wondering why SD software misplays AKQ10xx opposite xx it
is because each individual sim uses DD play to pick the best card(s),
and DD you can pick up all 3-2 breaks + Jxxx onside. Since DD you can
make more contracts if you cross to dummy after the ace (the risk of a
ruff is smaller than Jxxx onside) the sim will say that is the best play
but then having got to dummy the next sim says low to the K is better
than a finesse. It is all because the DD test for each monte carlo sim
will always know when to finesse and when not to after crossing to dummy.

Indeed, the software uses an intelligent Monte Carlo method - it allocates cards based on bidding (both pairs), lead, type of game (IMPS/MP), and of course what's been played, and then falls back on DD for the eventual line decision. For example, based on the contract, the program determines a number of potential lines via Monte Carlo (allocating high cards and distribution based on bidding or lack thereof), and then plays each line 50-250 times DD (depending on 'skill' level). It scores each line (success %).

For example - here is typical output for a specific declarer play in 3N (in decreasing likelihood of frequency):

03C 05D 08H 17S - DA DQ -> 9
03C 55D 07H 17S + DA D8 D7 D6 D2 - DQ DT -> 9
11C 55D 44H 33S + HA HK H9 - HT -> 9
11C 66D 33H 33S + HA HK -> 9

This line makes whenever LHO has 0-3 clubs, 0-5 diamonds, 0-8 hearts and 1-7 spades with RHO holding the DAQ OR LHO having 0-3 clubs, exactly 5 diamonds, 0-7 hearts and 1-7 spades with DA8762 and RHO holding the DQT OR LHO with 3=4=5=1 distribution and the HAK9 with RHO having the HT OR FINALLY LHO with 3=3=6=1 holding the HAK.

In this case, based on the bidding and other conditions, was scored a 96% line. Ultimately it compares each line and goes with the best success percentage.

But as I indicated in my post, the area where SD falls 'short' is in the inference area. But that is made up by 'always' knowing the long game, and frankly, not making mistakes.

If someone has a a large electronic set of 'human played hands' which could be automated to be played SD, it would give us a quantitative basis for comparison. But outside of that, I have full faith in the software and have yet to find a hand where the results were inconsistent with quality human play.

And finally, if you'd like to give me a few hands where you expect the SD results to be suspect, please let me know and I'll contact you offline. Also, we would need to quantify how often (in the course of all hands), these particular set of circumstances arise. Remember, in the scheme of things, I believe we only need to be within +/-2% to have confidence in the method. I doubt that these few anomalies would account for more than a fractional precentage of all hands.

Kurt

--
Posted by Mimo Usenet Browser v0.2.5
http://www.mimousenet.com/mimo/post

Steve Willner

2017-07-28 21:51:47 UTC

My understanding is that all SD calcs are done using monte carlo sims
with DD calcs for each sim. This can lead to odd results in certain
circumstances eg it is well known that most SD software will play a
trump suit of AKQ10xx opposite xx by cashing the ace, crossing to dummy,
and leading to the K taking an unecessary risk of a ruff when crossing
to dummy.

I thought GIB had fixed that. Kurt: could you easily set up a test?

On the original problem, game is so close to 50% that I don't think
there are on average many IMPs riding on whether one passes, invites, or
blasts game. A deciding factor for me is that an unlucky game bid may
be doubled and down a bunch, so the usual IMP odds for bidding
vulnerable games don't apply.

Robert Chance

2017-07-17 16:54:12 UTC

Post by Fred.
Dealer North, NS non-Vul, EW Vul, IMP scoring.
N E S
1NT(12-14) (P) ?
QJ
J6
KJ95
KJ752

I can't bring myself to pass a 12-count at IMPs, but 3NT is too much of a punt, so 2NT it is.

Apart from anything else, 2NT could prove to be pre-emptive. You could wind up making 2NT with a major suit partial available to the other side.

t***@att.net

2017-07-19 21:30:44 UTC

In terms of defense, an expert once told me that absent the bidding clearly indicating a lead, he could get the suit right about 25% of the time. Things like 1NT-3NT don't indicate much (maybe lead a Major). It's a bidder's game.

Bertel Lund Hansen

2017-07-20 06:21:40 UTC

Post by t***@att.net
In terms of defense, an expert once told me that absent the
bidding clearly indicating a lead, he could get the suit right
about 25% of the time.

That is the percentage you get if you choose randomly.

--
Bertel
bertel.lundhansen.dk fiduso.dk obese.dk

p***@infi.net

2017-07-20 15:23:31 UTC

Post by Bertel Lund Hansen

Post by t***@att.net
In terms of defense, an expert once told me that absent the
bidding clearly indicating a lead, he could get the suit right
about 25% of the time.

That is the percentage you get if you choose randomly.

I think that's exactly what the expert meant -- an exaggeration, of course.

Dave Flower

2017-07-20 11:26:46 UTC