Maybe obvious

[Finch]

I think polish leads are superior to standard and solve also those problems (as well as many others).

Those are:

A = AK

K = KQ

Q = QJ or AQJ

J = HJTx or JTx

T = Tx or HT9x

9 = T9+

2nd from xxx+ and Hxx, 4th from Hxxx+ and low from xx (including 9x but not Hx)

 

I am yet to see one hand where playing "strong ten" is superior to those but the ones when it helps declarer are plentiful (cause imo you really want to play J from HJTx and JTx)

 

 

If you really think that there are NO hands where strong 10s gain, then you aren't playing enough bridge. There are plenty of layouts where knowing at trick one that partner has made a strong lead is enormously helpful. There are layouts where it is enormously helpful to declarer, the point is whether you can work out from the auction which is which and lead accordingly.

[Finch]

So, results: The K is better even with no possible entry? Thank you bluecalm and kgr.

 

It wouldn't totally surprise me that even with no possible entry the K is better than a low one in the suit (although a different suit may be even better).

 

One huge advantage of the king lead is that declarer is often forced to duck and then you can switch successfully, a low one is very committal.

There is no mis-guessing upside from a low one either - one advantage of low from AKxx(x) (although it's going out of fashion) is when declarer gets a Q10 guess wrong (or low from KQ8xx declarer gets a J9 guess wrong).

 

- If the suit is 5332, partner having a doubleton, then the king is at least as good as a low one on every possible layout except Jxx in dummy and Ax with partner.

- If partner has 3 or 4, then the king is always at least as good as a low one

- If partner has a singleton, then it's the wrong suit to lead, but at least the king will usually allow you to switch; it costs when partner has singleton honour, but small singleton is 7:2 more likely

- If partner and dummy have doubletons and declarer has 4 then a low one gains when partner has precisely Ax; it loses when dummy has Jx; it loses a tempo when declarer has the AJ

[mfa1010]

On a related subject, we've recently been discussing the merits of strong 10 leads. These can be a disaster when e.g. dummy has AKx and declarer Jxx; they work well in other situations, notably when dummy is short in the suit - partner leads a strong 10, dummy has singleton Queen, you know to put the king on from Kxxx, but if partner leads the 10 that could be from 1098x or from A109xx you don't know what to do.

There doesn't seem to be a problem covering dummy's Q from Kxxx regardsless, but I suppose you think of the situation where dummy has a small singleton. Then flying king would be bad opposite T9xx.

[JLOGIC]

Really? Obviously (as has previously been pointed out) Europe is not homogeneous... aggressive leads at IMPs are very common among good English players. MPs is of course a different game.

 

Yes he is absolutely right imo, top americans lead too aggressively imo in many spots at imps. The Italians, Poles, Scandanavians all seem to lead more passive. Gavin and I were talking about this recently that one of the biggest changes in bridge at like the nationals is more passive leads, and we think this is largely from the amount of Europeans that come to nationals now plus vugraph. It is interesting to note that Meckstroth and Rodwell in particular have always seemed to lead more passive than their American counterparts. Of course, this is all just my opinion, maybe I just lead too passively, despite what this thread might suggest!

[han]

The day after the hand with Justin came up I had the following:

 

xx J1087xx Kxx xx

 

RHO opened 1NT, LHO bid 2C Stayman, RHO bid 2S, LHO bid 2N (not promising a 4-card major) and RHO bid 3NT. Should I lead low or an honor?

[bluecalm]

Obviously (as has previously been pointed out) Europe is not homogeneous... aggressive leads at IMPs are very common among good English players. MPs is of course a different game.

 

This is only my observation from vugraph. As I saw more vugraph hands than probably anybody else and analyzed play of many top pairs extensively (and made some stats about first lead success for example) it's more than just casual observation but still only an observation of someone who never played at top level (or close to it).

When I talk about "European players" I mean elite players which by definition is very narrow group of the best players who win a lot in international competition and are frequent guests on vugraph.

[bluecalm]

If you really think that there are NO hands where strong 10s gain, then you aren't playing enough bridge. There are plenty of layouts where knowing at trick one that partner has made a strong lead is enormously helpful. There are layouts where it is enormously helpful to declarer, the point is whether you can work out from the auction which is which and lead accordingly.

 

I am not claiming there are no layout like that obviously. I just think it matters so rarely that it's not worth it.

I also don't believe there are many (or any) people out there who could make agreements like: "if bidding suggest we profit more from strong T then we use it and if bidding suggest declarer profits more we don't)" and making more gains from it than losing due to misunderstanding.

 

xx J1087xx Kxx xx

 

RHO opened 1NT, LHO bid 2C Stayman, RHO bid 2S, LHO bid 2N (not promising a 4-card major) and RHO bid 3NT. Should I lead low or an honor?

 

J♥ - 369

8♥ - 357

4♥ - 356

 

On 1000 hands sample (dd simulation) assuming opener has 4spades and max 3 hearts and responder doesn'thave 4 spades and is in 8-9hcp range.

 

I went quickly through some hands and one funny example of when leading low was right is Ax in dummy and 9xx in hand. One example of J being right was KQx in dummy and 9x in hand which points that low heart may be better than dd simul suggests.

Also if there is K9 in dummy and Qxx in hand declarer will often play the K (same with Q9 opposite Axx) - those cases seem to be quite common and could tip the balance toward leading low.

Interesting stuff, unfortunately I am busy with some other things right now so I had to postpone writing my own simulator which would make easy to just list every hand when it mattered which would be helpful to improve our intuition about this stuff.

[JLOGIC]

The day after the hand with Justin came up I had the following:

 

xx J1087xx Kxx xx

 

RHO opened 1NT, LHO bid 2C Stayman, RHO bid 2S, LHO bid 2N (not promising a 4-card major) and RHO bid 3NT. Should I lead low or an honor?

 

I would lead low. Again I feel like bluecalms simulation confirms this, on top of the KQx 9x variation, there's also Q9 in dummy and Ax(x) in hand type variations or Qx A9(x).

[bluecalm]

With 5 card suit instead of 6carder it's even closer dd (J♥ - 362, low - 356). It would be really nice to have a feature which counts various ♥ configurations so we could improve on dd analysis but for now the only way is to just go manually through the hands and count them :(

 

Also matchpoint analysis could be better than imps analysis, because then even on smaller sample the difference will be more apparent. Unfortunately those takes forever and I can't run too many cases.

I made 1000 hand simul with 5 carder (xx JT87x Kxx xxx) and the results were:

 

J♥ - 768

7♥ - 725

4♥ - 723

 

I went through first 250 hands, and there were following situations:

-Q9x - xx, would declarer play 9 or Q ? (probably Q more often) x2

-Qx to A9x - probably again declarer would play the Q x2

-Kx to Q9x; usually declarer plays a K from it

 

So in first 250 hands there were 5 cases where leading low is no worse but dd simuls shows it's worse. This of course suffesrs from sample size issues but it shows it's very very close.

 

Funny case which I encountered when going through the hands where the one where low is better with this configuration:

AQ96 vs K2 where leading low upsets declarer's communication

[whereagles]

This simulation scheme is nice. Bluecam might re-write the standard leads book based on them :)

[han]

The scoring was Patton btw, but I was too afraid to mention that. Dummy had Qx and declarer A9x. Of course declarer played the queen and my diamond king stopped the diamonds so it made for an easy set. The lead of the jack or ten gives declarer a chance.

[benlessard]

for a sim of 1000 hands

 

south lead with

76,JT87xx,KJx,76

 

 

East has 16-17 pts bal with exactly 4S but not 4H. West has 8 pts not 4S not 5H

 

 

i get

 

9 where leading J or low H make a difference double dummy.

 

JT wins

Q9 vs Ax (eliminated because IRL he would play the Q)

K2 vs Q93 (eliminated)

KQx vs 93 (elim)

qx VS a92 (elim)

KQ6 vs 93 (elim)

Q9 vs Ax (elim)

 

 

 

AQ92 vs k6 (north has axxx,x,qxxx,AJT9 )

 

9xxx VS AQ

 

Q93 vs A62 (eliminated because north has obvious lead directing X of 2clubs)

 

so for the 1st 1000 its 0-2 in favor of leading low.

 

The next 1000 set ill remove the J of D.

[benlessard]

for a sim of another 1000 hands

 

south lead with

76,JT87xx,Kxx,76

 

 

East has 16-17 pts bal with exactly 4S but not 4H. West has 8 pts not 4S not 5H

 

i get

 

6 cases where leading high or low H make a difference double dummy.

 

JT wins

 

Q6---A9

AQ62---K95 (this is only legitimate case where leading high is best.)

Q9---K65

Q5---A96

Q5---A9

Q9---A5

 

LOW win

none

 

 

so far for 2000 its 1-2 in favor of leading low.

[benlessard]

leaading High win

 

Q5---A9 (elim)

q6---A9 (elim)

A9x---KQx (elim -- leading low give declarer an important 2nd entry but IRL the contact go down 100%)

 

Q9---Kx (leading low give declarer a very important entry so its eliminated)

Q9---A62 elim

 

leading low win

9xxx--AK (eliminated since declarer can only make double dummy crazy endplay)

 

1-2

 

now for the last 1000

 

edited

[Cascade]

J♥ - 369

8♥ - 357

4♥ - 356

 

These numbers surprise me.

 

Am I misinterpreting them. It seems that almost every time the card led was crucial since they add up to only slightly more than 1000.

[benlessard]

High win

 

Q9---A2 (eliminated)

Q65---A9

KQ62---A9 (however north has a 1S opening so its only good if East is dealer)

 

low win

A952---q6 (elim there is a double dummy endplay IRL 100% going down)

 

So for 4000 hands it end with 3-2 in favor of J wich simply mean that you wont get rich by leading 1 or the other.

 

Its really even.

[benlessard]

i also made a sim for

 

97,QJT54,A65,765

 

against (1Nt)--(3nt) and high H is clearly better

 

http://www.bridgebase.com/forums/topic/46404-simulation-lead-vs-1nt-3nt/

[bluecalm]

These numbers surprise me.

 

Am I misinterpreting them. It seems that almost every time the card led was crucial since they add up to only slightly more than 1000.

 

This is "amount if hands where given card defeats the contract" not "amount of hands where given card is the best lead".

[bluecalm]

Btw:

 

NEC cup 2011, semi final:

 

[hv=pc=n&s=s87542ht7dat75ck9&w=sajthaj5d8643cqt5&n=sk6hkq932dk92c876&e=sq93h864dqjcaj432&d=w&v=b&b=4&a=1d1h1sp1nppp]399|300[/hv]

 

Bocchi chose low ♥ and 1NT made instead of going 2 down.

Simulation (winning lead):

 

K♠ - 556

K♥ - 677

3♥ - 521

2♦ - 492

7♣ - 633

 

Clear enough. Even the best of the best cardplayers could learn something from those simuls ;) (according to double dummy simul his choice was about 1imp/hand worse than the best choice so it's quite a lot)

[MrAce]

Personally I think Bocchi made the correct lead. It was mentioned b4, but i will say it again, double dummy analysis is extremely poor to reflect the real outcome, when it comes to low level contracts for defense. It is very hard for defense in real life, to figure out the best defense, they have very limited info b4 the lead and even after the lead about the shape and strength of opponents.

 

Some may argue that it is same for declarer, which i don't believe. Declarer, whether he plays correct or wrong, 2 hands he controls will be in cooperation with each other, while for defense thats not the case since it involves 2 people.

 

Simulations are great tool, no doubt. But i think they will do their real work when they are capable of performing the BEST % single dummy play and defense instead of double dummy play and defense.

 

KJTx vs A98x playing 7 of this suit everything else is solid and it is split 3-2 or 4-1, simulation will show it as "makes" % 100 of the time, but in real life without any other info, people will fail to make it % 50 of the time. This is a huge gap.

 

In fact, a human defender, if sure they play in a 4-4 fit, can lead trump from Qxxx and reduce declarer's chance to almost % 0

[han]

KJTx vs A98x playing 7 of this suit everything else is solid and it is split 3-2 or 4-1, simulation will show it as "makes" % 100 of the time, but in real life without any other info, people will fail to make it % 50 of the time. This is a huge gap.

 

A good declarer might play a high honor first, catering for a singleton queen.

[Zelandakh]

I have noticed that especially American players lead too aggressively. I even remember that Lawrence in one his books wrote that leading from xxx is usually terrible preferring active leads from honors. This is wrong and modern players lead more passively, especially ones from Europe.

Hi blue,

 

I was meaning times when the "best" card in a given suit is against the tradional advice. For example, the tradtional lead from QJTxx (3 sequence) is the Q (or J playing Roman); similarly from QJ9xx (broken 3 seq) but low not from QJ8xx. So I am interested in times where this traditional advice is wrong. An example is the Bocchi hand where low from KQ8xx is the traditional lead but this was wrong here - but was it wrong in general?. What would be really nice is examples of when it is right to lead low from suits headed by KQJ, KQT, QJT, QJ9, JT9, JT8, HJT, HT9 and conversely when it is right to lead an honour from holdings like KQ9, KQ8, QJ8, JT7, something the Italians seem to do more than other nationalities despite the Bocchi example given.

 

As far as Americans leading aggressively and Europeans passively goes I think this varies. Again, looking at traditional leads, even Culbertson was quite fond of leading xxx in an unbid major. One thing that has changed is that invites have gone out of fashion on marginal hands. Once upon a time players (generally) led aggressively after 1NT - 3NT but passively after 1NT - 2NT - 3NT. Now when the first sequence is likely to be on 23-24 hcp this has gone by the wayside and leading passively makes more sense. That is precisely one of the advantages of this bidding style after all! I am not sure such style issues are so easy to classify using a double dummy analysis method but the question of the best card in a given suit (honour or small) seems to be something that we could easily classify and make new sub-rules for which could be of benefit even to low intermediates.

[MrAce]

A good declarer might play a high honor first, catering for a singleton queen.

 

Ohh sorry that closes the gap by a lot indeed. What u said is no different than spell check. Next time i write % 47 time fail :P

[Cascade]

xx KQT7x Qxxx xx playing imps

 

1N p 3N all pass.

 

Basically I'm curious what is the cutoff point where we stop leading the HK. Or do we always lead it, even without the DQ? Obviously it wins on AJ doubleon or Ax+Jx when they don't have 9 runners which seems easy enough to quantify. The harder part to quantify is how often we have an essential entry, and how often we can beat it by just being passive. I feel like the DQ with nothing else is a pretty good cutoff, with a king or even some stray jacks I would feel good about the HK lead, and with just a jack or nothing I would feel fine about leading a low heart but this one felt tough.

 

100 hands single dummy using GIB as declarer/defender

 

♥K vs ♥7 (very similar for ♥x which did you intend as the choice)

 

284-70 IMPs

 

I am rerunning the simulation for ♦xxxx now.