[mike777]

Your lead at MP and why?
Your lead at IMP and why?

1nt=3nt

9752
Q76
T974
J3

[Blofeld]

My leads tend to be awful, but I'll try the ♦T at MP, and the ♥6 at IMPs.

[Edit because I can't tell my suits apart]

[inquiry]

mike777, on Sep 13 2005, 07:53 AM, said:

Your lead at MP and why?
Your lead at IMP and why?

1nt=3nt

9752
Q76
T974
J3

At matchpoints I don't want to blow a trick. I will lead the a ♠, which I view as safer than a diamond.

At imps, I want to beat the contract. Parrtner will need a lot of something to beat it. I have to many spades to be useful. An attacking lead is best. I see two options. A low heart so partner can return one (we play partner for good hearts). Or lead the club JACK (we play partner for a long clubs and hope the jeck is deadly). The fourh diamond makes teh chances dummy is long in ♣ too great, so I start the ♥.

[P_Marlowe]

Hi,

Jack of clubs, trying to hit partners
suit, solving all declarer problems

Playing MP, the ten of diamond
is probably your safes lead, but
...

With kind regards
Marlowe

[ArcLight]

Partner has a good hand.
The opponents probably have a lot in the minors.

At IMPS
♥Q.

♥6 is standard. But I'm broke, and probably wont get in again, unless I can stop Diamonds. If pard has the A and declarer the K, the suit can become blocked on the lead of the low heart. Leading the Q allows a continuation if declarer ducks. Also, if dummy has the K and you lead the Q and pard a tenace, you may make 5 hearts.

At MPs I don't want to blow a trick, but I still think declarer has the minors and I dont want to lead them. The field will probably lead a heart, so I'd do the same as above.

[paulg]

At IMPs I'd lead the ♥6 but not with confidence and I would not be surprised to see that a spade worked.

Although there has been mention of leading the ♥Q I feel this is worse than a low card. It is certainly the lead that gets into the papers when it works, but on the occasions when partner does not have 5 solid hearts the ability to lead back to your ♥Q could be very important. It also retains its value as an entry if a heart is not the killer.

Paul

[luis]

Heart, automatic.
I did a study/simulation on the lead after 1N-3N with a bad hand.
A 3 card major is better than a 4 card major when your suit is weakish,
a 3 card major is better than a 4 card minor or a 3 card minor.
Heart, automatic!

[Chamaco]

Low heart at IMPS, diamond Ten at MP.

I think/hope leading from T97x won't waste a trick often, and still might be a decently aggressive lead.

[sceptic]

would lead a heart due to luis's simulation on his web site, made sense to me in that ezact same auction.

aslo can you put ur site back up luis

[keylime]

At MPs, I'm a big believer in leading spades versus 1NT-P-3NT.

At IMPs tho, it's a tossup. I rather like a heart myself.

BTW, nice to be back after my enforced hiatus. :-)

[han]

luis, on Sep 13 2005, 08:58 AM, said:

Heart, automatic.
I did a study/simulation on the lead after 1N-3N with a bad hand.
A 3 card major is better than a 4 card major when your suit is weakish,
a 3 card major is better than a 4 card minor or a 3 card minor.
Heart, automatic!

did you get these results for matchpoints too?

[mikeh]

At imps a small ♥. My experience is that we almost never beat these hands when you hold this type of hand, but if you do, it will be partner taking the tricks. So lead the suit in which you hold the most help. You will usually blow an imp (altho against a good team, playing similar methods, you may expect a push) but you maximize the chances of a beat.

At mps, I go with a ♠: the spot depending on my methods: usually 2nd highest from this holding. The ♦ is NOT safe(neither is the ♠, but there are fewer holdings on which the ♠ gives declarer something to which he is not entitled). The ♣ is both unlikely to hit partner's suit ( responder rates to hold 4+♣) but may well blow a trick: picture dummy with Q109xx opp declarer's Axx

[Jurek S]

♥ on IMP, seems quite standard,
♥ on MP also but will accept the blame if it's wrong without post-mortem fight

[Chamaco]

I post here 20 simulated hand taken from a 50 hand subset (I have / can make more, but I do not want to clutter the thread too much).

The simulation constraints were:

South = 15-17 hcp, any 5332, 4432, 4333, and 5422 with minors. I did not include 6m322 in this set, I might in the future

North = 9-15 hcp, any hand without 4 card major OR any 4333.

I am sure in the real world there are deviations from such constraints, but I considered it to be more or less representative of the "field"

As I am not an expert, you can try your own analyses.

--------------------------------

From a quick glance, I could estimate that in a a fair number of hands the lead would be irrelevant.
In some cases a major suit lead would be appropriate, but in almost the same number of cases it could waste a trick.
(Basically, often opener had one or both major with a tenace waiting to trap East honors that would otherwise hard to guess, especially when dummy had a singleton in the suit)

I could estimate more or less the same of a diamond lead: sometimes it did waste a trick when dummy had H8x(x) in diamonds, but it did not waste a trick, even given that holding, in all hands where NS held AKQJ in diamonds.
Perhaps the biggest downside of the diamond lead was that, in some cases, even if not wasting a trick directly, it did not attack declarer's weak spot, although in some cases the diamond Ten would be the only lead defeating the contract.

As a whole, the major suit lead (even spades) tended to be riskier than the diamond lead, but there were more casew where it would defeat the contract (although in a few cass, the diamond lead would be the only good one even at IMPS)

In a few cases the best lead would be a club.

-----------------------

Hand 1

Scoring: IMP

---

West North East South

 -     -     -     

Hand 2

Scoring: IMP

---

West North East South

 -     -     -     

Hand 3

Scoring: IMP

---

West North East South

 -     -     -     

Hand 4

Scoring: IMP

---

West North East South

 -     -     -     

Hand 5

Scoring: IMP

---

West North East South

 -     -     -     

Hand 6

Scoring: IMP

---

West North East South

 -     -     -     

Hand 7

Scoring: IMP

---

West North East South

 -     -     -     

Hand 8

Scoring: IMP

---

West North East South

 -     -     -     

Hand 9

Scoring: IMP

---

West North East South

 -     -     -     

Hand 10

Scoring: IMP

---

West North East South

 -     -     -     

Hand 11

Scoring: IMP

---

West North East South

 -     -     -     

Hand 12

Scoring: IMP

---

West North East South

 -     -     -     


Hand 13

Scoring: IMP

---

West North East South

 -     -     -     


Hand 14

Scoring: IMP

---

West North East South

 -     -     -     


Hand 15

Scoring: IMP

---

West North East South

 -     -     -     


Hand 16

Scoring: IMP

---

West North East South

 -     -     -     


Hand 17

Scoring: IMP

---

West North East South

 -     -     -     


Hand 18

Scoring: IMP

---

West North East South

 -     -     -     


Hand 19

Scoring: IMP

---

West North East South

 -     -     -     


Hand 20

Scoring: IMP

---

West North East South

 -     -     -     

[kfgauss]

I put this hand in Jack and had it simulate 1000 deals. This is double dummy, of course, so we lose interesting info like when we have erased guesses for declarer which do show up often as Chamaco's simulation suggests. (Does anyone know simulators that have some limited ability to correct for "guesses" like this -- e.g. deciding a 2-way finesse makes half the time? Perhaps GIB or Jack does this to some extent -- it seems they would need to know something about 2-way finesses for their play engines.)

Also, the output is expected (i.e. average) total points, which means it's only really relelvant for imps. Again, I repeat my request for suggestions for a better simulator (more user configurable, with output giving the number of times each result was achieved for each lead would be great).

Jack's definition of 3NT is:
1) 9 hcp, a 5-7 card minor, no 4+ card major [edited]
2) 10-13 hcp, no 8+ card minor, no 4+ card major
3) 14-15 hcp, no 6+ card minor, no 4+ card major

Expected total points (opps non-vulnerable):

♠2 = -388.8
♥6 = -397.4
♦10 = -403.4
♥Q = -407.0
♣J = -415.5

For those who know something about Jack, I have adjusted ♥Q to the correct double dummy value (Jack originally gave -417.0) -- Jack docks systemically incorrect leads 10 points to help itself decide which lead to make -- i.e. it makes the "incorrect" lead from a suit if its expected total point score is more than 10 points better, which is interesting & reasonable, but not of interest to us right now, though I guess we should note that the ♥Q lead could cause partner to go wrong. Of course, the ♥6 lead could cause partner to go wrong too. I led low from ♦A52 against 3NT recently (it was the unbid suit), hitting partner's ♦K8xxx and he had no entries and had a very tough decision of whether to duck when ♦10x showed up in dummy. If anyone wants to comment on this, does anyone advocate leading the ♦5 from my holding for some reason (even though we play standard leads vs NT), or does anyone advocate systemically leading 2nd from 3 systemically vs NT, so ♦5 would be the systemic lead? (Of course, tell me why too.)

Andy

[FrancesHinden]

kfgauss, on Sep 14 2005, 12:42 PM, said:

Expected total points (opps non-vulnerable):

♠2 = -388.8
♥6 = -397.4
♦10 = -403.4
♥Q = -407.0
♣J = -415.5

Given the range of errors from mis-guesses etc, that just tells you that 3NT is making whatever you lead.... so isn't enormously helpful!

[kfgauss]

FrancesHinden, on Sep 14 2005, 06:17 PM, said:

kfgauss, on Sep 14 2005, 12:42 PM, said:

Expected total points (opps non-vulnerable):

♠2 = -388.8
♥6 = -397.4
♦10 = -403.4
♥Q = -407.0
♣J = -415.5

Given the range of errors from mis-guesses etc, that just tells you that 3NT is making whatever you lead.... so isn't enormously helpful!

As these things go, this 10 point difference is fairly significant. Notice that the ♣J is only 12 points worse than ♦10, even though this seems to be a much worse lead (at least to me). If the only possible results were -430 or +50 (I choose -430 because 397 is frightfully near 400), this 10 point difference would translate to beating the contract 8.9% of the time instead of 6.8% of the time. (This is a silly hack computation, but I apologize for the fact that Jack has silly output.)

Also, it seems to me that the spade lead needn't erase declarer's guess more frequently than the heart lead away from Qxx. Maybe that's so, and I certainly favor thinking over rigidly accepting the output of flawed simulations, but it's at least rather interesting that the spade lead beats 3N more often than the heart lead double dummy, especially given that the vast majority of the responses favor the heart lead.

The majority of responses favor the heart lead because they think it's more likely to set, but will often blow a trick. Perhaps this is true (even double-dummy), but then the 1 imp or so we're gaining by leading the spade is worth more in the long run (accepting the above double-dummy analysis) because at total points you should be even more likely to go for the set than at imps.

If people still believe the heart lead is best, perhaps it'd be useful for them to think about and explain why a heart lead is likely to have/preserve a larger gain for the defense over pure double-dummy play than a spade lead.

Andy

[hrothgar]

kfgauss, on Sep 14 2005, 08:42 PM, said:

Expected total points (opps non-vulnerable):

♠2 = -388.8
♥6 = -397.4
♦10 = -403.4
♥Q = -407.0
♣J = -415.5

Hi Andy...

Any chance that you could provide a more precise explanation regarding how you ran the simulation... Based on your post, I'm assuming that you did something like the following

1. Deal 1000 hands consistant with the auction 1N - 3N
2. Go to hand 1 and Force Jack to lead the ♠2. Record the score
3. Repeat for the ♥6, ♦10, ♥Q, and ♣J
4. Go to hand 2, hand 3, hand 4, ... and repeat
5. Calculate the total points by summing the scores across each opening lead
6. Divide by 1000 to get the expected value for each lead

One quick question:

Does Jack permit you to easily run any tests to determine if the results are statistically significant?

[kfgauss]

It wasn't manual at all. Jack has a built in "analyze this situation" button which will deal out 1000 deals consistent with the bidding (and play, though there hadn't been any at this point) and display the expected total point score for each action --i.e. it takes each action on each of the 1000 deals and (using the double dummy result of each lead) averages the total points achieved. This is rather different from having Jack play all the hands -- double dummy is of course quite different. I don't have an easy way of doing what you suggest (it would be very nice to be able to do that though).

This feature seems to be one that Jack uses internally to decide how to proceed in various situations (such as bidding "judgement" problems, opening lead problems, much of the play).

I haven't really looked around, but I'd be very interested in programs that can do this with more user input (it only uses its definitions for the bids) and also which display the number of times each result was achieved as opposed to the irritating averaging going on. It has no (built-in at least) ability to run tests to see if the results are statistically significant, but I will say that the results rarely change by more than 5 total points from around deal 200 on (you see the results change as it deals more hands -- it takes a minute or two total).

One can also run it several times and see if you get roughly the same answer. Here are the results of another run in case that's interesting:

♠2 = -387.3
♥6 = -397.2
♦10 = -403.6
♥Q = -406.4
♣J = -414.8

I would give another few runs, but it seems to have a random number seeding issue and keeps giving the same results [even if i restart the program]. I've rebooted my computer since my previous post which has apparently reset it, but it'd be an annoyance to do this several times. In any case, these results are fairly close to the original ones and hopefully give you an idea of the accuracy. Let me know if there's anything else you'd like to know.

Andy