[paulg]

I reran the simulations to include the best opening lead when the contract makes, which gives a matchpoint flavour to the results.

The first included those hands where 3NT would be bid with any 4333 hand (so may have a 4-card major). The contract failed on 237 deals and the best leads were:

♦K - 673
♠x - 640
♥J - 633
♥x - 610
♦x - 424

On the second, the 3NT bidder did not have a 4-card major. The contract failed on 271 deals and the best leads were:

♦K - 634
♥J - 623
♠x - 628
♥x - 613
♦x - 374

On the third, North did not have a 4-card major nor precisely 3-1 in the majors (as many have a system bid for this). The contract failed on 255 deals and the best leads were:

♦K - 657 hands
♠x - 645
♥J - 631
♥x - 615
♦x - 401

Of course there are many variables in play for such simulations (as you see the number of failing contracts vary), but the 'performance' of the low diamond lead compared to a high diamond looks significant for matchpoints.

[Free]

Just wondering here: isn't it better to calculate the total number of tricks taken by opponents, and then see which lead gives opps the lowest number? Or will we get more AVE+/AVE- instead of top/bottoms?

[jdonn]

Wow, I'm certainly sold on the high diamond over the low diamond anyway. Maybe low is better without the 9 since we often give declarer a guess with the J and 9 in the two hands, and lose the opportunity to smother JT doubleton.

[Fluffy]

jdonn, on Jul 30 2010, 03:13 PM, said:

Wow, I'm certainly sold on the high diamond over the low diamond anyway. Maybe low is better without the 9 since we often give declarer a guess with the J and 9 in the two hands, and lose the opportunity to smother JT doubleton.

I think the big problem with low one is that we make no trick in the suit when partner has doubleton.

[cherdanno]

JLOGIC, on Jul 30 2010, 05:29 AM, said:

Obv a diamond is the best lead to beat it/at imps.

Oh wait this is matchpoints?

[pooltuna]

jdonn, on Jul 30 2010, 10:13 AM, said:

Wow, I'm certainly sold on the high diamond over the low diamond anyway. Maybe low is better without the 9 since we often give declarer a guess with the J and 9 in the two hands, and lose the opportunity to smother JT doubleton.

I suspect holding the KQT♦ instead of KQ9♦ might have a bigger impact on the results in this particular case as this will destroy some of the holdings where Jxx opposite xxx produce a stopper against the K leaders holding KQ9

[Siegmund]

In a double-dummy analysis, K from KQxx(x) and Q from QJxx(x) substantially outperforms low, against almost any notrump contract, with or without good spots. I remember being quite surprised when I saw this back in December or January, and we had a thread discussing reasons why small was still (in general) the better lead single-dummy.

[gwnn]

heart lead is best, spade is awful. diamond is in the middle.

[whereagles]

JLOGIC, on Jul 29 2010, 09:01 PM, said:

I would feel really confident that a double dummy simulation would find a diamond the worst suit to lead by a ton.

I would feel really confident you're wrong

Small diamond for me, because I got no side entries. King if I had.