[kgr]

We had a discussion Yesterday. 4 persons/3 different opinions. What is best play with:
AQ9xxx-xx
One said A first, one small to Q and I said small to 9. (Small to 9 better if LHO play small from KJTx)
I verified it in suitplay and small to 9 seems best.
But if I force LHO in suitplay to play small from KJTx then suitplay says that this is not best defense. So suitplay assumes that LHO will never play small from KJTx? Why is small to 9 then best play?

This is complete hand for play in 4H (I'm not sure about the bidding):

Pairs/MPs

---

Pairs; North leads ♣7.

[gnasher]

Small to the 9 gains a trick when LHO has J10x.

[gwnn]

As LHO you mean the person before the long suit i.e. South here?

[gnasher]

 gwnn, on 2015-May-10, 03:38, said:

As LHO you mean the person before the long suit i.e. South here?

Yes, the same player as kgr called "LHO" in his original post. That was just an answer about how to play the suit combination.
On the actual hand I wouldn't do that, because it risks a trump promotion.

[gwnn]

 gnasher, on 2015-May-10, 04:39, said:

Yes, the same player as kgr called "LHO" in his original post. That was just an answer about how to play the suit combination.
On the actual hand I wouldn't do that, because it risks a trump promotion.

Yes sorry, I was trying to clarify the OP mainly. Interesting suit combination.

[kgr]

 gnasher, on 2015-May-10, 04:39, said:

Yes, the same player as kgr called "LHO" in his original post. That was just an answer about how to play the suit combination.
On the actual hand I wouldn't do that, because it risks a trump promotion.

Thanks!
You mean: Trump promotion if North started with doubleton C and f.i. ♥J8?
North had:

---

It wins here to play A and small; OR 9 and A. But neither of these plays should be found I think?

[awm]

This might be an interesting suit combination for GIB.

Holding AQ9xxx opposite xx, the best line is to play low to the nine and then low to the queen. Of positions it's possible to pick up for one loser, this fails only against KJ/KT offside.

Playing the ace first is inferior, as you will end up losing to Kx and JT offside (assuming you lead to nine next if an honor falls, and to queen next if not, which is best). However, double dummy it can never cost to play the ace first! Will GIB get this wrong?

Another interesting thing is, suppose you start low to nine and it loses to the ten or jack. Then you lead towards the long hand again and see a small card. The only relevant positions are Jxx, Txx, Kxx in front of the long hand. If opponents would always play two low cards on all these holdings, it is better to play ace now (Jxx/Txx twice as likely as Kxx). But the best defense is to play an honor on the second round with Jxx/Txx, so against expert defense you should play to the queen! Again, it is interesting whether GIB gets this right, and also what the right play is at the table against less than expert defense.

[MrAce]

 awm, on 2015-May-10, 13:32, said:

Another interesting thing is, suppose you start low to nine and it loses to the ten or jack. Then you lead towards the long hand again and see a small card. The only relevant positions are Jxx, Txx, Kxx in front of the long hand. If opponents would always play two low cards on all these holdings, it is better to play ace now (Jxx/Txx twice as likely as Kxx). But the best defense is to play an honor on the second round with Jxx/Txx, so against expert defense you should play to the queen! Again, it is interesting whether GIB gets this right, and also what the right play is at the table against less than expert defense.

No

Why is Jxx/Txx is twice as likely when T or J was already played previous round? I am not even mentioning 1 extra undertrick.

[gnasher]

 MrAce, on 2015-May-10, 14:37, said:

Why is Jxx/Txx is twice as likely when T or J was already played previous round?

Because it's a restricted choice situation. Or, more intuitively, because the J and 10 are equivalent.

[MrAce]

 gnasher, on 2015-May-10, 14:54, said:

Because it's a restricted choice situation. Or, more intuitively, because the J and 10 are equivalent.

I can see the restricted choice. I just can't see the "twice as likely" odds for playing A which I disagreed.

KJ84-T = You are going extra down %2.83
K84-JT = You are going down % 3.39
J84- KT =You are making %3.39
84-KJT = wash %3.39

It is a MP event and we are playing game with adequate values. What am i missing?

[nate_m]

 MrAce, on 2015-May-10, 16:58, said:

I can see the restricted choice. I just can't see the "twice as likely" odds for playing A which I disagreed.

KJ84-T = You are going extra down %2.83
KJ8-JT = You are going down % 3.39
J84- KT =You are making %3.39
84-KJT = wash %3.39

It is a MP event and we are playing game with adequate values. What am i missing?

Your statement indicates that you don't see the restricted choice. Your argument is essentially the same as somebody who says "I have Axxx vs K10xxx and see the J fall, hooking wins vs. J and rising wins vs. QJ, why anything to do with the Q relevant?" Of course, if you hook every time you see an honor, you win vs. Q, J, and only lose to QJ, and that must be taken into account.

After low to the 9 and J/10, you cannot pick up 4-1 splits but rising ace wins vs. KJ and K10 offside, losing only to J10 offside. That's "twice as likely." You do lose an extra undertrick on the 4-1 splits, so it's not quite 2 to 1. This all assumes the opponents never find high from 10xx or Jxx. All this was ofc already noted by mikeh above.

[MrAce]

 nate_m, on 2015-May-10, 19:53, said:

Your statement indicates that you don't see the restricted choice. Your argument is essentially the same as somebody who says "I have Axxx vs K10xxx and see the J fall, hooking wins vs. J and rising wins vs. QJ, why anything to do with the Q relevant?" Of course, if you hook every time you see an honor, you win vs. Q, J, and only lose to QJ, and that must be taken into account.

After low to the 9 and J/10, you cannot pick up 4-1 splits but rising ace wins vs. KJ and K10 offside, losing only to J10 offside. That's "twice as likely." You do lose an extra undertrick on the 4-1 splits, so it's not quite 2 to 1. This all assumes the opponents never find high from 10xx or Jxx. All this was ofc already noted by mikeh above.

I know exactly what it is. N could take 9 with J or T from JT, while he will take with only T from KT.
I just don't put my money on restricted choice as much as Adam on this one. At least you acknowledged and did not totally ignore the possibility of undertricks in a MP game when we expect most people to be in it.
It was not MikeH btw.
Restricted choice assumes that people take 9 with either J or T randomly. In real life this is not even remotely true. I can bet that at least % 75 will take it with J. While in a bad field I expect it to be the opposite. I can not support this with a scientific data of course. I wish there was a tool in BBO in which we can search JT doubletons in defense and see what do people usually win with and who they are.