Friday, June 19, 2009

PABF Championships

I played interclub last night so missed the second half of our Juniors against Australia on Vugraph. Since BBO saves its Vugraph sessions I was able to download it and construct a text file report of the match which is below.

It's incomplete but I think they had some technical problems. The commentary is in a block at the end of each hand rather than mixed in with the play and this means that sometimes the comments refer to the previous hand. I also included a Deep Finesse style double dummy analysis.

Here is the link: PABF_R1-2.txt

UPDATE:
I've added two more, the NZ women against Australia and the open team against Singapore:
Here is the link: PABF_R1-4.txt
Here is the link: PABF_R1-7.txt

The first one is definitely worth a look. I'm not the kind of person who normally knocks women's bridge but there are two hands in there that are really quite special. Board 19 features an eight card spade suit headed by AK that was not bid beyond the two level at one table and not bid at all at the other table. On board 25, Australia came close to making 3NT at both tables.

Thursday, June 18, 2009

Responses to 1NT

One of my pet hates is 'practice Stayman', where a hand with no interest in the majors has to start with Stayman in order to invite 3NT. It helps the defence too much, especially on the opening lead. I'd rather guess between pass and 3NT than do that. There's no need for it because usually the 2S and 2NT bids are assigned meanings that seldom provide any real benefit.

I would suggest the following structure of responses to 1NT:

2C Stayman, always has a major if balanced
2D/H Transfers
2S Range finder or weak with long minor or GF 3 suiter with both minors
2NT Both minors, either weak or GF
3C/D Natural invitational
3H/S Natural slam try 6+ suit

After 2C:
2C-2D-2H: Both majors weak
2C-2D-2S: 5S and 4H game interest but not forcing
2 any - 2NT: invitational
2 any - 3C/D: Natural GF 5+ suit may have no major.
2D-3H/S: strong major 3 suiter, 3H=C shortage, 3S=D shortage
2H/S - jump in a new suit is a splinter.
Other major (i.e. 2C-2H-2S or 2C-2S-3H) is slam try with 4 card support and no shortage).

After 2D/H:
New suit game forcing except 2D-2H-2S which is forcing to 2NT or 3H/S only
Transfer then jump (eg 2D-2H-3S/4C/4D) is a splinter (in support of own suit :). With 6+ major, slam interest and no shortage start with 3H/S over 1NT.
Superaccept in new suit by opener shows values, or 2NT with scattered values but I like to just bid three of responder's suit to avoid telling them anything. May superaccept with a good maximum and only three card support.

After 2S:
Opener bids 2NT with min and 3C with max, then:
3C/D - Signoff
3H/S - Three suiter showing shortage

After 2N:
Opener bids better minor and natural after that. 3H/S by 2NT bidder is a fragment not a real suit as minor three suiters start with 2S.

After 3C/D:
Opener will usually pass or bid 3NT. 3 of a suit implies a fit and willingness to play 4/5 of the minor if stoppers for NT are not there.

After 3H/S:
Cue bidding

I think this handles everything pretty well and is not much to remember. Information about declarer's hand is withheld where reasonably possible. All hand types are covered. One suiters start with transfer if a major or 2C if a minor. Major two suiters start with transfer and bid other suit. Minor two suiters start with 2NT. Major/minor 2 suiters start with transfer if 5 major and 2C if 4 major. Three suiters start with 2C if both majors and 2S with both minors.

If they double our 1NT

Suit bids show a five card or longer suit
Pass forces redouble and bids then show spades and the suit bid
Redouble shows a two suiter without spades. Opener bids best minor then natural.

The advantage of doing it this way is that the hands we are most likely to able to compete on are the ones where we have a 5-4 fit or a 4-4 spade fit. These are identified immediately to allow opener to act if there is further bidding. Despite their double the hand may still belong to us or we could gain from pushing them a level higher.

If they double in 4th seat (ie 1N-P-P-X-P-P) then redouble with one suiter and bid lowest suit with 2 suiter.

Use the same method if they double our 1NT overcall or natural 2NT overcall.

Sunday, June 14, 2009

Simulations

I'm going to start a series of posts simulating outcomes in various bidding situations. I'll try to clearly list the assumptions in each case.

One problem that I can't easily do anything about is the difference between double dummy analysis and results that occur at the table. I'm not sure if any research has been done into quantifying this, but if it has, please let me know.

The big advantage for declarer playing double dummy are that missing cards (e.g. queens) can always be located and singleton kings offside always drop. The bid advantage for defenders are they always find the best opening lead and single dummy defence is just hard because you can't see your side's combined resources in they way that declarer can.

In slam contracts, I think declarer will take more tricks double dummy. This is just because there are fewer decisions for the defence when they hold fewer high cards. Not that defence against slams isn't important, but declarer gets a higher percentage of the tough decisions in slams compared to games and part scores. As a general rule, I'd guess that the higher the contract, the more double dummy analysis inflates the number of tricks taken by declarer. In a contract such as 1NT I expect the defence to do better double dummy than in real life.

If there is a significant difference between tricks taken double dummy and in real life, my conclusions will be wrong by that amount.

Inviting game after single major raise

This post focuses on the hands that are borderline between passing and making a game try after you open 1H/S and partner raises to 2. I'm making the following assumptions:

1. Partner will have three card support only.
2. Partner's HCP range is about 4-8 (but not 4 HCP if 4333). If you play 2/1 GF then you probably respond 1NT on very weak hands with support so your range is more like 6-9. I'm assuming 2/1 responses can be only around 9+ and will include the hands with three card support and 9 HCP. If not, adjust the strength of the opening hand accordingly.
3. Partner will always decline the game try with 4-5 HCP and always accept with 8. With 6-7 he will choose the correct action 70% of the time (because your game try will enable him to evaluate more accurately).
4. For the IMP calculations, I assume they pass at the other table.
5. I'm also disregarding the possibility that they will balance if you pass. This is obviously quite likely and argues in favour of bidding, not so much to shut them out, but because you no longer have to worry about losing IMPS in the case where the other table just make the two level contract your way.

5431 Hands

Tricks7 or less8910 or more
AK873HCP=42.01%3.81%2.78%1.10%
KJ65HCP=52.88%6.28%6.79%3.20%
AT5HCP=61.33%5.37%8.61%5.98%
4HCP=70.68%3.72%10.19%10.52%
HCP=80.29%1.77%7.45%15.26%
Expected IMP gain from inviting: Vul 0.52, Non Vul -0.08

The numbers in the table are the percentage of hands in each category according to the simulation. Partner's HCP are on the rows and the number of tricks you take on the columns.

So you show a slight loss by bidding on nonvul but it's worth continuing vul. This was a little surprising to me because I would never consider passing that hand, regardless of vulnerability. Possibly I have not increased my standards for making game tries while I have been progressively lowering my standards for single raises. Let's try with a slightly weaker hand:

Tricks7 or less8910 or more
AK873HCP=43.04%4.03%1.97%0.37%
KT65HCP=53.16%6.95%5.49%1.62%
AT5HCP=62.37%7.35%8.49%3.90%
4HCP=70.73%5.75%10.66%7.76%
HCP=80.36%3.36%9.77%12.87%
Expected IMP gain from inviting: Vul -0.66, Non Vul -0.89

So this one is a clear pass. A rule of thumb is that 5431 shapes need a good 15 to invite (or 39 on the 11-7-4-2-1 scale).


6322 Hands

Tricks7 or less8910 or more
AK8732HCP=41.64%3.60%3.08%0.83%
K6HCP=52.50%7.15%6.14%2.20%
AT5HCP=61.42%6.52%9.24%5.38%
43HCP=70.56%4.78%9.72%9.38%
HCP=80.30%2.63%9.38%13.55%
Expected IMP gain from inviting: Vul -0.08, Non Vul -0.5

This one has a good 14 HCP and 37 11-7-4-2-1 points. It's a tossup vulnerable and a pass nonvul.

Tricks7 or less8910 or more
AK8732HCP=41.03%3.34%3.64%1.40%
K6HCP=51.79%6.04%7.37%3.68%
AJ5HCP=60.74%4.78%9.36%7.55%
43HCP=70.33%3.08%8.85%12.04%
HCP=80.12%1.75%7.09%16.01%
Expected IMP gain from inviting: Vul 1.06, Non Vul 0.27

Changing the D10 into a jack makes it worth another move. It's now about an average 15 and has 38 11-7-4-2-1 points.


5332 Hands

Tricks7 or less8910 or more
AK873HCP=43.17%3.89%2.44%0.56%
KT2HCP=53.85%7.04%5.38%1.99%
AQ5HCP=62.58%7.67%8.89%4.28%
43HCP=71.23%5.06%9.60%8.58%
HCP=80.45%2.81%8.17%12.37%
Expected IMP gain from inviting: Vul -0.51, Non Vul -0.77

This above average 16 HCP hand is not worth inviting, though you wouldn't find many players passing it. Let's make it a bit better:

Tricks7 or less8910 or more
AK873HCP=42.38%4.04%3.69%1.26%
KJ2HCP=52.65%6.40%7.34%3.33%
AQ5HCP=61.42%4.85%8.51%6.81%
43HCP=70.66%3.40%8.73%11.36%
HCP=80.31%2.07%6.40%14.39%
Expected IMP gain from inviting: Vul 0.65, Non Vul 0.01

It's now a typical 17 and is worth a move, but nonvul it is still right on the margin. So 5332 hands need quite a lot to invite. There's also less chance they will balance so you may get to play 2S when it makes just eight tricks. However you will also be able to make 3NT sometimes when 4S goes down so a 2NT game try if available could still be worth a shot.

The minimum to invite with 5332 is a very good 16 or 42 on the 11-7-4-2-1 scale.


5422 Hands

Tricks7 or less8910 or more
AK873HCP=41.98%3.42%3.77%0.67%
A2HCP=52.37%6.09%8.76%2.48%
KQT5HCP=61.35%4.71%10.66%5.55%
43HCP=70.81%3.23%9.53%10.61%
HCP=80.31%1.71%7.24%14.73%
Expected IMP gain from inviting: Vul 0.54, Non Vul -0.06

This is a good 16 and has 41 11-7-4-2-1 points. It's just barely worth a move.


6421 Hands

Tricks7 or less8910 or more
AK8732HCP=42.11%3.67%3.03%1.01%
K2HCP=52.71%6.08%5.90%3.34%
QJT5HCP=61.96%5.73%7.68%6.06%
4HCP=71.25%4.80%9.17%9.65%
HCP=80.71%3.18%8.34%13.64%
Expected IMP gain from inviting: Vul 0.06, Non Vul -0.38

This is borderline with 13 HCP and 33 11-7-4-2-1 points.


6331 Hands

Tricks7 or less8910 or more
AK8732HCP=40.96%4.19%3.42%1.44%
A32HCP=51.87%6.20%6.77%3.46%
QJ5HCP=61.05%5.18%9.37%6.40%
4HCP=70.51%3.68%9.39%10.87%
HCP=80.23%2.12%7.95%14.94%
Expected IMP gain from inviting: Vul 0.57, Non Vul -0.06

Here you need 14 HCP or 35 11-7-4-2-1 points.


5521 Hands

Tricks7 or less8910 or more
AK873HCP=41.48%4.24%3.87%0.74%
A2HCP=51.78%6.24%6.97%2.59%
QJT52HCP=61.14%5.96%9.53%5.52%
3HCP=70.41%3.89%9.67%10.59%
HCP=80.22%2.34%8.18%14.64%
Expected IMP gain from inviting: Vul 0.35, Non Vul -0.21

This is right on the margin, again a hand that few players would consider passing. This is possibly the most striking example, partly because people routinely overbid 5-5 hands. It has a good 14 HCP and 36 11-7-4-2-1 points.


Summary

The following is the minimum strength to invite for each hand shape:
ShapeHCP11-7-4-2-1
64211333
63311435
5521Good 1436
6322Average 1538
5431Good 1539
5422Good 1641
5332Very good 1642

A further interesting point is how little the vulnerability affects the decision. Even on the marginal hands above, there are only a couple where bidding is right if vulnerable but not if non-vulnerable. Certainly the difference is small, maybe less than a ten. This needs further investigation.

Saturday, June 13, 2009

Anti Percentage

The following hand came up today in the Nickell vs Welland match to decide who plays Fleischer for the second USA spot in the Bermuda Bowl. Others may prefer to write about doubled grand slams making with two top losers, or the correct way to bid a 0607 shape, however the more mundane but higher frequency hands like this one tend to interest me.

How much would you bid with the South hand opposite a weak notrump?

Below is the Meckstroth-Rodwell auction but South at both tables jumped to game after discovering four hearts in the North hand. My instinct was that inviting is enough so I decided to run a simulation. The simulation is based on North as declarer with a balanced 11-14 including 4 hearts. I chose North on the basis that you either play a weak NT or transfer responses to 1C. This probably improves the expected number of tricks a little.

Results are as follows:

HCPNum HandsNum Makes%
1111120169615.25%
129720293530.20%
138371434251.87%
146784449566.26%

This suggests that inviting is plenty, and opposite an 11-13 hand even pass is possibly correct, especially when you also take into account hands where eight tricks are the limit. Certainly non-vulnerable it is right to pass if partner has a maximum of 13 and just invite if his maximum is 14.

Dealer: N Q 10 3
Vul: N/S A 9 8 7
K 10
Q 8 7 5
A K 9 5 7 6 4
2 J 10
8 7 6 5 3 Q J 9 4
10 9 6 A K J 3
J 8 2
K Q 6 5 4 3
A 2
4 2
WNES
1P1
P2P4
PPP