This blog provides supplementary thoughts and ideas to the site. If you haven't seen the main site, there is a lot there including the Martel and Rodwell interviews, photos, and articles. This blog is focused on advancing bridge theory by discussing the application of new ideas. All original content is copyright 2009 Glen Ashton.

Saturday, November 28, 2009

A Balanced Viewpoint

Ulf Nilsson writing in the November issue of The Bridge World (instead of his bridge blog ) makes the strong case for treating 5-3-3-2s with a five card major as balanced as the primary feature, and not the major suit. That means incorporating these hand types into the openings for the balanced hands in the appropriate range, something that the ETM Gold system used as a key lynchpin of the system.

Say one has 11-12 high card points, 5-3-3-2, and a five card major. System design could include these possible options:

Open 1 of a suit, in the most natural way possible given the system (i.e. if playing five card majors, only open 1M with five)
Open 1 of a suit, in the most natural way possible given the system except open 1C if 4-4-3-2 exactly
Open best minor even if 5cM
Open 1C except if 4-5Ds, then open 1C
Open 1D except 1M if five card major
Open 1D
Open 1C except 1M if five card major
Open 1C
Open 1NT except 1M if five card major
Open 1NT
(this is not an exclusive list - for example there are systems that open 2M with 5-3-3-2s, and we will see others below)

If you were just looking at the balanced hands themselves, then natural is best - get the suit bid puts one in the best competitive advantage. However as Nilsson begins in the article:
Responder's knowledge of opener's hand-type is far more useful than any other immediate distributional message opener can send. Why? Because optimal bidding methods vary substantially opposite balanced and unbalanced holdings. … Effective competitive strategy also varies, because balanced hand are generally more defensive in nature than unbalanced hands.
This implies, as the article discusses, that one wants to achieve separation of hand types in the opening framework. In other words including balanced hands in a natural suit opening contaminates it from a competitive aspect. For example it goes 1D-1S(overcall)-Double(you)-2S;-P-P-? - if opener can be balanced, the decision to bid again is quite different than if you know 1D is unbalanced.

Systems where both 1NT and a specific one level suit opening were reserved for balanced hands have been rare. Granovetter-Rubin toyed with an unpublished system a few years ago (seen practiced online), and more recently a system was discussed with 1C and 1D openings showing specific majors, and 1H was balanced in a certain range, while 1NT handled another balanced range, talked about on David Collier's blog here:

Now let's look at this candidate system:

1C: 11-13 or 17-19 bal
1D: unbal Cs or Ds
1M: 5+, unbal
1NT: 14-16 bal
Rest: standardish

Now 1C is pretty cool, while 1D will have troubled waters in competitive situations, such as 1D-3S(overcall)-?.

Note that for the system frameworks in this post, hands with a six card minor, and no singleton or void, will be considered balanced (abbreviated "bal", and "unbal" for unbalanced). Likewise 5-4-2-2s with a five card minor will be considered balanced in all frameworks.

Starting with the candidate system, redesign can move some of the unbal C hands into a 2C opening, using a trick to get a wide range: if the 2C opening promises 6-4+ or 7+Cs, then it can be played as wide ranging for two reasons. First it is almost always safe for responder to ask opener's hand type since opener will have good playing value. Second, the high-card point range of the opening will be limited since the high card point range between a preemptive hand and a game force is less on shapely hands (e.g. it takes less points to make a game forcing hand when there is a lot of shape). In addition let's shift the singleton diamond hands with 4 to 6 Cs into 1C, since these will be close to balanced hand types.

This results in:

1C: 11-13 or 17-19 or 22-24 bal , or 4-6 Cs unbal with singleton D
1D: 3+Ds, unbal, either Ds or longer Cs
1M: 5+, unbal
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2D: Any game force
2NT: 20-21 bal

Taking a Polish base, one could move the Ds hands into 1C, and the balanced hands to 1D:

1C: 3+Ds, unbal, either Ds or longer Cs, or 22+ bal or any game force
1D: 11-13 or 17-19 bal
1M: 5+, unbal
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2NT: 20-21 bal

That 1C opening produces the neat sequence 1C(variety)-1D(negative);-P, which is not legal under the ACBL GCC, which requires the artificial 1D response to be forcing.

Modifying a big club, five card major base:

1C: 15+ unbal or 17+ bal
1D: 11-13 bal or 11-14 4-4-1-4 or 4-4-4-1 exactly
1M: 5+, unbal, 10-14
1NT: 14-16 bal
2C: 10-14, 5+Cs
2D: 10-14, 5+Ds
2NT: 20-21 bal

Taking the 4 card majors out of the 2C and 2D openings (and the 4-4-4-1s out of 1D):

1C: 15+ unbal or 17+ bal
1D: 11-13 bal
1M: 4+, unbal, 10-14
1NT: 14-16 bal
2C: 10-14, 5+Cs, no 4cM
2D: 10-14, 5+Ds, no 4cM
2NT: 20-21 bal

In the above framework and some of the others below, one could want to play 1NT 10-13 when not vulnerable in 1st, 2nd or 3rd, and then have 14-16 open whatever 11-13 opens when vulnerable.

Utilizing the 2M openings:

1C: 15+ unbal or 17+ bal
1D: 11-13 bal
1M: 4+, unbal, 10-14, only 6 or longer in M if 4+ in other major
1NT: 14-16 bal
2C: 10-14, 5+Cs, no 4cM
2D: 10-14, 5+Ds, no 4cM
2H: 10-14, 6+Hs, not 4Ss
2S: 10-14, 6+Ss, not 4Hs
2NT: 20-21 bal

Coalescing some of the above:

1C: 3 to 6 in each minor, unbal, or 22+ bal or any game force
1D: 11-13 or 17-19 bal or 11-20 singleton/void m with 4-5 in other minor
1M: 5+, unbal
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2D: 10-18, 6+Ds and if only 6Ds then a four card major
2NT: 20-21 bal

Making 1D, 2C and 2D more limited, at the price of more stuff piled into the trunk of the 1C opening:

1C: 3 to 6 in each minor, unbal, or 17+ bal or 18+ any
1D: 11-13 bal or 11-17 singleton/void m with 4-5 in other minor
1M: 5+, unbal, 10-17
1NT: 14-16 bal
2C: 10-17, 6+Cs and if only 6Cs then a four card major
2D: 10-17, 6+Ds and if only 6Ds then a four card major

Limiting 1D only:

1C: 3 to 6 in each minor, unbal, or 19+ unbal with a minor or 22+ bal or any game force
1D: 11-13 or 17-19 bal or 11-18 singleton/void m with 4-5 in other minor
1M: 5+, unbal
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2D: 10-18, 6+Ds and if only 6Ds then a four card major
2NT: 20-21 bal

Interchanging the 1C and 1D:

1C: 11-13 or 17-19 bal or 11-18 singleton/void m with 4-5 in other minor
1D: 3 to 6 in each minor, unbal, or 19+ unbal with a minor or 22+ bal or any game force
1M: 5+, unbal
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2D: 10-18, 6+Ds and if only 6Ds then a four card major
2NT: 20-21 bal

Using 1C as the balanced outside-of-the-1NT-range bid:

1C: 11-13 or 17+ bal or 11+ singleton/void m with 4-5 in other minor
1D: 3 to 6 in each minor, unbal, or any unbal 18+
1M: 5+, unbal, 10-17
1NT: 14-16 bal
2C: 10-17, 6+Cs and if only 6Cs then a four card major
2D: 10-17, 6+Ds and if only 6Ds then a four card major

It's tempting to use 2H and 2NT to make 1C limited:

1C: 11-13 or 17-21 bal or 11-18 singleton/void m with 4-5 in other minor
1D: 3 to 6 in each minor, unbal, or any unbal 19+
1M: 5+, unbal, 10-18
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2D: 10-18, 6+Ds and if only 6Ds then a four card major
2H: 22-24 bal
2NT: 24/25+ bal, game force

We can move the 2D hand types into 1D, since if responder bids clubs, opener, if holding short clubs, can correct to diamonds on the same level. This allows the 2D to handle strong hands and make both 1C and 1D non-forcing:

1C: 11-13 or 17-21 bal or 11-20 singleton/void m with 4-5 in other minor
1D: 3 to 6 in each minor and/or 6+Ds, unbal, less than a game force
1M: 5+, unbal
1NT: 14-16 bal
2C: 10-18, 6+Cs and if only 6Cs then a four card major
2D: Any game force
2NT: 22-24 bal

Another tack is to use diamonds as the only suit somewhat moved out of the one level:

1C: 3+Cs, unbal, less than a game force, can have 5Ds with shorter Cs
1D: 11-13 or 17-21 bal or 11-21 singleton/void C with 4-5 in Ds
1M: 5+, unbal
1NT: 14-16 bal
2C: Strong, 16+ with 6+Ds unbal or any game force
2D: 10-15, 6+Ds unbal
2NT: 22-24 bal

This has the cool 2C(Strong)-2D(waiting but limited);-P sequence. 2C could start at 15, making 2D just 10-14, something I would definitely do in ACBL mid-chart event where all 15+ artificial openings are legal.

In mid-chart events one could try something like:

1C: 4+Hs, unbal, forcing
1D: 4+Ds, unbal, less than a game force, can have longer Cs
1H: 15+ bal or 4-3-1-5/3-3-1-6 exactly
1S: 5+, unbal
1NT: 12-14 bal
2C: 10-15 6+Cs or 11-14 4-3-1-5 exactly
2D: Strong, denies 4+Hs, 16+ with 6+Cs and no other suit except Ss possible, or any game force without 4+Hs
2NT: 21-22 bal

In the land of the system free:

1C: 4+Hs, unbal, or 10-15 4-3-1-5 exactly, or 17+ bal forcing
1D: 4+Ds, unbal, less than a game force, can have longer Cs
1H: 10-13 bal
1S: 5+, unbal
1NT: 14-16 bal
2C: 10-15 6+Cs
2D: Strong, denies 4+Hs, 16+ with 6+Cs and no second suit or 5+Cs and 4+Ss, or a game force with Ss or Ds
2NT: 21-22 bal

In the land of the price reduction, taking advantage of the US "Black Friday" sales yesterday I bought a Roman atlas. Here's a touch of Roman in the last two frameworks:

1C: 6+Cs 10-14 unbal, or any unbal 15+ or 17+ bal
1D: 10-13 bal
1M: 5+, unbal, 10-14
1NT: 14-16 bal
2C: three suiter with no 5cM, 10-14, often a 5-4-3-1 with a 5c minor
2D: 10-14, 6+Ds
2NT: 10-14, 5-5 minors, no 3 card major

1C: 6+Ds 10-15 unbal, or any unbal 16+ or 17+ bal
1D: 10-13 bal
1M: 5+, unbal, 10-15
1NT: 14-16 bal
2C: 10-15, 6+Cs
2D: three suiter with no 5cM, 10-15, often a 5-4-3-1 with a 5c minor
2NT: 11-15, 5-5 minors, no 3 card major

The last again has the cool 1C-1D(negative);-P sequence, when allowed. For these last two I would open 1NT with 10 to a poor 13 when not vulnerable in the first three seats, making 1D a decent 13 to 16, and with a 10-15 three suiter with a singleton A, K, or Q, I would open 1D, not the Roman two bid showing a three suiter.

Here's a bonus framework:

1C: Either:
a) three suiter with no 5cM, 10-14, often a 5-4-3-1 with a 5c minor
b) any unbal 15+
c) 17+ bal
1D: 10-13 bal
1M: 5+, unbal, 8-14
1NT: 14-16 bal
2C: 10-14, 6+Cs
2D: 10-14, 6+Ds
2H: 10-14, 6+Hs
2S: 10-14, 6+Ss
2NT: 10-14, 5-5 minors, no 3 card major

And finally the balanced hands completely separated:

1C: Either:
a) three suiter with no 5cM, 10-14, often a 5-4-3-1 with a 5c minor
b) 10-14 6+Cs
c) any unbal 15+
1D: 11-14 bal
1M: 5+, unbal, 8-14
1NT: 15-17 bal
2C: 18-21 bal
2D: 10-14, 6+Ds
2H: 22-24 bal
2S: 10-14, 6+Ss
2NT: 24/25+ bal

For any of the frameworks given above, if you know a close match with an existing system please comment, providing any information and links you may have, and thanks in advance to all!


Sunday, November 22, 2009

Structure Revelations

As I mentioned in the last post, using the robots might be the fastest method of testing a new bidding structure or complete system. Last month I played thousands of hands where I opened 1NT, and the playtesting of the robots notrump structure had a couple of revelations.

I've been a big believer in responder showing the singleton/void to the notrump opener, but it surprised me the ratio of how many times that was useless to how many times it was useful. On the vast majority of hands that responder wants to get to game with, if there is no major fit the contact will be 3NT, and showing the singleton/void rarely changes that.

For hands without a four card major (4cM), the robot notrump structure has two common ways to show a singleton/void on the way to 3NT. These are:

1) Use minor suit Stayman (1NT-2S) with both minors, and then show a specific major suit singleton/void.
2) Transfer to the minor (2NT to Cs, 3C to Ds), then bid a new suit at the three level to show a specific singleton/void (there is no way to show long Ds, short Cs, no 4cM, at the three level).

From playtesting the structure, it would be better to give opener the opportunity to refuse to know the singleton/void and hand types. Let's remap the 1NT-2S response to be game forcing, no 4cM, and a hand with a singleton/void. Opener can rebid 2NT to learn about responder's hand type, or rebid 3C, which says not interested unless lots of extras in shape and/or points.

Here's the detailed structure:


2NT: what you got for me?
--3C: 6+Ds, any singleton/void. 3D asks shortness.
--3D: 6+Cs, major suit singleton/void. 3H asks shortness.
--3H/S: both minors, singleton/void in bid major.
--3NT: 6+Cs, singleton/void diamond.
-- 4C: 6+Cs, singleton/void diamond, good slam interest or better.

3C: not interested.
--3D: a long minor with any singleton/void, extra shape/values. 3H asks suit (3S=Cs, 3NT and above = Ds)
--3H/S: both minors and extra shape/values, singleton/void in bid major.
--3NT: to play

Based on playtesting many hands, most auctions will go 1NT-2S(GF, no 4cM, singleton/void any suit);-3C(not interested)-3NT.

This frees up the sequences 1NT-minor transfer then new suit bids, which can be employed in various ways. One thing that is missing from the above use of 2S, compared to the robot structure, is the good slam tries with no singleton/void (1NT-minor transfer and then 3NT is a mild slam try). These could be implemented via the minor suit transfer:

1NT-2NT(transfer to Cs);-3C-?
--3D: good slam try in Cs, no void and often no small singleton
--3H: slam try with both minors, no void and often no small singleton
--3S: slam try with both minors, excellent clubs

1NT-3C(transfer to Ds);-3D-?
--3H: good slam try in Ds, no void and often no small singleton
--3S: slam try with both minors, excellent diamonds

Likewise the sequences showing 4-4-4-1s/5-4-4-0s (no 5cM) by bidding 1NT-3D, 1NT-3H, or 1NT-3S to show the shortness are a waste of time - using Stayman works almost all the time. After Stayman finds no major fit, if opener has a 4cM opposite a major suit singleton 3NT is usually as good as spot as any, and if opener has denied a 4cM, bidding 3C/D with the longest minor can investigate the stopper situation.

One option is to play 1NT-3D as showing long clubs and a 4cM - over 3D opener can ask with 3H (3S=Hs, 3NT+ = 4Ss), bid 3S stopper showing, bid 3NT to play, or bid 4C forcing to 5C. Then 1NT-2C;-2D-3C shows an unspecified singleton/void in any suit but clubs - 3D would ask, while 3NT would unask - that is opener does not want to know what it is.

The other revelation from playtesting was in competitive sequences, where we open 1NT, and an overcall is made directly before responder's first bid. Using Lebensohl many times shows that it is folly for responder, with a weak hand and a long suit, to be unable to immediately bid the suit.

For example in standard Lebensohl: 1NT-2S(overcall)-2NT(Lebensohl)-3S;-?. Opener will want to compete opposite some hands, safely pass in others, and double with good spades and no fit for responder. All that is impossible when opener does not know responder's suit, if any, for the Lebensohl bid.

Note that most partnerships now use "system on" after the 1NT-2C overcall, and thus we will just consider when the overcall is 2D, 2H, or 2S.

There has been increased use of Rubensohl (spelled various ways), and Transfer Lebensohl to allow responder a way to immediately show a suit. The basic format is:

2NT: transfer to Cs, to play or game forcing
3C: transfer to Ds, to play or game forcing
3D: transfer to Hs, to play or game forcing
3H: transfer to Ss, to play or game forcing
3S: transfer to 3NT, with no stopper in opponent's suit
3NT: to play with at least a partial stopper in the opponent's suit

If a suit could have been bid naturally at the two level, the transfer at the three level shows at least invitational values. A transfer to their known suit is game forcing Stayman, and if opener bids the transfer suit it denies a stopper.

Here's a more detailed discussion:

A version called Rumpelsohl has this main twist: 2NT, the transfer to clubs, is either clubs or a game invite in any non-club new suit that could not be bid naturally at the two level. For example 1NT-2S(overcall)-2NT(transfer to Cs)-P;-3C-P-3H is a game invite with 5+Hs.

I prefer invites to have some direct bid, and to lose the invite if unavailable. This is like Rubensohl as described above, but with this switch:

1) A transfer to their suit is a natural invite
2) 2NT, the transfer to clubs, followed by 3D is game forcing Stayman, and then bidding their suit denies a stopper.

For examples:

1NT-2S(overcall)-3H: invite with 5+Hs.

1NT-2S(overcall)-2NT(transfer)-P;-3C-P-3D: Stayman.

--Double: partnership option (penalty, negative, optional, values etc.)
--2S: natural, 5+Ss, non-forcing
--2NT: transfer to Cs, could be Stayman if next bid 3D
--3C: transfer to Ds
--3D: game invite in Ds
--3H: transfer to Ss, game invite or better
--3S: transfer to 3NT, no stopper
--3NT: to play, at least a partial stopper

It would be neat to reprogram the robot notrump structure to try this out.


Sunday, November 15, 2009

The Robot Battles Part VI - All for the best

"Are you alive? … Prove it"

The opening lines of the BattleStar Gatlactica series.

This is the last of a series of posts on battling the BBO Robots, although the next posting will discuss notrump structure based on robot tourney experiences. The robots don't bid or play the best, but they might be the fastest method of testing a new bidding structure or complete system.

The robot tourneys give you the best hand, or at least a hand that has the most or equal number of high card points as any other hand at the table. This has certain implications for the bidding, but one has to tread carefully.

When you have the best of the best, you can open 2C, but you have to live with the robot getting to notrump first by bidding 2C-2NT, and now the system is awful, with all suit bids showing 5 or longer and the robot sometimes hiding support. When it goes 2C-2D you've dodged the 2NT response bullet, and will usually have a nice auction.

Having the worst of the best, a hand in the 11-13 range, gives you key information: nobody has a better hand. It robot partner overcalls, and you have an 11 count, you know that robot opener likely has a shapely 11 count to open, and that partner has a minimum overcall. There is no need to jump the bidding or cuebid here, as you can pass or manage the bidding to the right spot.

Likewise opening a "weak two" with a poor 12, or any 11, tends to work well since partner will not have many hands with game interest, and this usually gets to the right spot quickly, avoiding robot silliness that can occur on slow auctions. However opening a weak two with just a five card suit frequently fails, as the robot's decisions on when to compete or not to compete is erratic. When a robot opens a weak two against you, that erraticism can work for you, but the lack of "system on" after your 2NT and 3NT overcalls is painful.

When you open 1H or 1S on 11-13, the robot will tend to bid 1NT forcing since it usually doesn't have enough to force to game. It is dangerous to pass this (assuming you have the five card major promised), since the robot can easily have three card support and a shapely game invite hand. However the 1NT forcing structure is imperfect - I've had a robot bid 1S-1NT;-2D-3C on a 2-4-3-4, and the definition of 3C just says 4+Cs - the robots don't like to correct back to 2M on doubleton support. If the robot bids a 2/1 after 1H or 1S, it is never safe to pass it, even if robot is a passed hand - they love to hide support - for example, P-1S;-2D can easily have 3Ss. If the robots need any convention it is Drury, and it wouldn't hurt for them to read (scan) a book on 1NT forcing.

If you open 1C or 1D on 11-13, the weaker you are the stronger the temptation will be to pass robot's suit response. First remember by passing a new suit you are having "trick disposal" robot play the hand, and results tend to be less than the best. However if robot "partner" has already bid the major suit your side is going to play in, then passing an 11 or poor 12 count can at least keep robot at a makeable level, and the robot opponents tend not to balance on the auction 1m-1M;-P. If you don't know if you are in the right strain, then don't pass unless you are shooting for tops - then you might pass an auction like 1C-1S with a 11 count and 3 spades, hoping to catch five spades and a hand that would push too high if you had bid again.

After an inverted minors raise, 1m-2m, you can pass if minimum since partner will have a worse hand and you are playing it. Passing the inverted minor when 12-14 balanced tends to work poorly, since 2NT and 3NT will usually be a better scoring parking spot: after the inverted minor, bid 2NT on most balanced 12-14s, even though the system will tell you that 2NT is just 14 - give yourself 1 or 2 bonus points for being declarer. If your hand is minimum and unbalanced, passing the inverted minor is best, though occasionally responder is so shapely that five of a minor turns out to be a nice spot.

When robot opens and one is in the 12-13 range, opener is quite limited (equal or less high card points than you) even though he doesn't know you know that. When you can place the contract do so - for example after a 1M opening, with a good fit, don't bother with Jacoby, but just bid 4 of the major. Likewise if you open 1H or 1S on 11-13, and responder uses Jacoby, slam is quite unlikely with two minimum hands, so either just bid 4M even with shortness, or bid a fake shortness (e.g. 1M-2NT;3C showing singleton/void in clubs, then retreat to 4M) to sow the seeds of robot confusion.

Sometimes you can have quite a weak hand, which means the points are about even around the table, and you have a questionable opener. Never pass out - if need be reduce your amount of alcohol consumption. Actually there is one case, aside from 20 tequila shots, when you should consider passing out: doubleton or shorter in both majors, and 11 (or even 10, but never saw that) - on these the robots will often compete in a major, and it's hard to bring in a plus score.

Sometimes the robot will assume you have the best of the best, but you don't. For example it goes 4S-P-P-? to you, (4S by left hand robot). If you hover your mouse over various bids you will see that partner will assume a bid of 5C by you now has a lot of playing value. When I first started playing the robot tourneys, it would go 4S-P-P-5C(me);-P-6C-All Pass down one with lots of swearing by me. Now after playing thousands of hands, it goes 4S-P-P-5C(me, I'm not here to defend);-P-6C-All Pass, down one but no swearing, as I now know the robots have great earplugs.

If the robots were to take those earplugs out, and pass along some recommendations to the programming team, here are my top three:

1) Eliminate the Bot Tell (discussed in a previous post).
2) Only give the South hand the best hand 95% of the time, removing the certainty that everybody has a worst hand.
3) When North becomes declarer, rotate the hands temporarily in order that the non-robot plays all North-South contracts - as discussed on BBO forums this would make these individual contests (see previous post on this) even more a matter of skill.

The bridge crew of Battlestar Galactica believe that the Cylons plan to eradicate all human participation in order to increase the tourney's skill level, and have one of the "final five" robots reach 100,000 masterpoints first. To prevent this bot plot, please encourage the ACBL to keep the tourneys alive: let humans remain in the online robot tourneys.

Wednesday, November 11, 2009

The Robot Battles Part V - Tactical Bids

This is the penultimate of a series of posts on battling the BBO Robots.

The only time you should psych in a robot tourney is when you are bored, need some entertainment, and want to relearn why you should not psych. It boils down to:

- It is pointless psyching when you have the best hand (okay, not pointless, but pointfull);
- Your robot partner will believe your psych until the bitter end, ignoring all evidence to the contrary.

However tactical bids work very well by telling little lies (when not vulnerable little white lies) that will put the robot opponents on the wrong trail, and that can put robot "partner" on the right one.

Some tactical bids involve nondisclosure, as disclosed in the last posting. For example, after a Jacoby transfer, 1NT-2D or 1NT-2H, if you want to superaccept I suggest you ignore the system (which involves showing a specific suit doubleton), and simply jump to 3 of the major regardless of where the doubleton is. The superaccept will still get you to game when you need to be there, and will often get the robot opponents to misdefend.

After 1NT-2S minor suit Stayman, it's best to always bid 2NT, even when holding a minor suit, since you want to put partner on the trail to 3NT, and not to five of a minor. Sometimes denying a four card or longer minor will get the robot opponents lost on the defense.

The most common tactical bid is the triple-x rated minor bid. When you have xxx in a minor suit (three little, or triple-x), bidding the suit can act as lead deflectional (okay, that's not in dictionaries just yet, but let's not get deflected). There are two common places for using the bid:

i) After robot "partner" opens 1H or 1S, bidding 2C or 2C with triple-x can put you on the trail to 3NT, and deflect the lead of this suit. A typical auction: 1M-2m;-2M(can be 5 if not extras)-2NT(game force, natural);-3NT. Even if robot raises the minor, he will often let you play in 3NT if that's your next bid.

ii) As an opening bid, with 11-13 high card points, xxx in the minor you bid, and a natural opening in the other minor (the one you ignore). There is some risk here, as it sometimes goes 1m-any;-1NT-2m: that is "partner" will pull your 1NT into 2 of your yucky minor, often a 4-3 fit and a poor score. However it works the majority of the time, and it is a useful way to shoot for tops - if your percentage is less than 57, it provides the opportunity to increase it without having lost too much when it fails. The auction 1m-overcall-3m seems to produce okay results, and don't retreat into 3NT as it often has no chance.

The key to tactical bids is little lies, not big ones. It's much like the notrump rule: a point and/or a card away: if you stretch that to two points (e.g. opening 1NT with 13) more bad things happen than good things. Likewise you can try opening 1C or 1D with two little, but you will soon find you have too little in your minor.

Tactical bids, or distortions as I sometimes call them, are useful to have in your arsenal for live bridge. When live defenders know you are an active bidder who can make these calls, when picturing your hand they have to consider a wider range of potential hands, which makes finding the best defense tougher. Consider using robot tourneys to practice up your tactical bids and get comfortable with them, and then make your live opponents less than comfortable playing against you.

Next and last: All for the best

Saturday, November 07, 2009

The Robot Battles Part IV - Full Nondisclosure

This is the fourth of a series of posts on battling the BBO Robots.

The Robots have some strange ideas on what to lead - they love to lead singleton trump honors, or trumps on no-fit auctions, or random suits instead of the suit you bid - perhaps this is just robot revenge rage from being cutoff in the bidding. However their specialty is the passive lead against notrump: they love to lead from Jxx, Txx, Qxx, Jxxx, Txxx, and Qxxx in an unbid major against notrump. Their passive leads can be quite effective against those who assume robots would lead fourth best from longest and strongest, but the leads don't work particular well against those who know to watch out for them.

Picture that you have KQT3 in hearts and 15 points balanced. Now you would love to have a lead away from Jxx or Jxxx into this holding. How can you make this happen more often?

Say you open 1NT, and "partner" bids Stayman - if you now bid 2H, the normal Stayman reply, and partner bids 3NT, hearts are no longer unbid, and the robot leader (not Optimus Prime of the Transformers leading the autobots, but the robot leading a card) will no longer heart you.

The solution I found was to always reply 2D to Stayman with a decent four card or longer major - this convention is called Staybot (formerly Stayrobot, thanks Chris for the name suggestion). The exceptions are: always give the correct reply to Stayman if you have both majors (strong chances you have a major fit), or if you hold a weak four card major (since you don't want this suit led). There are a three management issues with this convention:

1) After 1NT-2C;-2D, if the Robot now uses Smolen, you cannot backtrack into your 4-4 fit - the Robot will always correct back to the five card major now that you denied 4 in the other major. For example, you hold four decent spades, and fewer than four hearts, and use the Staybot convention: 1NT-2C;2D-3S(Smolen 4S&5Hs);-4S-5H;-5S-6H;6S-7H: regardless of how many times you bid spades, once you bid 2D, it will always put you back in heart land: the auction needs to go either 1NT-2C;2D-3S;-3NT or 1NT-2C;2D-3S;-4H.

2) You miss your 4-4 fits to play in notrump. However much of the time you will get a nice lead (not Optimus for the robots), and make the same number of tricks in notrump as you would have in the suit contract.

3) After 2NT-3C;-3D-?, the robot sometimes gets discombobulated and bids 4C or 4D on assorted junk - if you cuebid 4H or 4S, sometimes it will bid 4NT Blackwood and you can pass that, or you may find yourself in 6NT. The bad news is that these 6NT contracts can have no play. The good news is you may still make them with the helping hands (claws?) of the robot opponents.

The use of the Staybot convention is an example of nondisclosure - we deny a major suit in the hope we get a helpful major suit lead. The other main place to use nondisclosure against the robots is after partner opens one-of-a-minor. On most hands, regardless of major suit length, or stoppers, just jump to 3NT: 1C-3NT or 1D-3NT. On average these bids are highly successful, although occasionally you will get to a terrible spot: don't write for refunds.

Over the major suit openings, nondisclosure is harder. Usually I bid 1NT (forcing), then 3NT, but there are two problems:

1) The robot opener can pass 1NT forcing after 1H-1NT;-? when holding 4-5 in the majors. The big plus (aside from 1NT making 6) is nobody taught the robots Flannery.

2) On auctions like 1M-1NT;-any-3NT;-?, the robot will sometimes pull to 4M with just 5 in his major, and now you are in robot hell: wrong contract played by your "partner", and now if you bid 4NT, its Blackwood. Don't write for refunds.

Another way to get to 3NT when robot opens 1M will be covered in our next posting: tactical bids.

Thursday, November 05, 2009

The Robot Battles Part III - Pig Out

This is the third of a series of posts on battling the BBO Robots.

This is a picture of a nice robot, a gift from our oldest daughter:












It cleans the house.

This is a picture of a bad robot:



It also sucks, but not as a cleaning action.

Your primary duty in the bidding is to prevent any robot from becoming declarer, and especially the very dangerous GIB N robot pictured above. To do this you want to bid lots, but there are two landmines to avoid:

1) If you make a takeout double, or make a Michaels cuebid, or unusual or sandwich notrump, you are making a bid that gets the robot to bid the suit first - danger Will Robinson danger! - instead use natural bids (btw a takeout double is not "natural" but conventional), even if the bids are far less than perfect.

2) If you open and the opponents interfere while bad robot above remains silent, then if you bid again, bad robot will assume you have an amazing hand for freely bidding. For example, say you open 1NT, the next hand bids 2S, it comes back to you and you bid 3C - the robot thinks you have 15 to 17 high card points still, but it assumes you have 20 or so of what it calls "total points" - lots of playing value. In these situations you are in a tough spot - either pass out 2S, defend with bad robot and get a bad score, or bid, have bad robot push you overboard and get a bad score perhaps. It's the perhaps that can work for you: if you do bid, and get overboard, the opponents can misdefend to turn straw into gold.

The number one tactic to becoming declarer is the same tactic that second rate pros use with their third strata partners:

bid notrump first
For me the rule was any hand that was a point away and/or a card away from a 15-17 notrump opened 1NT. That meant a 1-3-4-5 14 count opened 1NT (one point away, and if you moved the club to the spade you would open 1NT). It meant a 5-4-1-3 18 count opened 1NT.

However a world class expert told me that he did not have good results in robot tourneys opening 1NT with 14s, so you are going to have to find your own comfort zone for opening 1NT, especially since the the bad robot pictured above can aggressively invite over your 1NT opening. I also used the point away/card away rule for 1NT overcalls, and I used it for 2NT 20-21 opening bids.

Thus a 12 board robot tourney might see me open 1NT 8 times, overcall 1NT once, open 2NT twice, and on the last of the dirty dozen have to bid a suit.

However there is yet another landmine: bad robot will never believe you have a five card or longer major if you open 1NT. Once I opened 1NT with a seven card major (okay more than once), and then bid my major, rebid my major, and rebid bid it once more (we were now several levels too high): the robot still assumed I only had four cards in the suit.

The good news is that your robot opponents, defending your contract after you opened notrump with a five card or longer major, will assume you don't have one either. For this to work well, hide your major length until later in the hand and the robots will often misdefend, assuming length in your hand in other suits.

Before you allow GIB N to play a hand remember this rule:

The No Dummy Rule: I'm not paying money to watch a robot misplay a hand
Something else to watch for is the "penalty double" by GIB N. If at all possible pull penalty doubles as GIB N will either have the wrong cards for it, and/or will make a silly lead, and/or will misdefend. Especially be careful of the situation where the opponents are bidding wildly, as they will often be shapely enough to bring home their contract (or in robot speak, bring the contract back to its docking station).

If you strive to hog the bidding, the robot tourneys become extra fun because you get to play the vast majority of the hands. I just brought in (from bridgebase's myhands) the hands I played from Oct 5 to the end of the month, and I was declarer on over 72% of them. Now that's pigging out on declarer play.

Next up: nondisclosure

Tuesday, November 03, 2009

The Robot Battles Part II - Disguise and the Bot Tell

This is the second of a series of posts on battling the BBO Robots.

In the late seventies at the University of Waterloo I attended a guest lecture by a world expert in artificial intelligence (AI). With incredible prescience he explained that the future of AI was not rule-based, by having a program judge a position by applying many rules (e.g. if the following twenty conditions are met, the best line will be a cross ruff), but by using processing power to examine end states (e.g. play out thousands of different lines, find out cross ruff works best in the majority of cases). Thirty years later, the BBO Robots with the GIB software use the processing power method to "think" - they are constantly running simulations of the current position to determine what will likely work best.

To run these simulations the BBO Robots take what they "know" - what information they have about the current layout - produce many example hands that match the knowledge, and then quickly play out each of the example hands to determine what seems to work best much or all of the time. It repeats this step ever time it has to play a card, and can even change its mind from card-to-card if new knowledge has been gained, or if the example set of hands becomes skewed in a new direction.

One of the best ways to combat the BBO Robots is to hinder them from knowing the true situation. If they create the wrong example hands, they will often play less than optimal. Thus one should try to disguise or hide the actual situation from the robots, and to only reveal key information when you need to.

This matches the situation in live bridge, where most players don't disguise their holdings enough. BBO Robot tourneys are a great place to practice your disguise techniques, given that Halloween is over and trying out disguises at your local bridge club may not be well received - okay, given that you want to try the techniques over and over again, to see when and how to use them.

Say you are declarer, and have 642 in hand in a suit, with decent length and strength in dummy. The vast majority of players will play the 2 first, the 4 next time, and the 6 last, even though saving the 6 and the 4 makes absolutely no difference. It can be better to play the 6 in some circumstances, so that your opponent, live or robot, doesn't know where the 4 and 2 is. For the robots, not knowing where the 4 and 2 is can change the example hands they produce to try to work out the best play to hurt you.

Against live opponents you want to use a mixture of disguise and truth - sometimes play the 2, sometimes the 4 and sometimes the 6. If you always played the 6 from 642, they would then know what you have or don't have when you play the 2 first. However Robots don't learn from hand-to-hand - they treat you as a new opponent. Thus you want to maximize your use of disguise against the Robot opponents, and that's why the tourneys are a great place to try out these techniques.

Say I opened 1NT with a six card major (more on this in the next post), and end there. If the Robots lead that suit, I will pick my cards to play the suit as if I had a nice three card suit, instead of true situation. The Robots will often be fooled by this disguise - they will assume, in their example layouts, that their partner is the one with the length in the suit, and defend, and mis-defend accordingly. Always look to assist the Robots in misreading the actual situation, by using disguise, and hiding information.

Say you are playing out a hand, and you will need to give up a trick to the Robots, and at that point they can hurt you with one play, and help with another. If you delay giving up the trick, the Robots gather more and more information, and then when you give up the trick their example layouts are very accurate. If you give you the trick early, and have hidden key information, the Robots will often form a wrong picture of the hand, and make the play that helps you.

This brings us to the "tell" and how it can help you. As Wikipedia tell us:

A tell in poker is a subtle but detectable change in a player's behavior or demeanor that gives clues to that player's assessment of his hand.
In live poker at the very cheap levels, most of your opponents will seemed to have gone to the Jessica Simpson school of acting. For the robots, early in the hand they don't know how to maintain a poker face, let alone read the song like Christopher Walken:

Early in the hand, if a robot has a number of choices to make, it generates examples and determines what seems to be the best play. If the robot has limited or no choice, it does not need to do this, and can play fast. Hence robots can quickly play singletons, or from cards that are equivalently the same (e.g. 345, or if the 7 has been played, 568). Thus the speed the robot plays a card is a "tell".

Say you have KT9764 in your hand and AJ3 in dummy. Early in the hand, you lead the 9, LHR (left hand robot) plays small quickly, you win the ace, and RHR plays small slightly slow. Now you led the jack, and RHR hitches (i.e. slow plays), before playing small in the suit. What is the robot thinking? If he didn't have the queen, there would be no need for robot thought, so let the jack ride.

Important warning 1: the same "tell" does not work later in the hand, since the robot does not have much cards left for thinking - the robot can play in tempo, and is quite capable of smooth ducks.

Important warning 2: the programmers can (and should) remove the "tell" by adding appropriate pauses into the robot play.

Watch for the bot "tell", and remember the longer the bot thinks, the more it has to think about.

The longest "tell" I've seen in the bidding is after 1H(bot)-P-1NT (forcing)-P;-?. If the robot opener has exactly 4Ss and 5Hs, he will now take a long time to rebid, sometimes rebidding 2H, sometimes passing, and sometimes bidding a minor. Most other hands provide for easy rebids following the structure, but this hand type does not fit into the rebid structure provided to opener.

Next up: hoggin' the hand.

Monday, November 02, 2009

The Robot Battles Part I - Pay Attention

This is the first of a series of posts (some of them mini-posts) on battling the BBO Robots. There's some recent discussion on BBO robots on (gasp!) the BBO forums:

Adam Meyerson (awm on BBO) said it well here:

… the competition is between human players, who are sitting in the same seat and playing with equally good (or bad) robot partners. This is essentially a pure form of individual tournament, and it does measure the caliber of play of a particular human when compared to the other competing humans. In principle even if GIB played quite poorly, this would still be a fair event because the humans compete against each other.
The Robot tourneys are actually human tourneys, with humans competing against each other for masterpoints. These tourneys have allowed for individual events to return to the bridge scene, and given partnership complications everybody has, this would make bridge even more accessible to the public.

The most important success factor in the robot tourneys is to pay attention - count all the suits, watch all the discards, know what is going on as best as you can figure out. As an experiment, albeit one that I knew how it would likely turn out, on Saturday evening, I played in some tourneys, but at the same time I was handing out candies, watching a TV movie and hockey games, and doing online surfing. In short my results suffered by 10 to 30% (yikes!). This happened even though I was playing the hands slower due to all the interruptions.

The robots can be tricky - if one is doing a squeeze against most opponents in real life, you tend to come down to the squeeze card, there is a squirm as the opponents decide what to discard, and it is easy to read the end position. Against the robots, they tend to see a squeeze coming up a few cards before, and will dump an important card early, in case you are not paying attention.

As a result of this one factor, I will now use Robot tourneys both for preparing for live tourneys and, assuming net access, at the live tourneys themselves. For the prep work, I believe playing many hands in the few weeks before a tourney makes one sharper. If you are not warmed up, then you waste too much mental energy on rote hands as you are not accustomed to triaging your hands: which ones are rote, which one require some thinking, which ones are complex.

When I go to a live tourney I tend to be, well, brain dead in the morning. Even for the Robot tourneys, the first one of the day I would usually have some sort of horrendous mistake, before I started to pay proper attention. For the live tourneys I used to try to kibitz part of a morning event, in order to get my bridge brain back. However kibitzing is a slow process for a zombie.

Now the next time I go to a live tourney, I plan to play one or two robot tourneys a couple of hours before playing the live bridge. That will take me 25 or 50 minutes, but now when I later play the live bridge, I'm back to paying attention. It's one or two dollars of online bridge, that's like buying a cup of coffee to wake up!

And next up: the "tell" of the robots - how to make your finesses!