[Chamaco]

Rob F, on May 10 2007, 07:44 PM, said:

Although no one seems to have mentioned this now, I will note that Bergen raises, for their preemptive nature, seem a much bigger plus if you're playing a limited opening bid system like precision.  This way it's more often the opponents that you're preempting (at least compared to standard).  If you play light and limited openings, say 9-15, you can really steal the auction sometimes when the opponent's have game but their points are evenly divided and the first opponent didn't quite have a call over 1M and the second isn't brave enough to double 1M-3M.

If you play light limited openings (say 9-15), you can gain and steal much more by preempting 1M:3M more often using a wide-range preempt (say 0-7/8-) rather than resort to the mixed raise.

In the limited opening context, the need to differentiate the NON GF raises into 3 ranges (preemptive, mixed, and limit) is much less crucial than in "standard" systems.

Most times, the mixed raise in a "regular" 2/1 context is needed to allow opener with a reverse (or equivalent playing strength) to bid game, but in the light opening system you mention, this is seldom an issue.

With a mixed raise, one can choose between a constructive single raise and a wide-range preemptive double raise according to the quality of the hand and the state of vulnerability (as well as tactical consideration)

This makes you preempt much more often 1M:3M although it is debatable whether this is really an advantage (one might argue that the wide-range of 1M:3M will avoid more oftem the doubling of the partscore, but I am not so sure about this)

[bhall]

Gerben42, on May 9 2007, 03:14 PM, said:

I've heard many people don't like Bergen raises. My question is: Why not and what do you play instead as raise structure?

Most of us recognize that the Law of Total Garbage is wrong about as often as it is right. It's a crutch for the cases where you can't get a better idea of the shape and location of high cards in the other hands. So a better question about Bergen is:

Why sell your birthright for a mess of pottage?

On the other side of this question, S. J. Simon (and mgr777, above) notes that you only want to exchange enough information to select the best practical contract for your side. Anything beyond that only gives aid to the enemy. Relay systems can go some way toward accomplishing this goal, but they run too quickly into the Fibonacci limit.

A reasonable compromise would be to tell partner you have a fit at the lowest possible level whenever the degree of fit is critical to the game decision. And to just bid game when it isn't. The standard 1M-2M raise violates this rule, and Bergen raises do so in spades (pardon the pun).

The raise to 2M at least leaves room to explore, although the exploration can prove costly. I think a better approach would be to "tighten up" the raise to show 4-card support and use the next-step response (1♥-1♠ or 1♠-1N) as a raise (typically 3-card) that needs room, either to obtain a fuller description of opener's hand or to complete its own description. That usage would, of course, increase the burden on the other cheap responses.

[TylerE]

Gerben42, on May 9 2007, 03:14 PM, said:

I've heard many people don't like Bergen raises. My question is: Why not and what do you play instead as raise structure?

What if you are playing the "Improving 2/1 GF" so 2NT is not free for a fit-showing bid?

I like to play 3m over 1M (and 2S over 1H) as fit-jumps, showing 4+ support for partner (occasionally something like Axx or KJx) and a good side suit (typically 5+ cards). Usuaully shortness somewhere. 7(++) - 9HCP.

[whereagles]

Rob F, on May 10 2007, 07:44 PM, said:

Although no one seems to have mentioned this now, I will note that Bergen raises, for their preemptive nature, seem a much bigger plus if you're playing a limited opening bid system like precision.

Actually, no. It much more dangerous for 4th seat to overcall a bergen raise if opener is unlimited because his (potential) extra strength may be enough to wield the axe on overcaller.

But it is true that bergen raises' preemptive effect is a plus for the convention. It is dangerous to enter the bidding when responder told opener very clearly what he has. Opener has more info than anyone else at table.

[whereagles]

bhall, on May 17 2007, 03:25 AM, said:

Most of us recognize that the Law of Total Garbage is wrong about as often as it is right.

Vernes was aware of this. His claim was not "total tricks = total trumps", but rather "total tricks = total trumps, ON AVERAGE".

The 'on average' is the key. On average means TT=TT more often than anything else, say TT=TT+1 or TT=TT-1, but this doesn't mean TT=TT in 50%+ of the time

[Chamaco]

whereagles, on May 23 2007, 10:36 AM, said:

Rob F, on May 10 2007, 07:44 PM, said:

Although no one seems to have mentioned this now, I will note that Bergen raises, for their preemptive nature, seem a much bigger plus if you're playing a limited opening bid system like precision.

Actually, no. It much more dangerous for 4th seat to overcall a bergen raise if opener is unlimited because his (potential) extra strength may be enough to wield the axe on overcaller.

Indeed, it depends. Opposite a limited opener Precision-like, responder should, IMO, raise preemptively even holding a bad 7-8 count, leaving opps i the dark on whether the "signoff-raise" was made on junk or on a decent hand.

Such "wide-range" bid makes it harder to defend against, and it may help to get rid, opposite a limted opening, of the mixed raise at the 3-level: just decide whether the hand is not worth game (so bid preemptively), or invitational (use whatever limit raise you play), or, if in between, just go via a constructive raise(even if it promises only 3 trumps, opener has a chance to make a decent game try)

[hrothgar]

Chamaco, on May 23 2007, 03:21 PM, said:

Such "wide-range" bid makes it harder to defend against, and it may help to get rid, opposite a limted opening, of the mixed raise at the 3-level: just decide whether the hand is not worth game (so bid preemptively), or invitational (use whatever limit raise you play), or, if in between, just go via a constructive raise(even if it promises only 3 trumps, opener has a chance to make a decent game try)

I think that there is an error in your logic:

For the sake of argument, I'm going to set up a very simple example.
Assume the following:

We're playing a strong club system. A 1♠ opening in first / second seat shows 5+ Spades and 11 - 15 HCP.

Lets compare two different response structures:

Response structure 1: A raise to 3M shows 4 Spades and exactly 8 HCP

Response structure 2: A raise to 3M shows 4 Spades and 6 - 10 HCP

You seem to be claiming that its more difficult to guessing how one should compete over the second response structure compared to the first. If one uses the second response structure, the total strength of the two hands can be anywhere in between 17 and 25 HCPs. In comparison, if one uses the first response structure, the total strength of the two hands is limited to 19 - 23 HCP.

However, it might be helpful to look at some more data: I did a quick monte carlo simulation defining the frequency of total HCP strength using each raise structure. Notice that the data set for response structure 1 is much more flat than the data set in response structure two. Response structure two has a (relatively) sharp peak at 21 HCP.

Opener's expect strength is negatively correlated with responder's expected strength. Accordingly, adding the two numbers cancels out a lot of the variance in the system.

I'm not sure that one can argue that a wide ranging bid opposite a wide ranging bid is harder to defend against than tightly defined bid opposite a wide ranging bid.

Response structure 1:

19 HCP = 2413 hands
20 HCP = 2334 hands
21 HCP = 1957 hands
22 HCP = 1678 hands
23 HCP = 1325 hands

Response structure 2:

17 HCP = 1661 hands
18 HCP = 3806 hands
19 HCP = 5884 hands
20 HCP = 8041 hands
21 HCP = 9390 hands
22 HCP = 7054 hands
23 HCP = 4738 hands
24 HCP = 2881 hands
25 HCP = 1090 hands

[Chamaco]

hrothgar, on May 23 2007, 01:30 PM, said:

Lets compare two different response structures:

Response structure 1:  A raise to 3M shows 4 Spades and exactly 8 HCP

Response structure 2:  A raise to 3M shows 4 Spades and 6 - 10 HCP

You seem to be claiming that its more difficult to guessing how one should compete over the second response structure compared to the first.

I claim that - in a strong context - 1M-3M "preemptive" should be 0- to a bad 8 (basically any hand that responder judges unworthy to make game opposite a maximum 1M opener-say a good 5-5 opening hand), and this range is quite different from "standard" Bergen, where many play, given the unlimited nature of the opening, 0-4, 0-5 or 0-bad 6 according to how sound the opening bids are


With a "good 7-8 hcp" I'd bid a constructive 1M-2M, or even 1M:2NT if a very good 8 hcp (say, 2 aces and a couple of assets such as shortness or some T9s)

The total hcp by opener's side after a 1M-3M "wide range preempt" would be 11-23

I am sorry if my previous post was unclear about this, sometimes I write too fast and english being not my language, the message gets messy for another reader

Having said that, I must confess that it is not easy for me to interpret the tabulated data you show, and even more difficult is to understand whether - in light of this post of mine - they reject my hypothesis or not....

Maybe you can explain in an easier form for me ? Thanks !

[jtfanclub]

hrothgar, on May 23 2007, 08:30 AM, said:

Response structure 1:

19 HCP = 2413 hands = 25%
-----------------------------------
20 HCP = 2334 hands = 24%
21 HCP = 1957 hands = 20% +/-1=61.5%
22 HCP = 1678 hands = 17.5%
------------------------------------
23 HCP = 1325 hands = 13.5% +/-2=100%

Response structure 2:

17 HCP = 1661 hands= 4%
18 HCP = 3806 hands= 8.5%
--------------------------------------
19 HCP = 5884 hands= 13%
---------------------------------------
20 HCP = 8041 hands= 18%
21 HCP = 9390 hands= 21% +/-1= 55%
22 HCP = 7054 hands= 16%
---------------------------------------
23 HCP = 4738 hands= 10.5% +/-2= 78.5%
---------------------------------------
24 HCP = 2881 hands= 6.5%
25 HCP = 1090 hands=2.5%

I put some math in the quote above

On the tight auction, if the opponent assumes you that the pair has 21 hcp, he is 20% likely to be dead on, within 1 (20-22) 61.5% of the time, and within 2 (19-23)100% of the time.

On the lose bid, if the opponent asssumes the pair has 21 hcp, he is 21% likely to be dead on, within 1 (20-22) 55% of the time, and within 2 (19-23) 78.5% of the time.

Being off by more than 2 on your estimate over 20% of the time is a lot. It's very difficult to come in at the 3 level when it's that unclear how many HCP the opponents have. I think if you do it for 0-8, where the opponents could have a combined point count anywhere from 17 to 29 hcp, and an average of about 22, you'll see just how very difficult it is to come in at the 3 or 4 level over that.

[SteelWheel]

I used to play Bergen Raises a lot....it was another one of those "crutch" conventions that helped a lot when I was an intermediate player playing against a lot of other intermediates.

These days, I'm not sure how helpful it really is. The idea of "quickly getting to the three level with a nine-card fit" is somewhat overrated, I think. The only time this really helps is when you open the bidding in 1st seat--assuming LHO passes, pard is able to make a preemptive action (via some kind of 3-level bid), and thus deprive RHO of some bidding room to make his butt-in lead-directing/save-suggesting call.

However: If we're in 2nd seat, open 1M, LHO passes, and pard makes some Bergen-ish noise, it would appear that the chance of RHO wanting to make a call of any kind go way down, I would think--RHO already rejected a chance to bid once, after all. How likely is he to want to bid on the 3-level, if he couldn't previously bid on the 1- or 2-level?

These days, I'm more taken with Fred's "Improving 2/1" concepts for raises as it allows for a natural 2NT response to 1M--an idea I believe has far more merit than current styles allow for.

Another thought: Many Precision (or other Strong Club) players use Bergen Raises--including Berkowitz/Cohen. I play them myself in a couple of Precision partnerships, just out of inertia--it's what we've played for many years, too much effort to change now since we don't play so often, etc. But I'm sure this is not theoretically best. After all, opener is "never" going to be able to go and shoot out a game over a 1M/3M auction. Wouldn't it be better to define the various jumps to the 3-level as different kinds of limit raises (e.g., balanced, shortness in one of the other suits, etc.)? I've toyed with this one for a long time--but again, I don't play seriously that much anymore, so haven't invested the effort to devise such a scheme.

[rbforster]

hrothgar, on May 23 2007, 08:30 AM, said:

I did a quick monte carlo simulation defining the frequency of total HCP strength using each raise structure. Notice that the data set for response structure 1 is much more flat than the data set in response structure two. Response structure two has a (relatively) sharp peak at 21 HCP.

Opener's expect strength is negatively correlated with responder's expected strength. Accordingly, adding the two numbers cancels out a lot of the variance in the system.

Thanks for the interesting example and simulation!