[enigmisto]

I'm currently reading the BFUN system book, and it uses a QP (quality point) scale for unbalanced hands: A=3, K=2, Q=1.

Unfortunately, the book doesn't describe what the QP targets should be for game and slam, or how to categorize hands as minimum / invitational / maximum. It says things like, "Jump bid with a maximum hand" but it gives no guidelines as to what constitutes maximum.

I can come up with some basic estimates by taking the HCP system and multiplying the numbers by 3/5, but that seems like it defeats some of the value of using this custom system which ignores jacks. Presumably if QP has independent validity from HCP, someone's done some computer simulations and come up with good thresholds for QP, which I imagine would be close to HCP * 3/5, but not quite the same. I'd like to know what the guidelines are for "thinking in QP".

Any suggestions?

[hrothgar]

This is a topic that I explore in my intro to symmetric relay.
https://www.dropbox....0Relay.pdf?dl=0


The main tool that I use to explore this is Hans van Stavern's Dealer.
I find it useful to experiment with scripts like the following


produce 1000
predeal south SAK3,HAT,DAK532,CT97

# Define slam points

slam_points_north =

3*(hascard(north, AS) + hascard(north, AH) + hascard(north, AD) + hascard(north, AC)) +
2*(hascard(north, KS) + hascard(north, KH) + hascard(north, KD) + hascard(north, KC)) +
1*(hascard(north, QS) + hascard(north, QH) + hascard(north, QD) + hascard(north, QC))

- 2 * (hascard(north, KS) and spades(north) == 1)
- 2 * (hascard(north, KH) and hearts(north) == 1)
- 2 * (hascard(north, KD) and diamonds(north) == 1)
- 2 * (hascard(north, KC) and clubs(north) == 1)

- (hascard(north, QS) and spades(north) == 1)
- (hascard(north, QH) and hearts(north) == 1)
- (hascard(north, QD) and diamonds(north) == 1)
- (hascard(north, QC) and clubs(north) == 1)

condition shape(north, 5431) and slam_points_north == 6

action

frequency "Spades" (tricks (south, spades), 0, 13),
frequency "Diamonds" (tricks (south, diamonds), 0, 13)

[enigmisto]

Thanks, that's interesting material! The link in your document (http://dealergib1.br...aler/dealer.php) seems no longer to be valid. Do you happen to know where this tool currently is accessible?

[johnu]

enigmisto, on 2021-June-14, 05:41, said:

I'm currently reading the BFUN system book, and it uses a QP (quality point) scale for unbalanced hands: A=3, K=2, Q=1.

Unfortunately, the book doesn't describe what the QP targets should be for game and slam, or how to categorize hands as minimum / invitational / maximum. It says things like, "Jump bid with a maximum hand" but it gives no guidelines as to what constitutes maximum.

Assuming what you say is correct, it sounds like a lousy book. I would toss the book in the trash or to the bottom of my obsolete book bin.

[pescetom]

enigmisto, on 2021-June-14, 14:31, said:

Thanks, that's interesting material! The link in your document (http://dealergib1.br...aler/dealer.php) seems no longer to be valid. Do you happen to know where this tool currently is accessible?

In a related discussion about a month ago, BBO explained that the server involved had been turned off and some solution (such as turning the server back on, or replacing the basic dealer with the enhanced dealergib version) would be evaluated.

[pilowsky]

hrothgar, on 2021-June-14, 06:56, said:

This is a topic that I explore in my intro to symmetric relay.
https://www.dropbox....0Relay.pdf?dl=0


The main tool that I use to explore this is Hans van Stavern's Dealer.
I find it useful to experiment with scripts like the following


produce 1000
predeal south SAK3,HAT,DAK532,CT97

# Define slam points

slam_points_north =

3*(hascard(north, AS) + hascard(north, AH) + hascard(north, AD) + hascard(north, AC)) +
2*(hascard(north, KS) + hascard(north, KH) + hascard(north, KD) + hascard(north, KC)) +
1*(hascard(north, QS) + hascard(north, QH) + hascard(north, QD) + hascard(north, QC))

- 2 * (hascard(north, KS) and spades(north) == 1)
- 2 * (hascard(north, KH) and hearts(north) == 1)
- 2 * (hascard(north, KD) and diamonds(north) == 1)
- 2 * (hascard(north, KC) and clubs(north) == 1)

- (hascard(north, QS) and spades(north) == 1)
- (hascard(north, QH) and hearts(north) == 1)
- (hascard(north, QD) and diamonds(north) == 1)
- (hascard(north, QC) and clubs(north) == 1)

condition shape(north, 5431) and slam_points_north == 6

action

frequency "Spades" (tricks (south, spades), 0, 13),
frequency "Diamonds" (tricks (south, diamonds), 0, 13)

Hans van Staverens dealer can be found here: https://www.bridgeba...aler/dealer.php
FWIW, I just tried copying the script above into it and it reported a "syntax error".
Here is the output of the small default script:

#
# Sample
#
produce 5
condition shape(north, any 4333 + any 4423) and hcp(north)>=13
action printoneline

which produces:

n AJ75.Q7.AKQ7.654 e KT8.98542.JT9.32 s Q2.AK.5432.AK987 w 9643.JT63.86.QJT 
n K94.Q962.AJT2.QJ e T32.T75.Q83.K865 s QJ75.J8.976.AT94 w A86.AK43.K54.732 
n K86.AK7.A96.JT62 e QJ7.QT43.JT7543. s .982.Q8.AKQ98754 w AT95432.J65.K2.3 
n KJT5.K5.KQ6.QT82 e Q986.A97.J854.K5 s 742.QJT2.A973.43 w A3.8643.T2.AJ976 
n KQ74.KJ2.KT9.AJ6 e 32.T63.AQ53.8542 s A95.754.862.KT93 w JT86.AQ98.J74.Q7 
Generated 51 hands
Produced 5 hands
Initial random seed 1623716891
Time needed    0.000 sec

It also works OK with pescetom's code from an earlier post about 1NT openings.

Is there a difference between the code format in the original site and the one above?

[PrecisionL]

I have been practicing the BFUN System with two partners for about a month and like most of the design.

QP evaluation is a paradigm shift and takes some practice. Together with known distribution it is powerful for slam bidding and sometimes to pick the best game.

However, the drawback is that sometimes the honor location can NOT be determined low enough in the bidding.

[foobar]

enigmisto, on 2021-June-14, 05:41, said:

I'm currently reading the BFUN system book, and it uses a QP (quality point) scale for unbalanced hands: A=3, K=2, Q=1.

Unfortunately, the book doesn't describe what the QP targets should be for game and slam, or how to categorize hands as minimum / invitational / maximum. It says things like, "Jump bid with a maximum hand" but it gives no guidelines as to what constitutes maximum.

I can come up with some basic estimates by taking the HCP system and multiplying the numbers by 3/5, but that seems like it defeats some of the value of using this custom system which ignores jacks. Presumably if QP has independent validity from HCP, someone's done some computer simulations and come up with good thresholds for QP, which I imagine would be close to HCP * 3/5, but not quite the same. I'd like to know what the guidelines are for "thinking in QP".

Any suggestions?

My rough rule of thumb is that you need about 19 QPs for a slam in a suit, but should explore with 18 and a 9-card fit. BTW, my recommendation is to use symmetric relays for the sheer mnemonic ease; everything else is so not worth it. There are many variants, including transfer based (example: TOSR), but the underlying principle is the same.

[enigmisto]

johnu, on 2021-June-14, 15:33, said:

Assuming what you say is correct, it sounds like a lousy book. I would toss the book in the trash or to the bottom of my obsolete book bin.

In fairness to the book, I think one the author's main ideas is that game/slam/etc for unbalanced hands depends a lot more on distributional factors and fit than points. So I suppose the idea is that QP gets you started with opening, but then you're using judgment beyond that. I suspect the author is rather intentionally refraining Still, I haven't developed that kind of judgment yet and need some help figuring out how to use this QP measure effectively to gauge whether I'm in range. The document hrothgar shared has some data that provides a good start for me.

[hrothgar]

pilowsky, on 2021-June-14, 18:33, said:

Is there a difference between the code format in the original site and the one above?

Sorry, I guess that BBO tweaked their Dealer code. They seem to do so from time to time.

It's been ages since I downloaded the standalone executable, however, there are definitely versions that support the "tricks" function.

[pescetom]

enigmisto, on 2021-June-14, 21:35, said:

In fairness to the book, I think one the author's main ideas is that game/slam/etc for unbalanced hands depends a lot more on distributional factors and fit than points. So I suppose the idea is that QP gets you started with opening, but then you're using judgment beyond that. I suspect the author is rather intentionally refraining Still, I haven't developed that kind of judgment yet and need some help figuring out how to use this QP measure effectively to gauge whether I'm in range. The document hrothgar shared has some data that provides a good start for me.

I always imagined that QP was more about honour distribution than strength, where it would make little sense as johnu argues. Given QP and the parity of Kings one can place the honours. But maybe that's only a small part of it.

[hrothgar]

I can not emphasize the following enough: You need to consider your bidding system holistically (and in turn, you can't consider things like slam points in isolation)

If a bidding system is using QP points or counting controls early, then this almost certainly shows up again in the auction termination section.

For example, in the case of MOSCITO, if the relay captain is asking about slam points, it means that

1. They already know partner's shape
2. They are interested in exploring slam
3. They are going to launch into a denial cue bidding sequence (presuming that there are sufficient controls)

[pescetom]

pilowsky, on 2021-June-14, 18:33, said:

Hans van Staverens dealer can be found here: https://www.bridgeba...aler/dealer.php
FWIW, I just tried copying the script above into it and it reported a "syntax error".

I just tried copying the script above into it and it reported: "can't open temporary file".
Which is what happens when dealer tries to invoke GiB to calculate tricks and fails (see the previous thread).

pilowsky, on 2021-June-14, 18:33, said:

Is there a difference between the code format in the original site and the one above?

Not to my knowledge, but the "tricks" function will only work with the 'dealergib' version.
Hence my suggestion to substitute the current version with the 'dealergib' version.
Or if this is considered difficult or risky, turn the related server back on again.

[pilowsky]

removed

[PrecisionL]

Quote from "Love Story" and it is not true. (Love is not having to say you are sorry?)

[DinDIP]

enigmisto, on 2021-June-14, 05:41, said:

I'm currently reading the BFUN system book, and it uses a QP (quality point) scale for unbalanced hands: A=3, K=2, Q=1.

Unfortunately, the book doesn't describe what the QP targets should be for game and slam, or how to categorize hands as minimum / invitational / maximum. It says things like, "Jump bid with a maximum hand" but it gives no guidelines as to what constitutes maximum.

I can come up with some basic estimates by taking the HCP system and multiplying the numbers by 3/5, but that seems like it defeats some of the value of using this custom system which ignores jacks. Presumably if QP has independent validity from HCP, someone's done some computer simulations and come up with good thresholds for QP, which I imagine would be close to HCP * 3/5, but not quite the same. I'd like to know what the guidelines are for "thinking in QP".

Any suggestions?

A helpful rule of thumb is the rule of 21: add the two hands' QPs and count working voids as 3, working singletons as 2 and working doubletons as 1. (These are the high cards that are not needed to provide control.) If the total is 21 or more then slam is worth investigating.

The most important thing to do is to bid lots of hands so as to get a feel for how to envisage possible slams. The next step is to think about whether it is possible in the available bidding space below the safety threshold to find out what you need to know. If in doubt remember that honours are more likely to be in longer than shorter suits (another application of vacant spaces).

While it uses 2-1 controls rather than QPs, the hundreds of examples in KK Relay and the commentary from the authors (Karen McCallum and Kit Woolsey) provide the best guide yet to what to think about in relay slam bidding. Inevitably, there are a number of DD auctions where the authors recommend bidding on and partner has the right hand when I think good relayers bidding just one hand would sign off.

[DinDIP]

hrothgar, on 2021-June-15, 03:59, said:

Sorry, I guess that BBO tweaked their Dealer code. They seem to do so from time to time.

It's been ages since I downloaded the standalone executable, however, there are definitely versions that support the "tricks" function.

If you download the executable then you need to have mdealer.exe (not dealer.exe) and bridge.exe (GIB's DD engine) in the same directory. Then the tricks and score functions work. (Thanks to Wayne Burrows for helping me sort out that I needed to use mdealer.exe rather than dealer.exe.)