[thepossum]

Dear all

Are BBO hands in clubs random or specially selected according to certain criteria. They certainly don't seem to fit the pattern of randomly dealt hands

regards P

[pescetom]

I can't answer the question, but beware about assuming that randomly dealt hands have a pattern or are easily identifiable in series. Randomly dealt hands are on average more "interesting" than manually dealt hands precisely because they are random and not influenced by the preceding order of the pack and dealing sequence.

[barmar]

All hands dealt by BBO are random. In "best hand" robot tournaments the South hand is swapped with the hand with the highest HCP after the random dealing is done.

What do you think the "pattern of randomly dealt hands" should be? By definition, random hands should have no pattern. People tend to have misconceptions about random sequences.

https://fs.blog/2015...ions-of-chance/

[The_Badger]

About a year ago I asked the same question of BBO hands, given how computer hands seem more distributional, etc. And guess what? I did my own analysis of a range of hands, over 100, compared them to statistical charts of frequency, and found that they virtually matched given a range of +/-1%. So it's fair to say that they are randomly dealt.

[barmar]

As I've mentioned in other previous discussions about this, we asked the author of BigDeal to analyze our hands. He said that they look OK as well. We do admit that our code is not as good as BigDeal. We've considered integrating it into our system, but haven't yet figured out a good way to do it. However, we consider it difficult to game our system because we're generating hands for so many different tables and tourneys at the same time, so even if you can figure out the sequence you can't tell where the hands will go.

[pescetom]

People who play at home or in clubs have problems believing random deals because they are used to manual deals, which tend more towards being balanced and having about 10 HCP. In random deals only 32% of hands are truly balanced (4333 or 4432) and another 15.5% semi-balanced (5332) - so it's actually more likely (52.5%) that you get an unbalanced hand instead.

[barmar]

All the clubs around here have been using computer-dealt hands for years. At tournaments we only deal by hand in Swiss Teams or KO.

[pescetom]

barmar, on 2018-October-26, 13:24, said:

All the clubs around here have been using computer-dealt hands for years. At tournaments we only deal by hand in Swiss Teams or KO.

Lucky you! Only the biggest clubs here can afford a dealing machine, so hand preparation is mainly manual even when the deal itself is by computer. Typically clubs play one or more national simultaneous tourmanents every week and the hands are either prepared by the Director or (for less important tournaments) by the players themselves in a non-playing first round. Other local tournaments and Teams are shuffled and dealt by hand.
There is an interesting alternative to traditional dealing machines which might catch on easily here if it becomes mature.
Of course with tablet based play looming, the whole issue may become academic.

[hrothgar]

thepossum, on 2018-October-25, 04:30, said:

Dear all

Are BBO hands in clubs random or specially selected according to certain criteria. They certainly don't seem to fit the pattern of randomly dealt hands

regards P

Your ideas are intriguing to me and I wish to subscribe to your newsletter.

[hrothgar]

barmar, on 2018-October-26, 09:16, said:

As I've mentioned in other previous discussions about this, we asked the author of BigDeal to analyze our hands. He said that they look OK as well. We do admit that our code is not as good as BigDeal. We've considered integrating it into our system, but haven't yet figured out a good way to do it. However, we consider it difficult to game our system because we're generating hands for so many different tables and tourneys at the same time, so even if you can figure out the sequence you can't tell where the hands will go.

The original poster felt that your RNG was biased.
You response is talking about predictability.

Hard to understand how it could be difficult integrate BigDeal.

1. Seed a high quality PRNG
2. Start reading off big numbers
3. Convert big numbers into hands
4. Store hands in a buffer

Reseed every now and then

[barmar]

hrothgar, on 2018-October-27, 10:59, said:

The original poster felt that your RNG was biased.
You response is talking about predictability.

Yeah, I was just thinking about all the potential problems that a poor dealing algorithm could have if it's not really random.

Quote

1. Seed a high quality PRNG
2. Start reading off big numbers
3. Convert big numbers into hands
4. Store hands in a buffer

There were some Linux compatibility issues we ran into. It also requires operator input to generate the initial entropy, we'd want to automate it using something like /dev/random. And we have to ensure that it can keep up with our need for new hands.

We had some email discussions about it a couple of years ago, but it's never seemed important enough to actually replace what we already have.