[wcneumann]

Based on many thousands of hands, I'm 99.5% convinced that the BBO deals I'm receiving in the ACBL daily and instant games are not even close to being random. Bad and really bad suit splits are much more frequent than my lifetime of bridge would suggest, and key honors are offside much more often than randomness would dictate.

Many of you have played random deals in tournaments and club games.

What is your impression? I wouldn't mind if they weren't random if BBO would so state, but I suspect they are distributing "goulash" type hands to some or all participants.

[smerriman]

This has been done to death. They're random. You're wrong.

[johnu]

You need to join the premium BBO Super Platinum club. Hands with bad splits and offside honors go to the ordinary BBO players and BBO doesn't care if those players get upset. BBO takes really good care of their Super Platinum players.

[hrothgar]

No

On so very many levels

No

[thepossum]

It needs another option "What do you mean by random?"

Sometimes I am certainly in the camp that thinks they are not random enough

[wcneumann]

 smerriman, on 2020-October-01, 22:57, said:

This has been done to death. They're random. You're wrong.

I'm quite willing to believe that they create deals that are random--in fact I'd be surprised if it were otherwise. What I'm questioning is whether EVERY user is randomly assigned ACBL Instant game sets and ACBL Daily Game hands. It would be a trivial bit of programming to select hands or 12-board sets that are something other than random. Once identified, one could easily assign such selections to a player or group of players based on any of a variety of metrics. If one did that, it would only affect a small number of players and one could, with a clear conscience state that the deals are "random."

My experiences over my several tens of thousands of hands suggest that something like this may be occurring--at least for some of us.

[hrothgar]

 wcneumann, on 2020-October-02, 05:42, said:

I'm quite willing to believe that they create deals that are random--in fact I'd be surprised if it were otherwise. What I'm questioning is whether EVERY user is randomly assigned ACBL Instant game sets and ACBL Daily Game hands. It would be a trivial bit of programming to select hands or 12-board sets that are something other than random. Once identified, one could easily assign such selections to a player or group of players based on any of a variety of metrics. If one did that, it would only affect a small number of players and one could, with a clear conscience state that the deals are "random."

My experiences over my several tens of thousands of hands suggest that something like this may be occurring--at least for some of us.

If you are willing to jump down that sort of rabbit hole and claim that BBO is specifically trying to victimize you (or other selected individuals)... well, it's hard to know what to say. Using that sort of logic you can go off, cherry pick data, and pretty much justify whatever stupid theory you want.

It's not even worth doing any kind of statistical analysis of your board results, because you can simply say "Well, BBO wasn't tweaking my hands on this specific set of hands".

[wcneumann]

 hrothgar, on 2020-October-02, 06:10, said:

If you are willing to jump down that sort of rabbit hole and claim that BBO is specifically trying to victimize you (or other selected individuals)... well, it's hard to know what to say. Using that sort of logic you can go off, cherry pick data, and pretty much justify whatever stupid theory you want.

It's not even worth doing any kind of statistical analysis of your board results, because you can simply say "Well, BBO wasn't tweaking my hands on this specific set of hands".

I didn't say or claim they were trying to victimize me. What I said was that my experiences over several tens of thousands of deals is that they are not as random as they should be. When I play in their live games, I do not see the same behavior. What I did say is that it would be very easy to distribute non-random boards to targeted individuals or groups of individuals for any of a variety of reasons.

In fact, doing so could be something that one might do to increase revenue. Distributing non-random deals or sets could increase revenue by allowing less capable players to do well when they miss a difficult to reach, but great contract that's going to fail because all three key cards are offside, or the trump suit breaks 5-0.

As I said originally, I don't object do them doing things like this to strengthen their business. But, if they are doing these types of things, I do object to them not letting their paying customers know.

If I get ambitious, I may investigate whether I can gather the data and do an analysis of sets and deals I've had to play over whatever period I can.

Your suggestion that the data from such an analysis could be discounted is specious. Data is data. Either the deals follow a random distribution or they do not.

My level of my play is high enough, and I play more than enough on BBO to have a pretty good sense about whether the deals I'm playing are random or not. I'm curious how wide spread the perception may be...

[hrothgar]

 wcneumann, on 2020-October-02, 06:58, said:

I'm curious how wide spread the perception may be...

The perception is incredibly wide spread, however, no one has ever been able to demonstrate any kind of compelling evidence.

Quote

My level of my play is high enough, and I play more than enough on BBO to have a
pretty good sense about whether the deals I'm playing are random or not.

I doubt that this is true. Not because I have any opinion regarding your skill at bridge or the number of hands that you play, but rather than researchers have demonstrated time and time again the humans are incredibly bad at judging whether or not data is random and (conversely) that they are so good at pattern recognition that they invent patterns in data that don't really exist.

However, I'll make the same offer to you that we do to everyone else who comes here with some random claim.


Bring us a hypothesis.
Explain how you believe that the hands that you are playing are biased.

After which, we'll look at the next week or so worth of hands that you play and see whether there is an statistical validity to your claim.

[hrothgar]

 wcneumann, on 2020-October-02, 06:58, said:

Your suggestion that the data from such an analysis could be discounted is specious. Data is data. Either the deals follow a random distribution or they do not.

A hypothesis that is generated after you look at a set of data needs to be treated very differently than a hypothesis that is generated before you collect a set of data.

There are a whole lot of ways in which data can exhibit bias.
Give me ANY set of data.

I can create some hypothesis that is consistent with bias.

[armantt2k]

 wcneumann, on 2020-October-01, 21:14, said:

Based on many thousands of hands, I'm 99.5% convinced that the BBO deals I'm receiving in the ACBL daily and instant games are not even close to being random. Bad and really bad suit splits are much more frequent than my lifetime of bridge would suggest, and key honors are offside much more often than randomness would dictate.

Many of you have played random deals in tournaments and club games.

What is your impression? I wouldn't mind if they weren't random if BBO would so state, but I suspect they are distributing "goulash" type hands to some or all participants.

I play in 20-30 games per month in VACB & ACBL games for ACBL MPs. I have no problems with deals in those games. But I agree with you when it comes to BBO dealt hands.

NOTE: I do not play in BBO point games. I am only reporting my experience when my partner and I practice with robots on Casual-Relaxed-MP tables.

We have been playing 18-36 boards per day, almost daily since the pandemic started, against 2 basic robots at a BBO Casual table. In those 12,000+ hands, a conservative back-of-envelope estimate (20 hands x 25 days/mo x 24 mos) we all too often get 4-1 and 5-0 trump splits (way more than the expected 28% + 4% of the time). Missing honors are also offside much more than 50% of the time.

I willingly concede that we are not expert players, not even close. But when we windup in the wrong contract (6-card trump suit, 19-hcp game bids), we are able to recognize it's our own damn fault. That is NOT the issue here.

We have played entire sessions without a single 'normal' 3-2 break in trump or a runnable suit in NT, or a single finesse being onside. It has reached the point where we are genuinely surprised if "things goes right" (i.e. 'playing the odds' actually works) on more than 2 or 3 hands in a session.

Like you, I'd be curious to hear from others - but even more, I'd actually like to hear from someone who has somehow captured all those deals and done the analysis. Until then, I can only say "Prove me wrong!"

[armantt2k]

Is the same software used to deal the hands used for the Casual/Bidding/Teaching tables as for the BBO ACBL Instant & Daily games?

[smerriman]

 armantt2k, on 2021-December-12, 21:37, said:

Like you, I'd be curious to hear from others - but even more, I'd actually like to hear from someone who has somehow captured all those deals and done the analysis. Until then, I can only say "Prove me wrong!"

Done this multiple times in the past on the forum.

But fine. Go to MyHands, extract your hands for the last 6 months. Copy and paste resulting source code.

Parse source code and only extract the MBC rows. Restrict rows to those with you, glotook, and two GIBs (and no passouts). 848 hands found.

Ignore no trump hands. 666 hands found.

Reduce hands to where you declared, since no doubt the bias is against you, not the bots. 388 hands found.

Reduce hands to where you had an 8 card fit. 174 hands found.

Count trumps in the opponents' hands. Results:

8-card fit: 118 3-2 breaks, 49 4-1 breaks, 7 5-0 breaks.

Which is 67.8%, 28.2%, and 4.0%.

There you go, proven wrong. If you suspect I'm making up these numbers, I'll happily send you links to the 174 hands so you can count them yourself.

Here are the 5-0s as a taster.

I could repeat it for finesses, but a) any finesse that's offside for one side is onside for the other, so it will work out to 50% even if there is bias, b) most people think they are losing lots of finesses because they take lots of finesses in dummy straight after the opening lead. This is nearly guaranteed to lose, due to fact that GIB leads passively, rarely from an honor. This isn't bias, but conditional probability.

[9NTredbl]

Of course they are random. All of bidding and play theory depends on this.

[pigpenz]

GIB sits around at night and picks the hands to be played

[johnu]

 pigpenz, on 2022-January-05, 13:35, said:

GIB sits around at night and picks the hands to be played

In fairness to GIB, the easy part is picking the hands to be played. The difficult part is figuring out which players deserve to get the bad breaks and lousy cards, and who should get all the good breaks, working finesses, and almost all the opening bids. Obviously the Super Double Gold Latinum Players Club members are going to the best hands with good breaks and working finesses. Everybody else has to go through a special program to determine their worthiness to get average or better hands.

[pilowsky]

It's a great question from someone called Neumann.