[FrancesHinden]

Deanrover, on Nov 14 2005, 02:59 PM, said:

Do you have any idea how much the edge a team of 4 experts would have over 4 average players in terms of IMPs/board? Final piece in the jigsaw!

It depends what you mean by "experts", and it depends what you mean by "average players". If you take the entire population that plays bridge sometimes, including those who only play 'for fun' (with friends or online, never in a tournament) then an 'average' player will be worse than your regular club duplicate attendee.

However, you'd never find a team of experts playing against such a team in a serious match. There's still a huge gulf between an 'average' player and an expert, and between an expert ("someone who wins national events") and a Bermuda Bowl finalist.

So let's think about the 'average' club player. Well, take my county's knock-out teams of four competition. With a paid up membership of about 2500, usually 25-30 teams actually enter. The draw is slightly seeded (there are four seeds put in the four quarters, other than that it is random). My team expects to win its first round match easily over 32 boards.

Against a team of 'average club players' from my locality, I would expect to win a 64-board teams-of-four match basically 100% of the time, and if we bothered to play all the boards with no change in tactics by either side I reckon I'd by up by over 100 imps at the end more than 50% of the time. On average I reckon to be about +1 imp/board against that sort of team.

In an 8-board Swiss Teams match, I have won against 'average club players' by 60+ imps a few times. I've also lost, perhaps one match in 10 or so (difficult to estimate, because many of the Swiss Teams matches I've lost have been against rather better than average players, that's the point of the Swiss tournament). If the Italian Open Team could be bothered to play, I'd expect them to lose an 8-board match maybe one time in 20.

But then looking at it the other way, I'd expect my team to lose to the Italian Open Team over 64 boards at least 90% of the time. 64 boards is actually an awful lot. An 8-board Swiss match, I'd reckon I had a 30% chance of winning, particularly if playing win/loss rather than VPs (i.e. if I had to play for a win). These numbers may be wildly out of course, I've never played them, only watched.

It's quite hard to estimate 'imps per board' - I was thinking about the multiple teams events I've played in, but the imps per board tend to be very high, as it's known you need a large score to win so the decent teams play a very volatile game (you usually need a net over over +100 over 24 boards to win that type of event). I've played in a few club multiple teams events, and average about +3 imps/board in an evening.

[helene_t]

hrothgar, on Nov 14 2005, 11:45 PM, said:

SoTired, on Nov 14 2005, 11:31 PM, said:

Interestingly, one of the ways that Las Vegas combats blackjack "counters" (professionals) is by having NON-RANDOM shuffles.[....] If the deck is deliberately non-random shuffled,then the professional can no longer rely on probability as his main evaluation tool.

I'd be VERY interested in seeing documentation to this effect.

Intuitively, this seems flawed. Professional Blackjack counters operate but estimately the 10s density in the deck. Non random shuffling simply transforms the game from simple counting into

I know X
I know you know X
I know you know I know X

Its unclear that this game would give a casino any better edge.

I can believe that a Casino might try seeding decks to detect card counters, but this is another story. I also believe that casinos might want to deliberate back load the 10s and combine this with a low penetration deck, however, this would probably not get by the gambling board.

I'm pretty sure that what the casinos are actually doing is that they build the decks simulating sampling with replacement, or, if you like, an infinity of decks.

Suppose (for simplicity) that you use only one 52-cards deck for blackjack. The conditional probability of drawing a 7 given that the previous card was a seven, is 3/51. Now we bias the draw so that a 7 is drawn with probability 1/13. Since each of the three remaining 7s now have a higher probability than any other card, the draw is, in a certain sense, less random than normally. However, ignoring the suit of the card drawn, the entropy is actually higher than before, so you could also say that the randomness is higher.

Such a method is probably legal if proberly announced.

[hrothgar]

helene_t, on Nov 15 2005, 03:13 PM, said:

I'm pretty sure that what the casinos are actually doing is that they build the decks simulating sampling with replacement, or, if you like, an infinity of decks.

This is (was???) quite common practice five years back. Many casinos were deploying "continuous shuffling" machines. They would start with a 5-6 deck shoe. At the end of each hand, all the cards were fed into a hopper. Each card was randomly placed inside the shoe. Any such system renders counting almost completely useless.

I heard rumors that many of these machines were actually being removed. A variety of explanations were offered ranging from wear and tear on the cards to mechanical problems. My own personal suspicion is that the casinos might be treating Blackjack as a loss-leader. The owners are willing to tolerate occasional losses to good card counters in order to draw more people into the casino. In particular, bad card counters could be a very lucrative business opportunity.

[SoTired]

The casinos normally use 6-8 decks and shuffle about 1/2 way thru. The large number of decks lowers the chance that the deck will ever go positive for the player.

To detect counters, the casinos just have to notice a player varying his/her bet. That is the tip-off.

Several years ago some teams of pros developed the system of a "counter" at the table making small bets. When the deck turned positive, he/she would signal a teammate and that player would come and play with larger bets until the counter signalled that the deck had turned negative again.

This system soaked the casinos. The casinos eventually ID's the teams. Part of their response to combat the teams was changing the shuffle. The casinos use a specific method of shuffling multiple decks. They changed their method several years ago that produces the non-random shuffle to combat counters. I think one team went thru a large losing streak and became more brazen and obvious and that is how they were spotted.

[hrothgar]

SoTired, on Nov 16 2005, 12:16 AM, said:

This system soaked the casinos. The casinos eventually ID's the teams. Part of their response to combat the teams was changing the shuffle. The casinos use a specific method of shuffling multiple decks. They changed their method several years ago that produces the non-random shuffle to combat counters. I think one team went thru a large losing streak and became more brazen and obvious and that is how they were spotted.

I'm an MIT grad and I knew some of the members of the second generation Blackjack team. I've heard a number of stories back in "the day". A lot of them revolved arround sex, several revolved arround drugs, and there was lots of booze. VERY few stories focused on non-random shuffles. I heard some claims regarding shuffle tracking, however, I never game them much credit. The "core" of the MIT Blackjack curiculeum focused on counting, counting, avoiding detection, and counting.

Step back and consider how a non-random shuffling system would need to behave in order to create an advantage for the house. Most counting systems focus on the 10's density in the Blackjack deck. The higher the 10s density, the greater the player advantage. Ideally, players want to vary the size of their bets based on the 10s density of the deck. In order for the house to gain an advantage, they would need to

1. Deal a lot of low cards to increase the 10s density and get players to raise their bets
2. Further increase the number of low cards dealt, futher increasing the house odds.
3. Retire the deck before the chicken came home to roost.

I don't buy this one minute.

First and foremost: The only reason that shuffle tracking could be possible is if time constraints force very imperfect shuffling by Dealers. Here, we asked to believe that the Dealers are miraculously able to far beyond a "perfect" shuffle and rig decks on the fly. If this were possible, I suspect that the Casinos would do a better job randomizing the shoes from the get go.

Equally significant: Back loading tens and then retiring decks early significantly changes the odds at the table. The Gambling Comission takes a very dim view when people fix slot machines. This behaviour is identical.

[mike777]

hrothgar, on Nov 15 2005, 06:12 PM, said:

SoTired, on Nov 16 2005, 12:16 AM, said:

This system soaked the casinos. The casinos eventually ID's the teams. Part of their response to combat the teams was changing the shuffle. The casinos use a specific method of shuffling multiple decks. They changed their method several years ago that produces the non-random shuffle to combat counters. I think one team went thru a large losing streak and became more brazen and obvious and that is how they were spotted.

I'm an MIT grad and I knew some of the members of the second generation Blackjack team. I've heard a number of stories back in "the day". A lot of them revolved arround sex, several revolved arround drugs, and there was lots of booze. VERY few stories focused on non-random shuffles. I heard some claims regarding shuffle tracking, however, I never game them much credit. The "core" of the MIT Blackjack curiculeum focused on counting, counting, avoiding detection, and counting.

Step back and consider how a non-random shuffling system would need to behave in order to create an advantage for the house. Most counting systems focus on the 10's density in the Blackjack deck. The higher the 10s density, the greater the player advantage. Ideally, players want to vary the size of their bets based on the 10s density of the deck. In order for the house to gain an advantage, they would need to

1. Deal a lot of low cards to increase the 10s density and get players to raise their bets
2. Further increase the number of low cards dealt, futher increasing the house odds.
3. Retire the deck before the chicken came home to roost.

I don't buy this one minute.

First and foremost: The only reason that shuffle tracking could be possible is if time constraints force very imperfect shuffling by Dealers. Here, we asked to believe that the Dealers are miraculously able to far beyond a "perfect" shuffle and rig decks on the fly. If this were possible, I suspect that the Casinos would do a better job randomizing the shoes from the get go.

Equally significant: Back loading tens and then retiring decks early significantly changes the odds at the table. The Gambling Comission takes a very dim view when people fix slot machines. This behaviour is identical.

As I said the MIT guys 15 years ago moved beyond counting. That was old school. They went to directing the the "ten" cards" onto the hands without touching the deck. Have no idea what new ideas they have come up with 15 years later. They had 3 main methods. The most basic one was literally seeing the bottom card when starting the shuffle, following the card through the shuffle and cutting of the deck and directing the card onto the hand.

OF course knowing the methods one can defeat them but if you are trying to stop basic counting that will not work. Also stopping many of these methods slows down the game and costs money as less hands are played.