Voodoo Archive 1

[ Read Responses | Return to the Index ]
[ Previous | Next in Thread | Next ]

Card-counting at baccarat-The truth (long)

Posted by Green Baize Vampire on 8 June 1998, at 3:21 p.m.

I've been away in the emerald isle recently so I've missed the recent thread on baccarat, which is fascinating stuff and would be more appropriate on a more credible page.

Let me say that I was impressed with Robert's series of posts titled Baccarat-Part X. If he were to develop his original and interesting angles on casino games further then he will become a valued contributor.

Similar thinking aroused my interest in baccarat. A word on the history to put things in their proper perspective.

The first serious work on card-counting at baccarat was done by Edward Thorp and William Walden in the 1960's. Thorp was supervising Walden's thesis, which was concerned with the application of Thorp's blackjack card-counting methodology to baccarat. The level of computer power at their disposal was very primitive, however, even a limited sample allowed them to demonstrate that a card-counting system for bank/player was impractical. There was however in those days two side-bets offered in addition to the main part of the game, paid off at 9-1, that the player or the banker would receive either a natural nine or a natural eight with their first two cards. It is easy to see how countable such a bet would be.

Thorp and Walden trained a team of players to exploit these sidebets, left for Nevada and made several thousand dollars before the team was collectively barred, Thorp was assaulted and the sidebets were removed, never to return. Shortly afterwards the tie bet appeared. Originally it was offered at 9-1, but over the course of decade devalued to 8-1. Thorp's work on the tie bet in his "The Mathematics of Gambling" states that he found the results of his attempts to design a card-count for the tie bet "disappointing" though he gives no hard data as to why this is the case.

The other notable work on baccarat advantage play was done by Peter Griffin, who in the "Theory of Blackjack" explains that computer-perfect play exhibits pitifully small returns, and linear card-counting in fact loses you money on the three bets, owing to its lack of acurracy.

It seemed to me that regardless of Griffin's conclusion it might be possible to design a winning system for baccarat. This is because attempts to design workable baccarat systems had largely concentrated on bank/player (see Tamburin, Thorp, Vancura, Black etc), which is a waste of time. Griffin's tie system lost money because it treated a non-linear phenomena (tie advantage) as linear.

Like Robert it seemed obvious to me that the removal of all the odd cards would give you an advantage, since only five totals would be possible, doubling the probability of a tie. However, you can't use a count to exploit this. Advantages only ocurr when ALL the odd cards are removed (when they become quite large). To see why, as a though excersize, take ten even cards from a bac deck. Play a few hands of bac. You'll see the tie has a huge advantage. Now add an odd card. The edge disappears. That's because if the odd card appears in either the bank or player hand the result CANNOT be a tie. My sims indicate that a player only betting a $1000 on tie whenever there were no odd cards would win about $1 an hour.

(I am incidentally extremely angry that the tiebreaker system, which is an inferior version of the above method, is being marketed. I am reasonably certain that the method was plagiarized from my own writings. Note that I published this method in blackjack review and in Dejanews (check the archives) long before Tiebreaker appeared. I would be interested in hearing from anyone who purchased the system to see if I have grounds for legal action.)

The deep penetration at bac causes the effects of removal to warp at low deck subsets, precisely when advantages occur. This means card-counting systems based on the removal of cards from a full-deck are of little use, you only get advantages on the tie when a number of card ranks are absent. With only tens in the deck you have a 800% edge, since the result must be a tie, yet a linear card-count would tell you that the tie was disadvantageous. (Math weenies:This phenomena is related to the "floating advantage", which is of real importance to baccarat where it is not at blackjack)

You can get round this by writing down the cards as they are dealt on the scorecard you are given. With a little training, it is not difficult to teach yourself when to bet on the tie. You can capture the lion's share of EV this way.

Unfortunately the system I devised only makes about $20 per shoe, if you stake a $1000 with every favourable situation,which is not far off Griffin's figures. Moreover, opportunities are extremely rare. Finally, when I did this research I was ignorant of optimal betting when uneven payoffs are concerned. I now reallize a tie bettor would have to bet more conservatively than any kind of reasonable profit would allow, leading me to abandon the room for improvement in my non-linear system.

Most of the profit comes from situations where only tens and a pair of cards remain in the deck. The advantage in these situations is two or three hundred %, sometimes even more, rare as these ocassions are.

Under very unusual circumstances the system I created can make some money. In some parts of the world only eight cards are cut out regularly. You can usually spot the bottom card if you sit in the right place. Sometimes only seven cards remain when you place your final bet. In this instance, and this instance only, baccarat can be quite favourable, though only to the well-financed.

I will be happy to provide the full system to anyone who asks for it. I am reasonably certain that it is the most powerful baccarat system in existence, pathetic though it's returns may be.

To make real money at baccarat requires something altogether more potent...



The Voodoo Archive 1 is maintained with WebBBS 2.24.060398.