Baccarat and card counting-is the Wizard wrong?
Thread Rating:
I run a baccarat player site (no link here, I read the policy on WOO carefully, so you'll find me if I'm worthy) and I think the Wizard is wrong when he says there's no point in counting cards in baccarat.
In Scoblete's Baccarat book, he mentions a system for counting by Tamburin and Rahm that lets you know when betting "Player" is a better bet than "Banker."
In short, this is the system:
The shoe starts at +14
card have values of:
8,9,10,J,Q,K = -1
2,5 = +1
3,4 = +2
A,6,7 = 0
Anytime the count drops below zero, you switch from betting Bank to Player and improve your odds.
I find it hard to believe Tamburin had it wrong, and while you can't have a positive EV bet in baccarat, reducing your losses is a good thing, and counting keeps your brain working.
Am i incorrect in asserting that counting has value in baccarat (and is fun) or has the Wizard made a rare mistake?
Counting in baccarat is basically a waste of time. You would need to track every card and use a computer to beat baccarat with card counting. For more details, see my baccarat appendix 2.
I definitely concede that a $100/hand bettor might only save 25-75 cents in expected losses per hour, but it IS fun to have a system that tells you when Player is actually a better bet.
I regret my post title, as I don't actually think you are wrong from a practical standpoint-but when has playing a negative expectation game for money ever been practical? I just like to switch between Bank and Player and have operated under the assumption that I had at least a passing justification.
So, is the Tamburin system correct in it's approach to switching between Bank and Player to *very* nominally reduce the house edge? Scoblete says "Player" becomes a slightly better bet roughly 10% of the time, and the WOO baccarat calculator shows that at least some feasible deck situations make Player a better bet.
When I play, I track like the counting math I presented is solid (though I didn't test it) so I hope I haven't been wasting my time.
Anymore than tracking for a quarter per hour is a time-waster!
Quote: ChipDeFerrariIn Scoblete's Baccarat book, he mentions a system for counting by Tamburin and Rahm that lets you know when betting "Player" is a better bet than "Banker."
In short, this is the system:
The shoe starts at +14
card have values of:
8,9,10,J,Q,K = -1
2,5 = +1
3,4 = +2
A,6,7 = 0
Anytime the count drops below zero, you switch from betting Bank to Player and improve your odds.
Even if counting baccarat for fun, I can't endorse that strategy. For one thing, the +1 for a five implies that fives are good for the Banker bet. My appendix when fives leave the deck, it favors the Player bet.
I would change the points per card as follows:
1 to 4 = +1
5 to 8 = -1
I don't know at what point you should start at to switch to Player below zero.
I can't help but say that if you're willing to work this hard, why not have an advantage, and count cards in blackjack?
Ah... now that is what you really enjoy! The challenge.Quote: ChipDeFerrariand counting keeps your brain working.
And maybe it does really "work" in that it gives you a very miniscule improvement but mainly it works because you enjoy doing it as a mental exercise. Its sort of like sticking your thumb out as an armored car goes by: you know they won't stop to pick up a hitchhiker, but its good for a laugh and it really doesn't cost you anything.
So my advice would be, sure... keep up your "card counting at Baccarat" routine. Even refine it somewhat from time to time. It keeps you happy, so why not do it. Just remember that your net gain in any session won't really cover one tip to the cocktail waitress much less a tip to the dealers. And ofcourse once you factor in the inevitable dealer error and player error ... well, there goes any benefit of your having counted cards.
Yet, its harmless and it won't make things worse for you. So why not just keep honing your counting skills and have fun with your hobby.
Yes, I imagine that would be even better, though counting at blackjack is more easily noticed and more likely to bring casino heat of some sort.Quote: WizardI can't help but say that if you're willing to work this hard, why not have an advantage, and count cards in blackjack?
My back of the envelope math says it saves the $100/hand bettor about $1 per hour, which is miniscule but does pay a few tips.
I do count cards at blackjack also, but it's more about the fact that betting Bank every time isn't optimal, you should bet Bank about 90% of the time and Player 10%.
Anyway, thanks for the replies, and thanks to the Wizard for both the forum and your proposed strategy optimization, I'll crunch the numbers and reduce my expected loss by another couple pennies an hour!
--
Also, when you reply to a thread it would be nice to see the whole thing, right now it's just a blank window and you need to open another tab to see the thread.
Quote: ChipDeFerrari
Also, when you reply to a thread it would be nice to see the whole thing, right now it's just a blank window and you need to open another tab to see the thread.
AMEN! (cough, vbulletin, cough)
This system isn't designed to get a house edge, it's designed to help you play the game correctly in all circumstances.
When I asked "Is the Wizard Wrong?" I really meant "Should you always bet Bank in baccarat if your goal is proper play?" And I *think* the Wizard agrees with me that you should mathematically occasionally bet Player to minimize the house edge--he just thinks it's not worth your time, ergo the advice to always bet 'Bank.'
Which I still think is wrong, from a proper strategy perspective.
Wizard, I know you smacked me down mightily here, but I'll be in Las Vegas Monday-Wednesday and really value your sites, so if I can buy you lunch for your efforts here and at WOO, shoot me an email at the address I registered.
Quote: ChipDeFerrariHello,
I find it hard to believe Tamburin had it wrong, and while you can't have a positive EV bet in baccarat, reducing your losses is a good thing, and counting keeps your brain working.
From "Beyond Counting" by James Grosjean, page 161:
"Stay clear of the book by Tamburin and Rahm, which displays such a weak understanding of statistics that I cannot trust their count system, which by their admission does not provide an edge anyway."
--Dorothy
----------------------------------------
DGs Minimally Offensive Signature
Quote: Jufo81About card counting in Baccarat: There is an online casino where the cut card is put ahead last 7 cards in the eight-deck shoe. So in best scenario the last Baccarat hand will be played with only 7 cards remaining. Given that the player can get the precise odds real-time using a calc such as this, wouldn't there be an advantageous scenario here? What I would be interested to know is the ratio of all shoes that yield some kind of edge (either for player, banker or tie) on the last hand with 7-12 cards remaining, because if the ratio is low, then it is simply too much effort to track every card in a shoe accurately.
Seems like if all that were remaining were 10s and face cards, you would clean up betting on ties. So the edge has to be high even if there were one or maybe two other cards. The wizard's work on baccarat generally says don't bother counting, but you may be describing a situation where it would be worth it.
Quote: odiousgambit
Seems like if all that were remaining were 10s and face cards, you would clean up betting on ties. So the edge has to be high even if there were one or maybe two other cards. The wizard's work on baccarat generally says don't bother counting, but you may be describing a situation where it would be worth it.
Yes, you would have 900% return on the tie bet if the last seven cards were all tens. However, the odds for that is low and only a fraction of all shoes happen to end with seven cards remaining for the last hand. However, doing a little playing with the Baccarat calculator, one can get many combinations with a player edge. For example if the remaining shoe is 4 tens, 2 fives and 1 eight, then the tie bet has a 46% player edge as well as 3.5% player edge for the banker bet.
It's obvious that there exists player edge situations with this game. The only interesting question is whether these situations occur frequently enough for the exact tracking of cards to be worthvile. If 1% of all shoes yields an edge, it is not worth the effort, but if 10% of shoes does it, then it might be.
Quote: MichaelBluejayHere's a seemingly good article about exploiting the Tie bet. Maybe the Wizard will comment about this on his Baccarat card counting page at WizardOfOdds.com.
That article in part says, " Assuming we make our last wager having seen all but a generous 8-13 cards." It is my understanding that in baccarat the cut card is placed 16 cards from the end of the shoe. When it comes out, they finish that hand, and then play one more. If it comes out as the first card of a hand, they finish that hand, and then shuffle. So, the best case scenario is it comes out after one card, and five more cards come out that hand. That would leave 11 unseen cards. The worst case scenario is 16 more cards. The following table shows the probability of all even cards according to the number of cards remaining.
| Cards Left | Probability |
|---|---|
| 11 | 1 in 209 |
| 12 | 1 in 339 |
| 13 | 1 in 551 |
| 14 | 1 in 895 |
| 15 | 1 in 1455 |
| 16 | 1 in 2364 |
On average, you'll get a positive bet once every 509 shoes. Doesn't sound like a practical advantage play to me. I think Thorp, Griffin, and me are right in saying that baccarat is, for all practical purposes, not countable.
I've said this before, but if there were an angle counting baccarat, it would be on the pair bets.
Quote: WoOAt the start of a new shoe, the dealer will turn over one card. This will determine how many cards the dealer will burn, according to the baccarat value, except a 10 or face card will result in 10 cards burned.
This quote is from the Wiz of Odds site. Unless these burn cards are burned face up, it adds 1 to 10 unseen cards.
Quote: mipletThis quote is from the Wiz of Odds site. Unless these burn cards are burned face up, it adds 1 to 10 unseen cards.
You're referring to the front of the shoe. The first card refers to the number of burned cards, which are shown to the players.
Quote: WizardYou're referring to the front of the shoe. The first card refers to the number of burned cards, which are shown to the players.
Ok, I assumed that burned cards were not shown to the player in baccarat as they are not shown blackjack.
Quote: TankoCard counting in Baccarat is totally useless. I tried it. Even a sustained count as high as 35, has no influence at all on the Hand outcome.
This is Elliot Jacobson's discussion on the subject:
Flim Flam.
Perfect tie bet counting yields a 0.5% edge with a 1-40 spread with decent conditions. That's not trivial given the amunt of money you can put down.
To be kind, it would be fair to say Eliot Jacobson knows nothing at all about the practice of counting cards at baccarat at a professional level.
ZCore13
Quote: Zcore13I will take Elliot Jacobson's opinion (which is the same as the Wizard's) of not wasting your time trying to count Bac over yours every day of the year. As the Wiz said earlier in this thread, if you're going to take the time to count cards and attempt to have an advantage over the house, why not play blackjack side bet games that are proven beatable. Why not play Bac sidebets that are proven beatable? You're wasting your time when the math tells you it's not worth the time to do it.
ZCore13
You can put down as much as $100,000 on tie at the baccarat table with a >100% edge. That's what "the math" tells you. You only need to be able to do that a few times a year to make a very good living. When you add in 20%+ rebates, the type of excellent conditions a VIP player can negotiate, heads-up vs the dealer you have a strong play.
If you are trying to scrape a bankroll together from dirt then, sure, go for the side bets. Personally, I would have a hard time finding high enough limits to make it worth my while to do that. If I did, it wouldn't be worth the exposure, since it is very easy to figure out a count play that simple. I'd rather do stuff which hasn't been blabbed all over the internet and get the boot everywhere.
There's a discrepancy between the goals of a low-roller and a high-roller. Low-rollers need a big edge and low variance just to pay their expenses. High-rollers have difficulty getting a decent return on investment, expenses are irrelevant. A 40% return on investment on a million-dollar bankroll is pretty good-a hedge fund manager that could deliver those returns would be feted to the heavens (or possible arrested on suspicion of fraud). A 40% return on a $10,000 bankroll is pointless.
A significantly better ROI than that is obtainable from baccarat using relatively crude methods.
Quote: GBVYou can put down as much as $100,000 on tie at the baccarat table with a >100% edge. That's what "the math" tells you. You only need to be able to do that a few times a year to make a very good living. When you add in 20%+ rebates, the type of excellent conditions a VIP player can negotiate, heads-up vs the dealer you have a strong play.
Ok I will Eliot Jackobson does evaluate a counting system for Baccarat that does give you a great then 1 ROR. Specifically a return of $8.15 per hundred hands per thousand dollars bet. Unfortunately with this system you are making a bet once every roughly 318 hours of play. So say you could make a 100,000 dollar bet on tie which seems high but ok. That means you are making 815 dollars an hour which sounds like a lot but its not really when you consider the element of risk involved. Also that is being generous saying a cut card placement of 12 cards you are more likely to find 14 cards where it is only equal to 315 dollars an hour. I seriously think your making the flawed assumption well I'd only need to win it a few times a year to be super happy I mean it carries a 60% edge when making the bet so thats equivilant to winning 60,000 dollars on the bet in terms of EV but really when staking hundreds of thousands of dollars in gambling you really probably have expenses greater then the 60,000 or 120,000 your talking about unless you are living well below your means.
As for the 20% loss rebate yes that makes the game positive EV though loss rebates can make near any game positive. I mean single zero roulette with a 20% loss rebate will win on average 19,881 dollars over the course of half an hour. For baccarat it is 22,500 dollars per 23 minutes of work. Given these numbers you can see the roughly 800 dollars won per hour is just swamped by the actually advantage from the loss rebate.
Quote: TwirdmanOk I will Eliot Jackobson does evaluate a counting system for Baccarat that does give you a great then 1 ROR. Specifically a return of $8.15 per hundred hands per thousand dollars bet. Unfortunately with this system you are making a bet once every roughly 318 hours of play. So say you could make a 100,000 dollar bet on tie which seems high but ok. That means you are making 815 dollars an hour which sounds like a lot but its not really when you consider the element of risk involved. Also that is being generous saying a cut card placement of 12 cards you are more likely to find 14 cards where it is only equal to 315 dollars an hour. I seriously think your making the flawed assumption well I'd only need to win it a few times a year to be super happy I mean it carries a 60% edge when making the bet so thats equivilant to winning 60,000 dollars on the bet in terms of EV but really when staking hundreds of thousands of dollars in gambling you really probably have expenses greater then the 60,000 or 120,000 your talking about unless you are living well below your means.
As for the 20% loss rebate yes that makes the game positive EV though loss rebates can make near any game positive. I mean single zero roulette with a 20% loss rebate will win on average 19,881 dollars over the course of half an hour. For baccarat it is 22,500 dollars per 23 minutes of work. Given these numbers you can see the roughly 800 dollars won per hour is just swamped by the actually advantage from the loss rebate.
The "system" Jacobson comes from an article I wrote fourteen (yes, fourteen) years ago. It is not a practical system intended for casino play. It is meant to illustrate how non-linear systems can outperfom linear systems such as those developed by Thorp and Griffin. That said, Jacobson's simulation is way off-base, partly because he ignores the potential for backcounting. Consider that if you are just counting even-valued cards, there will be a lot of situations where the evens/other card ratio becomes so poor that it is better to leave the game for a fresh pack. By doing so, you increase the frequency with which you encounter all-evens subsets.
You mention variance: something Jacobson clearly doesn't understand is that the optimal bet size with a very advantageous wager can be very high. With an all-evens subsets the Kelly bet is 8% of your entire bankroll. $1000 may sound aggressive to some: in fact you would be massively under-betting in many of the more favourable advantage situations. To put it into perspective, in terms of risk and reward a single 65% edge on the tie wager is equivalent to playing a 2-1 blackjack promotion for roughly an hour.
To actually beat baccarat you need to be able to develop a heuristic to identify other significantly favourable situations. This more recent article goes into more detail: http://forumserver.twoplustwo.com/showpost.php?p=4843782&postcount=1.
This obviously needs repeating: you can't beat baccarat as a practical matter with a simple system, you have to perform something akin to mental combinatorial analysis.
On game conditions: there are better conditions available than most theorists seem to assume. In the UK for the best part of a decade it was quite easy to find games dealt down to the last card but one, and you could often see the back card. The evolution live gaming software also dealt down to the same level: rumour has it William Hill alone lost 8 figures before they changed the rules. There are other opportunities which currently exist that I'm not going to compromise by mentioning here.
Rebates: Yes, a rebate can potentially make any gambling game positive EV. As a practical matter though, most rebates are designed to leave a tiny edge for the casino. There will usually be some stipulation about a minimum number of hands-otherwise you could just bet the whole lot on a single hand, and the minimum hand requirement eats into your profit. Even when a casino gets greedy/screws up and gives you the best of it, then it often isn't enough to bet into by itself. The difference between a 0.1% edge and a 0.5% edge is as night and day. You need all the edge you can get. In most situations you'll be able to break-even with rebates and then freeroll their money on very infrequent, very advantageous situations. At the end of the year you find out your profit is pretty close to your EV on those few dozen hands.
There are many other possibilities here which make baccarat counting a viable play. Stuff I've been able to get away with: argue my bank commissions away, negotiate 100% penetration, get the dealer to shuffle-up on request, exploit trackable shuffles etc etc. It all adds up.
Quote: GBVQuote: TankoCard counting in Baccarat is totally useless. I tried it. Even a sustained count as high as 35, has no influence at all on the Hand outcome.
This is Elliot Jacobson's discussion on the subject:
Flim Flam.
Perfect tie bet counting yields a 0.5% edge with a 1-40 spread with decent conditions. That's not trivial given the amunt of money you can put down.
To be kind, it would be fair to say Eliot Jacobson knows nothing at all about the practice of counting cards at baccarat at a professional level.
With your perfect tie bet counting and 1-40 spread with decent conditions, what is the betting frequancy ? What is your EV/shoe ?
Quote: GBVThe "system" Jacobson comes from an article I wrote fourteen (yes, fourteen) years ago. It is not a practical system intended for casino play. It is meant to illustrate how non-linear systems can outperfom linear systems such as those developed by Thorp and Griffin. That said, Jacobson's simulation is way off-base, partly because he ignores the potential for backcounting. Consider that if you are just counting even-valued cards, there will be a lot of situations where the evens/other card ratio becomes so poor that it is better to leave the game for a fresh pack. By doing so, you increase the frequency with which you encounter all-evens subsets.
You mention variance: something Jacobson clearly doesn't understand is that the optimal bet size with a very advantageous wager can be very high. With an all-evens subsets the Kelly bet is 8% of your entire bankroll. $1000 may sound aggressive to some: in fact you would be massively under-betting in many of the more favourable advantage situations. To put it into perspective, in terms of risk and reward a single 65% edge on the tie wager is equivalent to playing a 2-1 blackjack promotion for roughly an hour.
To actually beat baccarat you need to be able to develop a heuristic to identify other significantly favourable situations. This more recent article goes into more detail: http://forumserver.twoplustwo.com/showpost.php?p=4843782&postcount=1.
This obviously needs repeating: you can't beat baccarat as a practical matter with a simple system, you have to perform something akin to mental combinatorial analysis.
On game conditions: there are better conditions available than most theorists seem to assume. In the UK for the best part of a decade it was quite easy to find games dealt down to the last card but one, and you could often see the back card. The evolution live gaming software also dealt down to the same level: rumour has it William Hill alone lost 8 figures before they changed the rules. There are other opportunities which currently exist that I'm not going to compromise by mentioning here.
Rebates: Yes, a rebate can potentially make any gambling game positive EV. As a practical matter though, most rebates are designed to leave a tiny edge for the casino. There will usually be some stipulation about a minimum number of hands-otherwise you could just bet the whole lot on a single hand, and the minimum hand requirement eats into your profit. Even when a casino gets greedy/screws up and gives you the best of it, then it often isn't enough to bet into by itself. The difference between a 0.1% edge and a 0.5% edge is as night and day. You need all the edge you can get. In most situations you'll be able to break-even with rebates and then freeroll their money on very infrequent, very advantageous situations. At the end of the year you find out your profit is pretty close to your EV on those few dozen hands.
There are many other possibilities here which make baccarat counting a viable play. Stuff I've been able to get away with: argue my bank commissions away, negotiate 100% penetration, get the dealer to shuffle-up on request, exploit trackable shuffles etc etc. It all adds up.
I'd just like to say though yes the tie bet and perhaps even other bets may be countable in baccarat the other advantages you are talking about far outweigh them. I mean with no banker commision just flat betting banker is a positive EV play why bother counting. With loss rebate flat betting banker is positive EV why bother counting. I mean with no commision you are talking a 1.2% advantage far better then your .5% you mentioned for tie and far simpler to accomplish. If you already have one of these advantages and felt the need to count out of boredom sure go for it and occasionally there will be positive scenarios but I'm willing to bet bulk of your wins come from other advantages.
Quote: TwirdmanI'd just like to say though yes the tie bet and perhaps even other bets may be countable in baccarat the other advantages you are talking about far outweigh them. I mean with no banker commision just flat betting banker is a positive EV play why bother counting. With loss rebate flat betting banker is positive EV why bother counting. I mean with no commision you are talking a 1.2% advantage far better then your .5% you mentioned for tie and far simpler to accomplish. If you already have one of these advantages and felt the need to count out of boredom sure go for it and occasionally there will be positive scenarios but I'm willing to bet bulk of your wins come from other advantages.
That's not how it works.
The edge from some situational plays may be larger than that from counting, but there's a huge amount of wasted downtime searching for those plays.
When you do find a good angle, you can't be choosy about, say, hands per hour, nor can you follow a specific dealer around, and of course staying under the radar is a priority. For these reasons and others you need a staple play to keep your profits up.
In addition, counting baccarat may be the "special" play, or is an essential component. If you find a game dealt down to the last card for example, your edge is a lot more than 1.2%.
Regarding bank commissions : about the only situation where you would truly have a 1.2% edge is when you get a promotion. Most of the time it is something that happens only after a heavy loss. It isn't a dependable source of profit.
It is dangerous to apply the type of analysis you get on the WOO site too literally to casino play. It is all very well to calculate the optimum edge under some generic set of conditions, indeed you have to do that to determine whether a game is playable. But, it is your hourly earnings under real-world conditions that determine how much money you actually make.
Quote: ssho88Quote: GBVQuote: Tanko
With your perfect tie bet counting and 1-40 spread with decent conditions, what is the betting frequancy ? What is your EV/shoe ?
It varies greatly with penetration but around 2% as an average figure when you play through every shoe.
However, you do not want to be playing through every shoe, as I discussed above.
The very low frequency of opportunity for the play-all approach is the primary reason why existing studies on counting cards at baccarat tend to understate the potential gains.
The EV/shoe is 0.5% of whatever your average bet is during the shoe.
most now agree that the count method is flawed in that book.Quote: charlespWhy do you start at +14?
iirc, that is a running count. but no matter to me.
Remember, they only wanted to sell books
and NOT use that system to play for real money in a casino.
IF they did use it, they, imo, did not play many shoes with it
better to stay with one showing data (and more than one agree on the values)
as you have seen also here
https://wizardofvegas.com/forum/gambling/tables/26218-baccarat-count-based-betting/
Sally
