Observed house edge in baccarat?

Quote: AcesAndEights

No, this is incorrect. Hold = Amount casino keeps / Total value of chips purchased. If someone buys $1000 in chips, then goes directly to the cashier to cash them out, the hold for this person would be 0%; this will skew the observed hold, and the person didn't even wager anything.

Ah I see. I'm approaching it from a sports betting background where hold has a slightly different meaning. Thanks for clarifying.

Quote:

Yes, the actual house edge depends on the proportions of each bet the players are taking. It's only mathematically possible to determine the exact house edge on a single bet at a time. The Wiz proposed a proportion of bets that would match with what the Macau casinos are reporting for the investors. But the actual proportions for each bet and thus the combined house edge is something that would vary from casino to casino, city to city, and would require a lot of observation and data collection. I'm certainly not an expert on the gambling industry, so I don't know if that kind of data exists or would be available. Based on the people I've seen playing Baccarat (very superstitions, try to play streaks, will bet the Tie if it hasn't come up in a while), the 2.85% they use in Macau doesn't sound too far off.

Thanks.

To clarify, table hold = win / drop. Drop = $ in drop box. Win = extra cash in the cage after all the chips have been returned. Most table game pits cannot track handle (i.e. total amount bet, total wager, action, "coin-in" in slot parlance, etc.) so the casinos use the data they *can* collect.

Realized house advantage = win / handle. This does converge on the house edge, but it's almost never measured (except for slots).

Quote: MathExtremist

To clarify, table hold = win / drop. Drop = $ in drop box. Win = extra cash in the cage after all the chips have been returned. Most table game pits cannot track handle (i.e. total amount bet, total wager, action, "coin-in" in slot parlance, etc.) so the casinos use the data they *can* collect.

Spot on.

Quote: MathExtremist

Realized house advantage = win / handle. This does converge on the house edge, but it's almost never measured (except for slots).

Also true - H.E is never directly measured for table games, as it is impossible. Each bet's transaction would have to be tracked and recorded, which cannot be done in the real world. What can we do then?

Hence the fairly lose reverse engineering done ("ball park") as follows on a "well, this is what we can do with the casino figures we got" basis: Assumed Table game House edge ~ hold * 0.1

If a table held 28% for a three month period, it is assumed to be performing with roughly a 2.8% H.E. for that period; so, if it's nominal (official) math report house edge is 2.5%, it is assumed to be performing within range, and acceptable.

This is how we "oversee" our products that are active in the field, for example:

Pai Gow game "x" has been performing in a three month period with a 27% house edge, on $850,000 drop at casino x with a table hold of $229,500; it has an official H.E. of 2.47% house plus it was install with side bet y that has a H.E. of 3.2%. So, it is performing within range, and no dealer training and no "best practices" need to be suggested to the casino operator. Fine. Baccarat table "y" is performing with a 16% H.E from $1,800,000 in buy-in drop, holding $288,000. It has an official house edge of 1.3%. It is performing slightly above spec. Again fine.

Table game "z" saw $90,000 worth of drop, and held $9,000. Table hold was 10% but it has an official house edge of 3.1%. Not good. It's performance indicates it has a small drop, and it is holding less than what could be expected, which would be closer to $28,000; the casino operator expressed a concern over "table game z." We examine with Casino operator "q" dealer training and procedure ("is there a misunderstanding on payout procedures or hand evaluation procedures?"), possible Advantage Play or dealer game protection ("is dealer flashing hole cards?") product flaw ("is side bet payout misprinted somewhere?) etc., etc., etc.,

Quote: Paigowdan

Spot on.


Also true - H.E is never directly measured for table games, as it is impossible. Each bet's transaction would have to be tracked and recorded, which cannot be done in the real world. What can we do then?

Hence the fairly lose reverse engineering done ("ball park") as follows on a "well, this is what we can do with the casino figures we got" basis: Assumed Table game House edge ~ hold * 0.1

If a table held 28% for a three month period, it is assumed to be performing with roughly a 2.8% H.E. for that period; so, if it's nominal (official) math report house edge is 2.5%, it is assumed to be performing within range, and acceptable.

This is how we "oversee" our products that are active in the field, for example:

Pai Gow game "x" has been performing in a three month period with a 27% house edge, on $850,000 drop at casino x with a table hold of $229,500; it has an official H.E. of 2.47% house plus it was install with side bet y that has a H.E. of 3.2%. So, it is performing within range, and no dealer training and no "best practices" need to be suggested to the casino operator. Fine. Baccarat table "y" is performing with a 16% H.E from $1,800,000 in buy-in drop, holding $288,000. It has an official house edge of 1.3%. It is performing slightly above spec. Again fine.

Table game "z" saw $90,000 worth of drop, and held $9,000. Table hold was 10% but it has an official house edge of 3.1%. Not good. It's performance indicates it has a small drop, and it is holding less than what could be expected, which would be closer to $28,000; the casino operator expressed a concern over "table game z." We examine with Casino operator "q" dealer training and procedure ("is there a misunderstanding on payout procedures or hand evaluation procedures?"), possible Advantage Play or dealer game protection ("is dealer flashing hole cards?") product flaw ("is side bet payout misprinted somewhere?) etc., etc., etc.,

Thanks for the explanation. Again, I'm coming from a purely sportsbetting background so I'm used to being able to track handle.

You seem to have a lot of experience with this...for a Vegas Strip casino, do you happen to know rough H.E. for Baccarat and Craps?

Quote: Paigowdan

H.E is never directly measured for table games, as it is impossible. Each bet's transaction would have to be tracked and recorded, which cannot be done in the real world.,

Slot HE manifests itself on a daily basis, table
games do not. If a casino could get by on just
offering slots, they'd do it in a heartbeat. Table
games are a nightmare, but they do make money
and cater to a certain portion of the gambling
public.

The HE for any bet placed usually refers to the long-term average of random distribution. Over short periods of time this HE varies due to randomness. (a.k.a. variance). Even a Slot machine with a 9% HE can have periods of +EV, and more likely periods of -EV.

Slots are cheap because they take reletively small area, have small, incidental human salaries and expenses per machine, and operate quickly, with usually a 7-12% HE. Some Progressives go 1% or less HE. Nonetheless, as previously posted there are far more slots than tables for this reason. Card-games as gambling will always have their place in a Casino. The future however definately points to automated gaming at tables (read: video tables with no Dealer).

Quote: 98Clubs

The HE for any bet placed usually refers to the long-term average of random distribution. Over short periods of time this HE varies due to randomness. (a.k.a. variance). Even a Slot machine with a 9% HE can have periods of +EV, and more likely periods of -EV.

In your 9% example, I don't see how a slot machine can have a period of +EV.

In the end, all games converge to HE, whether it's slots, BJ, baccarat, sports betting, Keno, etc. My original question was the long-term HE on games that have varying HE bets within them (baccarat, craps).

Quote: JonDough

In your 9% example, I don't see how a slot machine can have a period of +EV.

In the end, all games converge to HE, whether it's slots, BJ, baccarat, sports betting, Keno, etc. My original question was the long-term HE on games that have varying HE bets within them (baccarat, craps).

Baccarat and craps can have wild swings in the short term, especially if a crap table has a Fire bet (huge payout side bet) on it.
What happens with Bac and crap tables is that they MAY have some wild & hard short term swings, but over the long haul are money makers and stay true to their H.E. of the main bets You have to trust the math over the long haul.

On a Long-term basis, the house edge on Bac is accurate at about 1.4% (14% table hold) with large cash drops, and craps is about 2% (20% TH), also with large cash drops. Pai Gow hold fairly steady at about 22%-30% table hold, and Blackjack in mature markets of experienced players like Atlantic City can have a slim house edge and table hold, sometimes scary, but with large drops. Side bet patronage helps the BJ house edge - and cash 'hold' of the games, - and BJ layouts will always have some sort of side bets on them to pull in extra income. A necessary evil from the casino operators POV.

Roulette seems to be very steady by comparison. Full tables where Everyone plays their own numbers and styles, the aggregate bets cancel each other out, - with the zeros pulling in a steady house edge. Roulette gets less dollar action than it used to; in American Markets it's BJ, craps, Pai Gow Poker, and carnival games like Three card poker and Ultimate Texas Hold 'em. In Asia, 80% of the action seems to be Baccarat. In Europe, a lot more Roulette action. And Asian tiles - Atlantic City has 50 or so tables.

If you search hard, a lot of gaming jurisdictions post their table drop and hold numbers like Missouri Gaming Commission, along with New Jersey. A TON of information can be determined from their data.

Quote: 98Clubs

The HE for any bet placed usually refers to the long-term average of random distribution. Over short periods of time this HE varies due to randomness. (a.k.a. variance). Even a Slot machine with a 9% HE can have periods of +EV, and more likely periods of -EV.

A slot machine can't have an expected value of +9%, unless there's some sort of jackpot. It might have an actual value of +9% over that period (looking backwards) but it's expectation cannot fluctuate, unless the conditions do.

Quote: Paigowdan

in American Markets it's BJ, craps, Pai Gow Poker, and carnival games like Three card poker and Ultimate Texas Hold 'em.

I checked the Missouri Gaming Commission and in 2011
every month the top 3 were BJ, craps and roulette. Every
month they had the biggest drop. Pai Gow was usually
4th or 5th.