THE MATHEMATICS
Casino House Edge and What Each Table Game Actually Costs
House edge is the percentage of every ringgit staked that a casino keeps on average across many resolved bets. On the U88 live floor it runs from about 0.5% for blackjack with basic strategy and 1.06% on baccarat Banker, through 2.70% on single-zero roulette and 5.26% on double-zero, to roughly 14.4% on the baccarat Tie.
The casino house edge is the share of every ringgit staked that the operator keeps on average over a long run. It is charged per bet resolved, not per session or per hour. Blackjack with basic strategy is cheapest at about 0.5%, then baccarat Banker at 1.06%, Player at 1.24%, European roulette at 2.70%, American roulette at 5.26% and the baccarat tie at about 14.4%.
- House edge is a price per bet resolved, so what you pay depends on total turnover rather than on how long you sat or how much you deposited.
- RTP and house edge are the same statement from opposite ends, since a 97.30% return is identical to a 2.70% edge on a single-zero wheel.
- Hit frequency is a separate question entirely, which is why the baccarat tie can land one round in ten and still be the worst bet on the layout.
- A cheap game played quickly costs more per hour than an expensive game played slowly, because pace multiplies turnover faster than price does.
- Rule details move the number more than most player decisions do, with a 6 to 5 blackjack payout adding roughly 1.4% to the cost of that table.
- No betting pattern changes the edge, because expected value adds up across bets no matter what order or size you place them in.
Everything else on this site refers back to this page, so it is worth being precise from the first line. The house edge is the average share of each amount staked that the operator retains over a very large number of resolved bets. It is not a fee, it is not deducted anywhere you can see, and it is not a prediction about tonight.
Is the house edge charged per bet, per session or per hour?
Picture two players. One deposits RM1,000, puts the entire amount on Banker in a single hand, and walks away. The other deposits RM100 and plays two hundred hands of RM10 on the same bet. They both staked RM1,000 in the first case and RM2,000 in the second, and it is the staked total that matters. The second player has exposed twice as much money to the edge despite depositing a tenth as much.
That is the mechanic almost every misunderstanding comes back to. The edge does not attach to your deposit, your balance, your session length or your bankroll. It attaches to turnover, which is stake multiplied by the number of bets resolved. A player who recycles the same RM100 through forty hands has staked RM4,000 in the eyes of the mathematics.
So the honest way to read any figure below is as a price list. At 1.06% the baccarat Banker bet costs roughly one ringgit for every hundred you put through it. At 5.26% a double-zero roulette wheel costs about five. Neither number tells you what will happen in an hour, because over any real session the swings are far larger than the price.
RTP is the same number standing at the other end of the sentence
Return to player and house edge describe one thing. RTP is what comes back on average; the edge is what stays behind. They sum to 100%, so a single-zero roulette wheel with a 2.70% house edge has a 97.30% RTP, and the two statements carry exactly the same information.
| Term | What it measures | Example |
|---|---|---|
| House edge | Share of turnover retained on average | 2.70% on a single-zero wheel |
| RTP | Share of turnover returned on average | 97.30% on the same wheel |
| Hit frequency | How often a bet wins anything at all | Just under half of all even-money spins |
| Volatility | How widely results scatter around the average | Wide on a straight-up number, narrow on red |
The convention differs by format for historical reasons rather than logical ones. Table games are quoted as an edge, machine games are quoted as RTP, and the same game can appear in both dialects. If you are ever comparing across the two, convert one into the other before you decide anything.
Hit frequency is a third number, and it is the one that fools people
How often a bet wins and what a bet is worth are unrelated questions. The clearest demonstration in the building is the baccarat tie: it lands roughly 9.5% of the time, which is about one round in ten and a half and feels frequent enough to chase. It pays 8 to 1 when it needs to pay slightly over 9.5 to 1 to break even, and that shortfall is the roughly 14.4% edge on the baccarat tie.
The reverse case is just as instructive. A straight-up roulette number wins once in thirty-seven spins on a European wheel, which feels hopeless, and it carries exactly the same 2.70% price as backing red. Rarity is not expensive and frequency is not cheap. What sets the price is the gap between the true odds of an event and the odds the table pays for it.
Games are designed to exploit the confusion. A high hit frequency with a reduced payout feels generous and can be priced worse than a rare, well-paid bet. That design is at its most aggressive on the optional wagers around the edge of the layout, which is why they get a page of their own: the casino side bets page sets out what each one pays and why it costs several times what the main bet costs.
The floor ranked from cheapest to most expensive
Every figure here comes from the published rules and payout schedule of the game. None of it comes from anyone testing anything, and none of it changes with your bankroll, your history or your luck.
| Bet | Conditions | House edge |
|---|---|---|
| Blackjack, basic strategy | Liberal rules, 3 to 2 naturals | About 0.5% |
| Baccarat, Banker | Eight decks, after 5% commission | 1.06% |
| Baccarat, Player | Eight decks | 1.24% |
| Roulette even money, French rules | Single zero with La Partage or En Prison | 1.35% |
| European roulette, any bet | Thirty-seven pockets, single zero | 2.70% |
| Sic bo, Big or Small | Loses on any triple | 2.78% |
| American roulette, any bet | Thirty-eight pockets, double zero | 5.26% |
| American roulette, five-number bet | Zero, double zero, one, two and three | 7.89% |
| Baccarat, Tie | Eight decks, paying 8 to 1 | About 14.4% |
Three observations from that ranking. The spread between the cheapest and the most expensive standard bet is a factor of nearly thirty, which is far larger than anything your play can influence. The two roulette wheels sit at opposite ends of the sensible range for the identical game, so wheel choice is the largest single decision on that layout; the comparison is laid out on the European versus American roulette page. And blackjack only holds the top position for players who use the chart, which is covered on the blackjack basic strategy page.
These numbers are computed, not observed
Baccarat and roulette figures are produced by enumeration. Because the drawing rules are completely fixed, every possible arrangement can be worked through and the outcomes counted, which is why baccarat probabilities are quoted to two decimal places rather than approximated. Nobody sat and recorded a million hands to arrive at 1.06% for the Banker bet; it falls out of the rules. The full working is on the baccarat odds page.
Blackjack is the exception and its figure carries an assumption. The published number of roughly 0.5% assumes correct basic strategy on every hand under a specific set of rules. Change the rules or change the player and the number changes, which is why blackjack is the one game where a published edge is a conditional statement rather than a property of the table.
The useful consequence is that these figures are checkable rather than promotional. Any European wheel produces 2.70%, anywhere, forever. If a source quotes something different for a standard bet, it is either describing a different rule set or it is wrong, and either way you can identify which by asking what deck count, what payout and what commission it assumed.
A cheap game played fast costs more than an expensive game played slowly
This is the part almost nobody works out, and it is worth doing with real numbers because the result is counterintuitive. Since the edge is charged on turnover, the cost of an hour is stake multiplied by rounds per hour multiplied by the edge.
| Scenario | Turnover per hour | Expected cost per hour |
|---|---|---|
| Baccarat Banker, RM20, speed table at 180 hands | RM3,600 | About RM38 |
| Baccarat Banker, RM20, standard table at 100 hands | RM2,000 | About RM21 |
| American roulette, RM20, live wheel at 30 spins | RM600 | About RM32 |
| European roulette, RM20, live wheel at 30 spins | RM600 | About RM16 |
Read the first and third rows together. The cheapest standard bet in the building, dealt at speed, costs more per hour than the most expensive common wheel dealt at a normal live pace. The 1.06% bet is five times cheaper per ringgit and still ends up costing more, purely because a speed table resolves six times as many rounds.
So the practical lever is not only which bet you choose. It is how many bets you make. Playing a slower table, sitting out rounds, betting every second shoe rather than every hand: each of those reduces turnover directly, and turnover is the thing the price is charged on. Most players optimise the 1% and ignore the 6x.
The same arithmetic explains why table minimums matter so much. At a RM3 table the identical pace produces less than a sixth of the turnover of a RM20 table, which is why matching the table to the bankroll is a financial decision rather than a comfort one. That reasoning continues on the casino table limits page.
Over a real session the edge is almost invisible next to the swings
A 1.06% edge on two hundred RM10 hands is an expected cost of about RM21. The realistic range of results around that figure runs into the hundreds of Ringgit in either direction, because the bet resolves close to a coin flip and runs of eight or ten in a row are ordinary.
Two things follow. A winning night proves nothing about a method, and a losing night proves nothing about a table being cold. Any outcome available in an evening sits comfortably inside normal variance for a near-even bet. And the edge only becomes visible in aggregate, which is precisely why the operator side of the equation is stable while the player side feels like chaos: the operator is running the long-run experiment continuously, and you are sampling it for three hours.
Rule details move the price further than most player decisions do
Two tables that look identical in the lobby can carry meaningfully different prices, and the difference is always printed in the rules panel.
- Blackjack paying 6 to 5 on a natural. Adds roughly 1.4% to the cost of the game, which is more than the entire edge you were playing against at a 3 to 2 table. It is the single worst rule change on the floor.
- Blackjack insurance. A separate bet on the dealer holding a ten in the hole, running at around 7% against the player. The answer is no, including when you hold a strong hand.
- Baccarat tie paying 9 to 1. A genuine improvement over 8 to 1, taking that bet from about 14.4% down to roughly 4.8%. Still several times the price of the main bets.
- Commission-free baccarat. Pays Banker at flat even money but settles a winning Banker six at half stake. That single carve-out costs slightly more than the 5% commission it replaces.
- French roulette rules. La Partage or En Prison return half your even-money stake when the ball lands in zero, cutting those bets from 2.70% to 1.35% on a single-zero wheel.
- The five-number bet. Exists only on the double-zero layout and costs 7.89%, the worst standard bet on either wheel.
Reading the rules panel takes about a minute and is worth more than any amount of table selection based on feel. Two seats in the same lobby, same game, same studio, can differ by more than a full percentage point.
No sequence of bets changes the number, and this is not a matter of opinion
Progressions such as the Martingale, the Fibonacci and the D Alembert all promise the same thing in different clothing: that varying stake size according to what has just happened converts a losing proposition into a winning one. It does not, and the reason is short. Expected value is additive. Each bet has its own negative expectation and the sum of a set of negative numbers is negative, no matter what order they are placed in or what size each one is.
What a progression genuinely does is reshape the distribution. Doubling after a loss produces many small wins and rare large losses, so a short sample usually shows a profit and the system feels like it works. The rare large loss is proportionally bigger than everything collected before it, and it also runs into the table maximum, which is the ceiling that stops the sequence exactly when it needs to continue.
The same applies to reading patterns. Baccarat road maps, roulette hot and cold boards and dragon tiger streak displays are all records of the past, and none of them shifts the probability of the next round. The cards have no memory. The wheel has no memory. A displayed history is a display, not a signal. The honest treatment of every popular system is on the baccarat strategy page.
A turnover-based rebate is the one thing measured in the same units
Most promotions are hard to compare with a house edge because they are denominated in something else. A rebate paid on turnover is the exception: the edge is charged on turnover and the rebate is paid on turnover, so they are directly comparable numbers.
U88 runs a daily cash rebate on live casino turnover with no maximum. The rate is set by member tier: 0.20% at Normal, then 0.30% at Bronze, 0.40% at Silver, 0.50% at Gold, 0.60% at Platinum and 0.80% at Diamond. Against a bet costing 1.06%, the Diamond rate offsets most of the theoretical cost while the Normal rate offsets under a fifth of it. Against a bet costing 14.4%, neither rate registers. None of that makes any game favourable, because rebate terms carry their own qualifying conditions and the edge applies to every round regardless. What it does mean is that the value of a turnover rebate depends on which bet generated the turnover and which tier you sit in, and generating it on expensive bets to collect a rebate is paying fourteen to receive a fraction of one.
What to actually do with the number
Three decisions, in order of how much they matter.
- Pick the bet. Within any game, take the cheapest standard wager and leave the decorated ones alone. That single choice can be a factor of ten.
- Read the rules panel. Wheel type, natural payout, tie payout, commission arrangement. A minute of reading beats an evening of intuition.
- Control the pace. Slower tables, fewer rounds, sitting out. This is the lever with the largest arithmetic behind it and the one almost everybody ignores.
What none of this does is make a game favourable, and it would be dishonest to imply otherwise. Every wager described on this page is negative expectation by design, which is why the operator is in business. The number tells you the price of the entertainment. It never tells you that the entertainment is free. The full game-by-game context is on the live casino hub, and the terms used here, from RTP to hit frequency, are defined in the live casino glossary.
Play responsibly. U88 is strictly for players aged 18 and over. Table games carry a built-in house edge that never switches off. Set a deposit and time limit before you sit down, never chase a loss, and step away if it stops being entertainment.
Frequently asked
What is the house edge in a casino?
Which casino game has the lowest house edge?
Is house edge the same as RTP?
Does playing faster increase the house edge?
Can any betting system overcome the house edge?
Why is the baccarat tie so expensive if it wins often?
Know the price before you take the seat.
Join U88