Baccarat strategy: what the odds actually allow
Baccarat strategy comes down to one honest sentence: bet Banker, skip the Tie, and manage how much you put at risk. There is no third card to choose and no decision to optimise, so everything else marketed as a system is a way of arranging your bets, not a way of changing them.
On this page
The only decision on the table
In blackjack or poker you can play a hand well or badly. In baccarat you cannot. Once the chips are down, the dealer follows a drawing rule that is identical on every hand, in every casino, for every player. Nothing you do between the deal and the result changes the outcome.
Each round deals two cards a side, and a two-card total of eight or nine is a natural: it ends the hand immediately and no third card is drawn by either side. Everything else follows a fixed table.
That leaves exactly three levers: which bet you take, how much you stake, and how long you sit there. There are only three betting options on the layout (Banker, Player and Tie), and only the first lever has a mathematically correct answer, which is the same answer every time. If you have not seen how the drawing rules work, the baccarat rules page walks through them card by card.

Bet selection, ranked by cost
These are 8-deck figures, which is what almost every live and online table uses.
| Bet | House edge | Cost per RM100 wagered, long run | Verdict |
|---|---|---|---|
| Banker | 1.06% | about RM1.06 | The bet to make |
| Player | 1.24% | about RM1.24 | Acceptable, slightly worse |
| Tie (8:1) | 14.36% | about RM14.36 | Avoid |
| Player / Banker Pair (11:1) | around 10% | around RM10 | Avoid |
That is the whole of baccarat strategy that carries any real weight. Stick to Banker on every hand and you are simply paying the cheapest fee the game charges; the gap between it and the Tie bet is larger than the gap between good and bad play in most casino games.
The potential payout is what makes the Tie tempting, and it is also the riskiest bet on the layout. At 8:1 it looks generous until you notice it wins under one hand in ten, which is why its house edge is more than ten times Banker's.
People occasionally argue for alternating between Player and Banker to keep the game interesting. That is a fine reason. It is simply worth knowing it costs you a fraction of a percent, not that it does anything useful.
Why the 5% commission still leaves Banker ahead
Banker winnings are paid at 1:1 minus a 5% commission, so RM100 returns RM95. New players read that as a penalty that must surely wipe out the advantage. It does not.
The Banker hand draws its third card after seeing what the Player hand drew. That one asymmetry is worth roughly 50.7% of all decided hands, against 49.3% for Player. Left uncommissioned, the bet would be profitable for the customer, which is why the commission exists at all. After it, the Banker edge is 1.06% against Player's 1.24%: smaller, but still the better side.
Two practical notes follow from this. First, the commission reduces every win, which is why doubling-up systems recover less than they appear to. Second, a no-commission table is not a loophole: it pays Banker wins with a total of 6 at half, and that swap makes the Banker bet a touch more expensive than the standard game rather than free.

The roadmaps: what each grid actually records
Walk past any baccarat table in Genting or open any live-dealer stream and you will see a wall of coloured dots and slashes. Those are the roadmaps (路), and they matter enormously to how the game is played socially in Malaysia and across Chinese-speaking Asia. They are worth understanding properly, because most explanations either treat them as mystical or dismiss them without saying what they are.
Every road is a record of results that have already happened. Red always means Banker, blue always means Player, green always means Tie. Here is what each one tracks:
| Road | Name | What it records |
|---|---|---|
| 珠盘路 | Bead Plate / Bead Road | The raw results, one mark per hand, filled top to bottom then left to right. Red for Banker, blue for Player, green for Tie, with small corner dots marking pairs. This is the shoe's transcript, nothing more. |
| 大路 | Big Road | The same results rearranged so streaks are visible. Each new column starts when the winning side changes; a run of Banker wins stacks downward in one column. Ties are drawn as a green line through the current mark rather than a new entry. A column that hits the bottom bends right and becomes a dragon, 长龙. |
| 大眼仔路 | Big Eye Boy | Derived entirely from the Big Road, starting one hand after the first entry of its second column. It does not record who won. It records whether the Big Road is behaving repetitively (red) or choppily (blue) compared with the column before it. |
| 小路 | Small Road | The same repetitive-or-choppy question, but comparing against the column two places back instead of one. It begins after the first entry of the Big Road's third column. |
| 曱甴路 | Cockroach Road / Cockroach Pig | Same idea again, skipping three columns back, and starting after the first entry of the Big Road's fourth column. Drawn as diagonal slashes rather than circles. |
So the last three roads are not separate information at all. They are three different summaries of the shape of the same Big Road, which is itself a redrawing of the Bead Plate. Nothing enters the system except results that have already been paid out.
What the roads can and cannot tell you
Reading the roads well is a genuine skill, and players who do it are not being foolish. It gives the game its rhythm, it is most of what people are talking about when they talk about baccarat, and spotting a dragon forming is one of the pleasures of the game. None of that is worth mocking.
What the roads cannot do is tell you what happens next. The payouts do not change when a column gets long. The drawing rules do not change. The cards remaining in the shoe do not know which colours were drawn on the display board, and the display board is not consulted before the next card comes out.
The instinct that a long Banker run must be due to break — or conversely that it must continue — is the gambler's fallacy, and it is worth naming plainly rather than hinting at. After six Banker wins, the seventh hand still resolves under the same fixed rules, and the chance of each result is essentially what it was on the first round of the shoe. There is a technical footnote: unlike a roulette wheel, a shoe is finite, so dealt cards do slightly change what is left. That is why baccarat card counting exists as a subject at all. Published analyses consistently find the effect far too small and too rare to be worth acting on, and crucially, none of the roads measure it: they track who won, not what cards left the shoe.
The honest summary: use the roads for the enjoyment of the game and its social side. Do not size a bet from them, and be suspicious of anyone selling a prediction tool built on them.

Martingale, Paroli, Fibonacci and the rest
Progressions are the other half of what gets sold as a baccarat betting strategy. They are worth understanding because the way they fail is instructive.
Martingale is the famous one: double your stake after every loss, so the first win recovers everything plus one unit. It works exactly as advertised right up until it does not. Two things break it. Table limits: a table with a 100× spread between minimum and maximum allows only about seven doublings before you physically cannot place the next bet. And bankrolls are finite in a way the maths of the system assumes they are not.
Put a number on it. Backing Banker, you lose a decided hand about 49.3% of the time, so seven consecutive losses arrive roughly once in every 140 attempts. When it does, you are down 127 units against a string of one-unit wins, and on Banker the 5% commission means each recovery comes back slightly short of whole.
Paroli reverses it, doubling after wins instead of losses, which caps your downside and makes losing streaks cheaper and winning streaks rarer but larger. Fibonacci climbs more gently and therefore loses more slowly on the way to the same place. 1-3-2-6 and the 1-3-2-4 just prescribe a fixed shape for a run of four bets.
All of them share one property: they change the distribution of your results without touching the expected value. Every individual bet still carries a 1.06% edge for the house, and a sum of negative-expectation bets is negative no matter what order you arrange the stakes in. A system can make a session more likely to end in a small win and less likely to end in a small loss, at the price of a rare large one. It cannot make the edge go away, and any product promising a guaranteed or sure-win formula is describing something that does not exist.
Bankroll and table choice, which do help
Nothing above lets you beat the game. Several things let you lose more slowly and enjoy it more, which is a reasonable goal in itself.
- Flat bet. The same stake every hand keeps your exposure predictable and removes the scenario where one bad run takes the whole bankroll.
- Set a loss budget before you start, in ringgit, and treat it as the price of the session rather than a target to win back.
- Take a table minimum you can bet a hundred times. Baccarat is fast: a mini-baccarat table can get through more than a hundred rounds an hour, and total money wagered is what the house edge acts on, not the size of any single bet.
- Skip the side bets. Pair bets, Lucky 6 and dragon bonuses cost several times what the main bets do.
- Prefer standard commission tables over no-commission ones, which quietly cost slightly more.
One practical note if you are playing online rather than in a pit: a loss budget only means anything if whatever is left can come back out, so the withdrawal conditions are worth reading before the deposit rather than after. How money actually moves in and out of a casino account is covered separately.
If you want a game where decisions genuinely change the result, the big two card game rewards skill in a way baccarat structurally cannot. The cards you choose to play and hold matter on every turn.

A last word on limits
Gambling in Malaysia is restricted under the Common Gaming Houses Act 1953 and the Betting Act 1953, and is additionally prohibited for Muslims under Sharia law. This page explains the mathematics of a game; it is not a suggestion that you play.
Wherever baccarat is legally available to you, the house edge is the fee for the entertainment and it is charged on every hand whether you win or lose that hand. Decide what the evening is worth to you in advance, keep that money separate from anything you need, and stop at the number rather than at a feeling. If chasing losses has started to feel normal, that is the signal to step away. 18+.
What is the best bet in baccarat?
Banker, on every hand, at a house edge of 1.06% after the 5% commission. Player is next at 1.24%. Nothing else on the layout is close, and no sequence of bets improves on simply taking the cheapest one each time.
Do baccarat roadmaps predict the next hand?
No. The Bead Plate, Big Road, Big Eye Boy, Small Road and Cockroach Road are all records or summaries of results that have already been settled. They are genuinely interesting to read and they carry no information about the hand that has not been dealt yet.
Does the Martingale work in baccarat?
Not as a way of beating the game. It converts many small wins into one occasional large loss. Table limits cap the doubling at around seven steps on a typical table, a seven-loss run on Banker turns up roughly once in 140 attempts, and the 5% commission means each recovery falls slightly short of even.
Is there any sure-win baccarat formula?
No. Every bet available carries a fixed negative expectation, and no arrangement of stake sizes or pattern reading changes that. Anything sold as a guaranteed system, prediction app or sure-win formula is selling something that cannot exist.
Is card counting worth trying in baccarat?
In practice, no. The shoe is finite so composition does drift slightly, which is why the topic exists, but the achievable advantage is tiny, appears rarely and near the end of a shoe, and is nothing like the counting edge available in blackjack.
Should I play no-commission baccarat to avoid the 5%?
No. The commission is replaced by paying Banker wins with a total of 6 at half, and that trade makes the Banker bet slightly more expensive than on a normal commission table.
