Baccarat Strategy: What Actually Works, and What’s Myth
Researched with AI tools, checked against the sources linked on this page and approved by our editor before publishing. How we work.
Baccarat strategy comes down to two decisions: which bet you back, and how much money you put through the table. Roads, streak systems, Martingale, card counting and prediction apps do not change the house edge. Picking the right main bet and refusing the Tie and side bets keeps your cost near 1.06% of what you stake; the Tie at 8:1 costs 14.36%, and some side bets nearly 30%. If the card rules are new to you, start with how to play baccarat.
The short version
| Tactic | Does it work? | What it actually does |
|---|---|---|
| Bet Banker on a standard commission table | Yes | Lowest main-bet edge: about 1.06% |
| Bet Player on a Super 6 table | Yes | 1.24%, against 1.46% for Banker when a winning Banker six pays 1 to 2 |
| Skip the Tie | Yes | Avoids 14.36% at 8:1 (4.84% at 9:1) |
| Skip the side bets | Yes | Avoids 2.65% to 29.98% in the examples below |
| Play fewer, smaller hands | Yes | Cuts turnover, the only cost you control |
| Flat betting | Neutral | Same edge; keeps each bet’s exposure fixed |
| Reading the roads (ikut jalan) | No | Records past results; does not predict the next hand |
| Martingale and other progressions | No | Same edge; much larger stakes after a losing run |
| Card counting | No | Trims the edge to about 1% at best; nothing to count after a reshuffle |
| “RTP hari ini” and prediction apps | No | A baccarat bet’s return is set by its rules, not a daily setting |
All figures assume eight decks, 5% commission on Banker wins and the payouts stated.
Match the bet to the table
| Table type | Banker rule | Banker edge | Player edge | Cheaper bet |
|---|---|---|---|---|
| Standard (commission) | Wins pay 1:1 less 5% | 1.06% | 1.24% | Banker |
| Super 6 (half-pay six) | Winning Banker 6 pays 1 to 2 | 1.46% | 1.24% | Player |
“Always Banker” is right on a standard table and wrong on a Super 6 one. A winning Banker six comes up about 5.39% of the time, and halving those wins costs Banker more than the 5% commission it replaces (Wizard of Odds gives 1.46%). Not every no-commission game works this way: EZ Baccarat pushes a Banker win with a three-card 7, which by our calculation leaves Banker at about 1.02%. Read the table’s rules panel; if it does not state the commission and payouts, treat the house edge as unknown. The RTP glossary explains what these percentages cost over a session.
Refuse the Tie
A Tie (seri) lands in about 9.52% of rounds, often enough to feel catchable, and 8:1 is the biggest number on the main layout. At that payout the house edge is 14.36%, more than 13 times Banker’s. Some tables pay 9:1, which cuts the edge to 4.84%, still more than four times Banker’s.
Side-bet names are not enough
These are eight-deck examples from published mathematical analysis, not offers confirmed at any casino. Payouts are profit odds; a different qualifying event or payout needs a new calculation.
| Side bet | Qualifying result and profit payout | House edge |
|---|---|---|
| Player Dragon Bonus | Player wins; full schedule below | 2.65% |
| Banker Dragon Bonus | Banker wins; full schedule below | 9.37% |
| Perfect Pair | First two cards of either hand match rank and suit; 25:1 | 13.03% |
| Player or Banker Pair | First two cards of the chosen hand match rank; 11:1 | 10.36% |
| Dragon 7 (EZ Baccarat) | Winning Banker hand has three cards totalling 7; 40:1 | 7.61% |
| Panda 8 (EZ Baccarat) | Winning Player hand has three cards totalling 8; 25:1 | 10.19% |
| Super 6 side bet | Winning Banker hand totals 6; 12:1 | 29.98% |
Dragon Bonus: a natural win pays 1:1 and a natural tie pushes; otherwise a winning margin of 4, 5, 6, 7, 8 or 9 pays 1:1, 2:1, 4:1, 6:1, 10:1 or 30:1, and other outcomes lose. The Pair figure is for 11:1; at 12:1 the same bet drops to about 2.89%, because a same-rank pair comes up 31 times in 415. The Super 6 side bet is separate from the half-pay main-bet rule above. Put in ringgit, a RM10 Super 6 side bet carries the same expected cost as about RM283 bet flat on Banker. Source: Wizard of Odds’ side-bet analysis; we recomputed the Pair, Dragon 7, Panda 8 and Super 6 rows.
Control turnover: the only variable you own
| Assumed dealt hands | Fixed stake | Turnover | Expected cost at 1.06% |
|---|---|---|---|
| 60 | RM10 | RM600 | RM6.36 |
| 150 | RM10 | RM1,500 | RM15.90 |
| 150 | RM50 | RM7,500 | RM79.50 |
The hand counts are round-number assumptions, not timings we measured. Expected cost follows turnover, so speed tables and multi-table play are spending decisions. Flat betting does not reduce the edge, but it keeps each bet’s exposure fixed and your expected cost in line with hands played. Our guide to stake sizes and stop rules shows how to set a flat bet against a session budget.
Road maps and streaks
Live tables show a scoreboard: the bead plate, the Big Road and derived roads such as Big Eye Boy, Small Road and Cockroach Pig, as Wizard of Odds’ guide to the scoreboards explains. Following them is called ikut jalan. They are an accurate record of past results and no guide to the next hand: the same analyst calls trying to predict the next hand from them “a waste of time”. Following a streak and betting that it will break are the same mistake in opposite directions.
The roads feel predictive because long runs are more common than intuition expects. In a simplified model of independent hands using the eight-deck Banker/Player split, ties set aside:
| A run on one side somewhere in the shoe | 74 resolved hands | 60 resolved hands |
|---|---|---|
| 4 or more in a row | ~99% | ~99% |
| 5 or more | ~93% | ~88% |
| 6 or more | ~70% | ~62% |
| 7 or more | ~43% | ~36% |
| 8 or more | ~24% | ~20% |
How many hands a shoe resolves depends on where it is cut. A six-long run appears in most shoes; it is ordinary, not a signal.
Martingale on Banker: a win can still leave a loss
Assume a RM10 first stake, doubling after each full loss, 5% commission on winning Banker bets, and ties pushing without advancing the step. The net result after the first win is:
| Win on step | Stake | Losses already taken | Net result |
|---|---|---|---|
| 1 | RM10 | RM0 | +RM9.50 |
| 2 | RM20 | RM10 | +RM9.00 |
| 3 | RM40 | RM30 | +RM8.00 |
| 4 | RM80 | RM70 | +RM6.00 |
| 5 | RM160 | RM150 | +RM2.00 |
| 6 | RM320 | RM310 | −RM6.00 |
| 7 | RM640 | RM630 | −RM22.00 |
| 8 | RM1,280 | RM1,270 | −RM54.00 |
From step six, a “successful” recovery still finishes below break-even, before any table limit or bankroll problem. Ignoring ties, seven Banker losses in a row happen about 0.71% of the time, roughly once in 141 sequences, and cost RM1,270 while chasing RM9.50.
Over a fully funded seven-step sequence, the average outcome is about −RM0.79 on about RM74.26 staked, counting stakes returned on ties: exactly the same 1.06%. Our bankroll guide finds the same on a roulette wheel (−RM0.87 on about RM32.10, or 1/37). Doubling changes how the loss arrives, not its rate.
Compare two players with RM1,500 each. A RM10 flat bettor risks RM10 a hand; the Martingale player wins small amounts in most sequences and, about once in 141, loses RM1,270, 85% of the bankroll. With RM500, the RM320 sixth stake cannot even be placed.
Other progressions
Paroli, 1-3-2-6, Fibonacci and Labouchere are rules for choosing the next stake. None changes the expected loss per ringgit staked; they change the shape of results (many small losses and occasional larger wins, or the reverse), not the edge.
Card counting
In a blackjack shoe dealt deep enough, counting can help because you can bet more and play differently when the remaining cards favour you; our blackjack guide explains why it fails against RNG and continuously shuffled games. In baccarat you make no playing decisions, and the effects of removing cards are tiny. Wizard of Odds’ counting analysis finds that choosing Banker or Player by the count trims the combined house edge from 1.06% to about 1.01%, still a house edge. Its effect-of-removal study concludes that “for all practical purposes baccarat is not a countable game”, citing Peter Griffin’s estimate of about 70 US cents an hour betting US$1,000 only when the count is favourable. Edward Thorp, quoted on the same site, called practical counting strategies “at best marginal, and at best precarious”, because shuffling early wipes them out. Against a continuous shuffler, or a game that reshuffles before every hand, there is nothing to count.
“RTP baccarat hari ini”, bocor tables and prediction apps
A baccarat bet’s return is fixed by the drawing rules and the paytable. There is no daily RTP setting or “hot table” configuration for anyone to leak, and past results do not predict the next hand. Treat any Telegram channel, agent or app selling today’s baccarat RTP, a bocor table or a guaranteed jalan as a front for a deposit link, a subscription fee or a login-harvesting page, and never share banking credentials. Our Scam Watch and blacklist explain particular warning signs. If you have already sent money, call your bank and the NSRC on 997 straight away, then lodge a police report anyway: a 997 call does not replace it (NFCC’s NSRC page).
Where you play matters more than any system
A staking system cannot cost you a withdrawal; an operator that will not pay can. Before depositing, check the operator’s licence on the regulator’s own register and read its withdrawal caps and bonus terms. Our Malaysia comparison and live-casino guide cover operators separately; a ranking is not a payout guarantee.
Malaysian law and support
Playing can be an offence for you. Section 6(1) of the Common Gaming Houses Act 1953 and section 6(1) of the Betting Act 1953 penalise players, each with a fine up to RM5,000, up to six months’ imprisonment, or both. In October 2023 the Court of Appeal upheld convictions for online gambling on laptops (Free Malaysia Today). An offshore licence is not a Malaysian one. See our Malaysian gambling-law guide for context.
18+ only. Every baccarat bet carries a house edge, so continued play is expected to lose; nothing here is a method for winning. Set a budget before you play, never use borrowed or essential money, and stop at your limit or earlier. Befrienders KL offers emotional support on +603-7627 2929, 24 hours, and Gambling Therapy offers free online support. If you keep breaking your limits, our self-assessment and responsible gambling resources can help.
Frequently Asked Questions
Hanya dua perkara yang membantu: memilih taruhan yang betul (Banker di meja komisen, Pemain di meja Super 6) dan mengawal jumlah wang yang anda pertaruhkan. Sistem seperti Martingale atau ikut jalan tidak mengubah kelebihan rumah.
Satu sesi boleh berakhir untung, tetapi papan road hanya merekod keputusan lepas dan tidak meramal tangan seterusnya. Dalam model ringkas kami, rentetan enam kali berturut-turut muncul dalam kira-kira 6 hingga 7 shoe daripada 10, jadi ia perkara biasa, bukan isyarat.
Banker wins pay 0.95 to 1, so a win recovers only 95% of the new stake. From the sixth step, a “successful” recovery still finishes below break-even, before table limits or your bankroll even come into it.
No. Past ties do not make the next one more likely, and the Tie costs about 14.36% at 8:1 and 4.84% at 9:1, more than Banker either way.
Pada bayaran 11:1, Player Pair atau Banker Pair (lapan dek) mempunyai kelebihan rumah kira-kira 10.36%, hampir sepuluh kali ganda taruhan Banker; pada 12:1 angka itu turun kepada kira-kira 2.89%. Semak jadual bayaran meja kerana taruhan dengan nama yang hampir sama boleh berbeza.
Neither changes the expected cost per ringgit staked. Paroli gives more small losses and occasional larger wins, Martingale the reverse, so choose by the experience you prefer, not by any expected advantage.
Not in practice. Wizard of Odds’ analysis finds card counting trims the edge only from 1.06% to about 1%, and Griffin put the profit from his count at about 70 US cents an hour betting US$1,000 when favourable. A continuous shuffler or a reshuffle before every hand leaves nothing to count.
Tidak. Pulangan taruhan baccarat ditetapkan oleh peraturan permainan dan jadual bayaran, bukan oleh tetapan harian. Sesiapa yang menjual “RTP hari ini” atau meja bocor untuk baccarat menjual sesuatu yang tidak boleh meramal tangan seterusnya; lihat Scam Watch kami.
It makes no mathematical difference if the new table has the same rules. Switching helps only if the new table has better rules, such as standard commission for a Banker bettor or a 9:1 Tie, and stopping is always an option.
Sources and limits
Desk research and our own arithmetic, not casino play, deposit or withdrawal testing. Main-bet and side-bet figures come from Wizard of Odds’ baccarat analysis and the pages linked above; we reproduced the main bets, streak table and Martingale averages ourselves from an eight-deck shoe. The streak and Martingale figures use the simplified independent-hand model stated with them. All sources checked 13 October 2026. Figures describe the stated rules, not every table or your likely session result.
Editorial Note: Felix McCoy
Read more articles and editorial information on the author profile.