Dragon Tiger Winning Strategies
Learn Dragon Tiger betting systems and what really works.
Dragon Tiger Strategies: What Works and What Doesn't
Quick Navigation
Dragon Tiger has no decisions after the bet. There is no card to draw, no hand to stand on, no dealer rule to exploit. That makes the honest strategy page short: nothing you do during a round changes the house edge, and the only choices that matter are made before it. This page works through the systems people use, shows what each one actually does to the numbers, and then lays out the few decisions that genuinely move your expected result.
The Three Choices That Matter
1. The table's tie rule
Half your Dragon or Tiger stake back on a tie: 3.73% house edge. All of it lost: 7.47%. Same game, double the cost.
Pragmatic Play, SA Gaming, On Air Entertainment and Deltin publish half back. Evolution, Ezugi and OneTouch do not state it; check the help screen.
2. Which bet you make
Dragon or Tiger: 3.73%. Tie at 11:1: 10.36%. One-card side bets: 7.69%. Suited Tie: 13.98%. Tie at 8:1: 32.77%.
Every bet other than Dragon or Tiger costs at least twice as much per dollar, and the 8:1 Tie nearly nine times as much.
3. How much passes through
Expected loss = edge × total wagered. At about 140 rounds an hour, a $10 flat bet is $1,400 an hour through the table and about $52 an hour lost on a half-back table.
Speed and Super Speed tables raise the hourly figure without changing the edge.
Choose the half-back table, bet only Dragon or Tiger, and decide in advance how many dollars you are willing to push through it. That is the entire strategy that mathematics supports. Everything below explains why the alternatives do not add to it.
Betting Systems, Tested Against the Numbers
A betting system changes how much you stake from round to round. It cannot change the expected value of any single round, which is −3.73% of the stake on a half-back table whether the stake is $1 or $1,000. Since the total expected loss is that rate multiplied by everything wagered, a system can only redistribute when the losses arrive, not reduce them. You can put your own stake and number of rounds into our probability calculator to see that expected loss for any table rules, or watch it happen in the classroom mode of our Dragon Tiger Simulator.
Martingale: double after every loss
Bet 1 unit on Dragon; after each loss, double; after a win, return to 1. Each completed sequence wins exactly 1 unit, so the system feels like it works, because most sequences complete. Counting a tie as a lost round (it costs half your stake and does not end the sequence), the chance of losing n rounds in a row is 0.537349n:
| Consecutive losses | Probability | About once every | Units lost when it happens | Next bet required |
|---|---|---|---|---|
| 6 | 2.41% | 42 sequences | 63 | 64 |
| 7 | 1.29% | 77 sequences | 127 | 128 |
| 8 | 0.70% | 144 sequences | 255 | 256 |
| 9 | 0.37% | 268 sequences | 511 | 512 |
| 10 | 0.20% | 498 sequences | 1,023 | 1,024 |
Starting at $10, the eighth doubled bet is $2,560. Most tables will not accept it: Evolution's Dragon Tiger at Stake, for example, lists maximums between $500 and $25,000 depending on the room, and the low end is reached after the sixth loss. When the limit or the bankroll stops the doubling, the sequence loses everything it has staked, and one such event erases dozens of one-unit wins. The house edge on every dollar wagered was 3.73% throughout.
Paroli and 1-3-2-6: press after wins
Positive progressions raise the stake after wins and reset after a loss. They cap the damage of a losing run at one base unit per loss, which makes them far safer than Martingale, but the extra units staked after wins are wagered at the same −3.73%. Expected result per dollar: unchanged. What changes is variance: bigger, rarer wins and more frequent small losses.
D'Alembert: add one after a loss, subtract one after a win
A gentler negative progression. Because Dragon and Tiger win less than half the time (46.27% each), the “one win cancels one loss” logic breaks: over a long session there are more losses than wins to cancel, and the average stake drifts upward. Expected loss per dollar is still 3.73%; the total wagered, and therefore the total lost, grows.
Fibonacci: one step forward after a loss, two back after a win
Bets follow the sequence 1, 1, 2, 3, 5, 8, 13 and so on. A loss moves one step up the sequence, a win moves two steps back, so one win recovers the two losses before it and the climb is gentler than Martingale's doubling. The bill still arrives: in our tester's check (40,000,000 rounds on Dragon, half the stake lost on a tie, sessions of 100 rounds, table maximum 500 units) Fibonacci finished 90.2% of sessions ahead, its worst session lost 2,679 units, and it cost 3.74% of every chip wagered against the exact 3.73%.
Labouchere: cross off the line
Write a line of numbers, say 1 2 3 4 5, and bet the first plus the last. A win crosses both off; a loss adds the lost bet to the end of the line. Finish the line and you have won its total, 15 units. A losing run makes the line longer and the bets larger with no ceiling but the table's. In the same check Labouchere finished 74.7% of sessions ahead, its worst session lost 6,725 units, its bets reached the 500-unit maximum, and it cost 3.67% per chip wagered: random noise around the same 3.73%.
Following the roads
Every studio shows bead roads, and Pragmatic Play offers an Ask Dragon / Ask Tiger view that projects what the roads would look like after each outcome. They describe the past. Cards are dealt without regard to it: after five Dragons in a row, P(Dragon) on the next card is still 46.27%. Betting the streak, betting against it, and betting at random have identical expected value; our Dragon Tiger Predictor Tool lets you watch eight ways of calling the round, roads included, meet the same coin toss.
Hedging Dragon and Tiger together
Betting both sides guarantees a loss: on the 92.53% of rounds that do not tie, one bet wins what the other loses, and on the 7.47% that tie you lose half of both. Expected cost is 3.73% of the combined stake for zero chance of profit. Pragmatic Play's help notes the table may block the combination outright.
Try any of them yourself: the betting system tester on our simulator plays all seven, flat betting included, on the same cards for up to 100,000 rounds and reports sessions ahead, worst session and cost per chip for each.
Counting the 7s
Dragon Tiger is sometimes described as a game where card counting matters. That is half true, and the half that is true is small.
The main bets: not countable in practice
Dragon and Tiger are symmetric no matter what remains in the shoe; the edge is ½ × P(Tie), and P(Tie) moves only slightly as ranks deplete (7.47% at a full eight-deck shoe, 7.40% at six decks). No composition of a real shoe turns the main bet positive.
The 7-trap side bets: countable, but the count is the 7
On a Big/Small, Odd/Even or Red/Black bet that loses on a 7 (Pragmatic Play, On Air Entertainment), the house edge at any moment equals the probability that the next card is a 7, plus or minus any imbalance between the two winning groups. With a fresh shoe that is 32 of 416 cards: 7.69%. As 7s leave the shoe the edge falls; if the two groups stay balanced, the bet is fair only when every 7 is gone.
| Half a shoe remaining (208 cards) | 7s remaining | House edge on a balanced Big/Small |
|---|---|---|
| Average depletion | 16 | 7.69% |
| 7s running out | 12 | 5.77% |
| Very rich shoe | 8 | 3.85% |
| Rare | 4 | 1.92% |
| All 7s dealt | 0 | 0.00% |
Even in the fourth row the bet is still worse than simply betting Dragon. A player advantage requires the 7s to be gone and the remaining cards to lean toward the group being bet, a state that is rare late in a shoe and exactly what the operators guard against: Pragmatic Play disables side bets after 50 rounds of each shoe, and every studio cuts the shoe well before the end. So the answer to “can you count Dragon Tiger?” is: you can count the side bets, you cannot get to a profitable one, and the table will not let you stay long enough to try.
The Lightning Fee
Lightning Dragon Tiger adds a 20% fee to every bet: a $5 bet costs $6. After betting closes, one suit is assigned a random multiplier from 2x to 8x; a winning card in that suit is multiplied, and a Tie or Suited Tie with two Lightning cards is multiplied twice, up to 1280x. Whether the multipliers repay the fee depends on the base payout table and on how often each multiplier appears, and Evolution publishes neither. We therefore publish no house edge for Lightning Dragon Tiger, and we treat any site that does as guessing. What is not in doubt: you pay 20% more per bet to find out, and the expected value of that surcharge is not yours to compute. The detail, and the questions Evolution has not answered, are in our Lightning Dragon Tiger article.
Where to Play Dragon Tiger for Real Money
The three lobbies we would start with. Whichever table you open, check its tie rule in the game help first: half back costs 3.73%, all lost 7.47%.
Bitcasino.io
100% up to 1,500 USDT (30x)
Accepts US players: No
Play at Bitcasino.io
Sportsbet.io
100% up to 300 USDT (40x)
Accepts US players: No
Play at Sportsbet.io
Jackpot City
Varies by country
Accepts US players: No
Play at Jackpot City
Affiliate links. Every figure above comes from the operator's own pages – see the full comparison for terms, minimums and how we verify what we publish. Yes* means some US states are excluded.
Pace and Bankroll
The variable most players ignore is speed. A standard live baccarat round runs about 48 seconds, roughly 75 hands an hour (Betway's live baccarat page quotes 48 seconds against 27 for Speed Baccarat); Evolution's standard Dragon Tiger runs about 25 seconds, roughly 140 rounds an hour, and its Speed and Super Speed versions more. Expected loss is the edge multiplied by everything you wager, so a 3.73% edge at 140 rounds an hour costs six to seven times as much per hour as baccarat's 1.06% Banker bet at 75. To see that cost for your own stake, table rules and session length, put them into our Dragon Tiger calculator.
- Set the limit in dollars lost, not in time or rounds. Rounds pass faster than you will notice.
- Flat-bet. A flat bet has the lowest total wagered for a given number of rounds, and therefore the lowest expected loss.
- Treat any welcome bonus as slots money unless the terms say otherwise. Where operators publish a live-table contribution at all, Dragon Tiger's is often not stated separately from baccarat's; see the notes on each card of the casino comparison.
- If the point is entertainment, choose the table with the most going on, not the highest limit. If the point is to win money, the numbers on this page say to pick a different game.
The Honest Summary
- Works: choosing a half-back table (3.73%, not 7.47%); betting Dragon or Tiger only; flat betting; a dollar loss limit set before the first card.
- Neutral: every betting progression, road-reading, streak-following, side-switching. They rearrange when you lose, not whether.
- Harmful: the Tie bet (10.36% at best, 32.77% at 8:1), the Suited Tie (13.98%), the one-card side bets (7.69%), hedging both sides, and Martingale at any table with a maximum.
- Unknowable: the true cost of Lightning Dragon Tiger, until Evolution publishes its multiplier table.
If gambling has stopped being entertainment, the Responsible Gaming page lists confidential helplines by country.