Banker Wins 50.68% (Excluding Ties): What It Means

If you read two different textbooks on baccarat probabilities, you will often see two seemingly contradictory numbers for the Banker win percentage in baccarat: one source states that Banker wins 45.86% of hands, while another states that Banker wins 50.68% of hands.

Neither number is a typo. Both describe the exact same eight-deck combinatorial distribution, viewed through two different denominators: 45.86% is the Banker hand's raw frequency across all coups dealt (including Ties), whereas 50.68% is the Banker hand's conditional win rate across resolved decisions only (once the 9.52% of tied hands—which push on Banker and Player wagers—are removed). Below, we derive the exact arithmetic behind both percentages, show why a 50.68% decided win rate still carries a −1.06% expected value after the 5% Banker commission, and walk through an 18-hand shoe inside our free Baccarat Calculator & Road Decoder.

1. Two ways to count the same shoe: raw frequency (45.86%) vs resolved decisions (50.68%)

In a standard eight-deck (416-card) Punto Banco shoe, there are 4,998,398,275,503,360 possible ordered six-card deals. Because the Banker hand acts second under the third-card tableau, those deals resolve into three exact buckets:

• Banker wins: 2,292,252,566,437,888 deals = 45.8597% (rounded to 45.86%)
• Player wins: 2,230,518,282,592,256 deals = 44.6247% (rounded to 44.62%)
• Tie: 475,627,426,473,216 deals = 9.5156% (rounded to 9.52%)

What happens when a coup ends in a Tie? Unless you placed a separate side wager on the Tie bet (which carries a 14.36% house edge at 8:1), a Tie is a push. Neither the Banker wager nor the Player wager wins or loses; your chips remain on the felt untouched.

When you want to know how often a Banker or Player wager actually wins given that the hand resolved in a decision, you divide each side's raw win probability by the 90.4844% share of hands that do not tie (45.8597% + 44.6247% = 90.4844%):

Non-Tie (Decided) Share  = 0.45859742 + 0.44624661 = 0.90484403 (90.48%)

Banker Decided Win Rate  = 0.45859742 / 0.90484403 = 0.50682479 (50.68%)
Player Decided Win Rate  = 0.44624661 / 0.90484403 = 0.49317521 (49.32%)

Notice how those two numbers work together: out of every 100 coups dealt at a live table, roughly 9.5 hands push as Ties, and out of the remaining 90.5 resolved hands, Banker wins 50.68% (about 45.86 hands) while Player wins 49.32% (about 44.62 hands).

2. Why does Banker win 50.68% instead of 50.00%?

If baccarat were a symmetrical game—such as flipping a fair coin or dealing two-card totals with identical drawing rules—resolved hands would split 50.00% Banker and 50.00% Player. The entire +0.68-percentage-point shift above 50% (creating a 1.36-percentage-point gap between 50.68% and 49.32%) comes from the asymmetric third-card drawing rules:

• Player acts first without information: Whenever neither side holds a Natural 8 or 9, Player must draw on 0–5 and stand on 6–7 before knowing how Banker will act. Because modulo-10 addition wraps sums of 10–18 down to 0–8, drawing onto a two-card 4 or 5 frequently degrades Player's hand.
• Banker acts second after seeing Player's third card: When Player draws a third card, Banker checks that card's exact point value before deciding whether to draw or stand on 3, 4, 5, and 6. Standing when Player's third card likely wrapped Player past 10—and drawing when Player pulled a strong 6 or 7—lifts Banker from 50.00% to 50.68% of decided hands.

3. Why a 50.68% win rate still produces a −1.06% expected value

Here is the single most important trap to avoid in baccarat math: seeing that Banker wins 50.68% of resolved decisions and assuming that a win rate above 50% gives the player an edge.

A win frequency above 50% only creates a positive mathematical expectation when winning wagers are paid at full even money (1:1). Because standard casinos deduct a 5% commission on winning Banker wagers—paying +0.95 units on a win against −1.00 unit on a loss—the break-even win rate required for a Banker wager is not 50.00%. We can solve for the exact break-even decided win rate (p) where expected value equals zero:

Break-Even Equation at 0.95:1 Payout:
  (+0.95 × p) − (1.00 × (1 − p)) = 0
  0.95p − 1.00 + 1.00p           = 0
  1.95p                          = 1.00
  p                              = 1.00 / 1.95 = 0.5128205 (51.28% of decided hands)

Look closely at that threshold:

• Break-even win rate needed at 0.95:1 payout: 51.28% of decided hands
• Actual Banker win rate in an 8-deck shoe: 50.68% of decided hands
• Shortfall below break-even: 0.60 percentage points (51.28% − 50.68%)

Because 50.68% falls 0.60 percentage points short of the 51.28% break-even line, the Banker wager carries a negative expected value on both a per-hand-dealt basis and a per-resolved-decision basis:

1. Expected Value Per Hand Dealt (Including 9.52% Ties):
   EV = (+0.95 × 0.458597) − (1.00 × 0.446247) + (0 × 0.095156)
      = 0.435667 − 0.446247
      = −0.010580 (−1.06% house edge per coup dealt)

2. Expected Value Per Decided Hand (Excluding Ties):
   EV = (+0.95 × 0.506825) − (1.00 × 0.493175)
      = 0.481484 − 0.493175
      = −0.011691 (−1.17% house edge per resolved decision)

4. Sample-size reality: why a single 70-hand shoe rarely lands on 50.68%

In a computer simulation of ten million shoes, the Banker hand's decided win share converges to 50.68% with razor precision. Inside a single eight-deck shoe at a casino table—which typically deals 70 to 75 coups before the cut card, leaving roughly 65 resolved non-tie hands—you will almost never see 50.68% on the display.

Why? Because 50.68% of 65 decided hands is 32.94 wins (you cannot even deal a fractional hand!), and the binomial standard deviation over n = 65 resolved decisions is:

Standard Deviation (Hands) = √(65 × 0.5068 × 0.4932) ≈ 4.03 hands
Standard Deviation (%)     = 4.03 / 65               ≈ ±6.20% per 1 SD
95% Confidence Band (±2 SD)= 50.68% ± 12.16%         → [38.5% to 62.8%] (25 to 41 Banker wins)

In plain English: in a completely fair eight-deck shoe with 65 decided hands, Banker winning anywhere between 25 hands (38.5%) and 41 hands (63.1%) sits right inside ordinary two-standard-deviation binomial noise. When you open our free Baccarat Calculator (or read How Baccarat Smart Works), the summary panel computes the live 95% Wilson score confidence interval around your shoe's non-tie Banker percentage so you can see immediately whether a lopsided shoe is simply normal small-sample variance.

5. Worked 18-hand shoe (BBBPPBPPBBPBBPPBTB): 58.8% decided Banker share & a Cockroach Pig Dragon Tail

Let's trace both counting conventions and all five casino roads on this engine-verified 18-hand shoe (BBBPPBPPBBPBBPPBTB—10 Banker wins, 7 Player wins, and 1 Tie):

BBBPPBPPBBPBBPPBTB

• Raw frequency (all 18 coups dealt): Banker 10 / 18 = 55.56%, Player 7 / 18 = 38.89%, Tie 1 / 18 = 5.56%.
• Resolved win rate (17 decided coups, excluding Hand 17 Tie): Banker 10 / 17 = 58.82%, Player 7 / 17 = 41.18% (with a 95% Wilson confidence interval of 36.0% to 78.4%, comfortably containing the theoretical 50.68%).

Scoreboard walkthrough: from Big Road columns to the Cockroach Pig Dragon Tail

As explained in our complete guide to reading baccarat roads, the Bead Plate and Big Road use red for Banker and blue for Player, whereas Big Eye Boy (lookback 1), Small Road (lookback 2), and Cockroach Pig (lookback 3) use Red for pattern repetition and Blue for pattern breaks— never Banker or Player. Here is what happens across BBBPPBPPBBPBBPPBTB:

1. Big Road (9 Columns — Depths 3, 2, 1, 2, 2, 1, 2, 2, 2): After Banker opens with a three-hand streak in Column 1 (depth 3), Columns 2 through 9 alternate between depths of 2 and 1 in a repeating rhythm (2, 1, 2, 2, 1, 2, 2, 2). Hand 17 is a Tie (T) dealt immediately after Banker starts Column 9 at Row 1, placing a green slash on Column 9, Row 1 before Hand 18 (B) extends Column 9 to Row 2.
2. Big Eye Boy (Lookback 1 — 13 Entries: R B B B B R R B B B R R R): Big Eye Boy begins on Hand 5 (Column 2, Row 2) using Rule 2 (Case 2A: both drop deeper): Column 1 (depth 3) reaches Row 2, so mark #1 is Red. Whenever a column of depth 2 drops to Row 2 right after a column of depth 1—at Column 4, Row 2 (mark #4) and Column 7, Row 2 (mark #9)—Column c−1 stopped one row above at Row 1 (Rule 2, Case 2B: first drop below the match), producing Blue. Then across Columns 8 and 9 (depths 2, 2 following Column 7's depth 2), Big Eye Boy records three straight Red marks (marks #11, #12, and #13).
3. Cockroach Pig (Lookback 3 — 10 Entries: R B R R R R R R R B) & Its 7-Mark Dragon Tail: Look at Big Road Columns 2 through 8: their depths are 2, 1, 2, 2, 1, 2, 2—a period-3 rhythm that repeats every three columns! Because Cockroach Pig compares each column with the column three positions to the left (L=3), starting at Column 4, Row 2 (mark #3, where Column 1 of depth 3 also reached Row 2) and continuing through Columns 5, 6, 7, and 8 (matching Columns 2, 3, 4, and 5), Cockroach Pig records seven consecutive Red marks in a row (marks #3 through #9). Because the derived-road grid is 6 rows tall, the sixth Red mark reaches the bottom at Row 6 and the seventh Red mark turns 90 degrees to the right along Row 6 to form a Dragon Tail—until Column 9, Row 1 (comparing Column 8 depth 2 vs Column 5 depth 2 for mark #9 Red, followed by Column 9, Row 2 checking Column 6 of depth 1 via Case 2B) finally breaks the streak with Blue (mark #10)!
4. Next-Hand Preview (Hand 19): Column 9 currently sits at depth 2, while Columns 8, 7, and 6 have finished depths of 2, 2, and 1:
   • If Hand 19 is Banker (Column 9, Row 3 — Rule 2): Big Eye Boy checks Column 8 (depth 2, stopped one row above at Row 2, Case 2B) → Blue; Small Road checks Column 7 (depth 2, stopped one row above at Row 2, Case 2B) → Blue; Cockroach Pig checks Column 6 (depth 1, already stopped further up at Row 1 so both Rows 2 and 3 are blank, Case 2C) → Red.
   • If Hand 19 is Player (Column 10, Row 1 — Rule 1): Column 9's finished depth (2) matches Column 8 (2) and Column 7 (2) but differs from Column 6 (1), so Big Eye Boy = Red, Small Road = Red, and Cockroach Pig = Blue.

Analyze shoe distributions in the free Baccarat Calculator

Step through BBBPPBPPBBPBBPPBTB in the interactive shoe sandbox right below this article, or open /calculator to inspect live non-tie win percentages, 95% Wilson confidence intervals, and six D3 randomness charts on any shoe you enter. Want guided lessons and timed drills on every road rule? Explore our full course curriculum and start our free Module 1 interactive preview.