Scott Delekta: Winning Strategy for Dragon Bonus Baccarat

[Scott Delekta]

The Dragon Bonus Bet is a side bet for the table game of baccarat. The paytable is shown below:

Win by 9 points (non-natural) 30 to 1
Win by 8 points (non-natural) 10 to 1
Win by 7 points (non-natural) 6 to 1
Win by 6 points (non-natural) 4 to 1
Win by 5 points (non-natural) 2 to 1
Win by 4 points (non-natural) 1 to 1
Natural winner 1 to 1
Natural tie Push
All other Loss

(8 Deck shoe)

House Edge Std Deviation
Banker 2.65% 2.30
Player 9.37% 2.47

Through the use of computer simulation this game can be beaten with card counting. Here is the first count I developed:

Dragon Master Count

Card Rank Count Value

Ace +2

2 +3

3 +3

4 +2

5 +1

6 +0.5

7 -2

8 -2

9 -2

10,J,Q,K -0.5

The simulation model used an 8 deck shoe and the shuffle point was 401/416. This is common penetration for live baccarat (14-16 cards cut position) since the primary game cannot be beaten with card counting. The Running Count (RC) begins at 0.

Here are the simulation results:
Each data point (min 200 million shoes simmed)

Dragon Player Bet (99% confidence level E =0.02%)

RC % occurence payout %

>=97 1.63 100.09

>=99 1.61 100.44

>=101 1.54 100.85

>=103 1.45 101.30

>=105 1.34 101.81

>=107 1.20 102.27

>=109 1.04 102.72

>=111 0.87 103.18

>=113 0.70 103.55

>=115 0.54 103.75

>=117 1.22 103.94

Dragon Bank Bet (99% confidence level E =0.02%)

RC % occurence payout %

>=114 0.33 100.02

>=115 0.54 100.32

>=117 0.40 100.80

>=119 0.82 101.06

The second count which I am currently running sims on should be mathematically equivalent but I wanted to verify via simulation. I will post the sim results soon.

Red Dragon Master Count

Card Rank Count Value

Ace +2

2 +3

3 +3

4 +2

5 +1

Red 6 +1

7 -2

8 -2

9 -2

Red 10,J,Q,K -1

This count is easier because the fractions have been removed and you have to count about 20% less cards.

[Tony Provezo]

Thnak you fro your work.
What is "E", expectation?
RC is running count I assume? Why do you not find it necessary to use True Count, ie. RC normalized to the # of unplayed cards?

> The Dragon Bonus Bet is a side bet for the table game
> of baccarat. The paytable is shown below:

> Win by 9 points (non-natural) 30 to 1
> Win by 8 points (non-natural) 10 to 1
> Win by 7 points (non-natural) 6 to 1
> Win by 6 points (non-natural) 4 to 1
> Win by 5 points (non-natural) 2 to 1
> Win by 4 points (non-natural) 1 to 1
> Natural winner 1 to 1
> Natural tie Push
> All other Loss

> (8 Deck shoe)

> House Edge Std Deviation
> Banker 2.65% 2.30
> Player 9.37% 2.47

> Through the use of computer simulation this game can
> be beaten with card counting. Here is the first count
> I developed:

> Dragon Master Count

> Card Rank Count Value

> Ace +2

> 2 +3

> 3 +3

> 4 +2

> 5 +1

> 6 +0.5

> 7 -2

> 8 -2

> 9 -2

> 10,J,Q,K -0.5

> The simulation model used an 8 deck shoe and the
> shuffle point was 401/416. This is common penetration
> for live baccarat (14-16 cards cut position) since the
> primary game cannot be beaten with card counting. The
> Running Count (RC) begins at 0.

> Here are the simulation results:
> Each data point (min 200 million shoes simmed)

> Dragon Player Bet (99% confidence level E =0.02%)

> RC % occurence payout %

> Dragon Bank Bet (99% confidence level E =0.02%)

> RC % occurence payout %

> The second count which I am currently running sims on
> should be mathematically equivalent but I wanted to
> verify via simulation. I will post the sim results
> soon.

> Red Dragon Master Count

> Card Rank Count Value

> Ace +2

> 2 +3

> 3 +3

> 4 +2

> 5 +1

> Red 6 +1

> 7 -2

> 8 -2

> 9 -2

> Red 10,J,Q,K -1

> This count is easier because the fractions have been
> removed and you have to count about 20% less cards.