Can We Beat RNG?

Baccarat, of course, is a card game offered by casinos both physical and online. Six or eight decks of cards are used to constitute the shoe. A set of rules determines how many cards are drawn and when, and the side, Player or Banker, closest to a sum of 9 wins the hand. In essence, the game is basically an even-money, 50/50 coin-toss, where the player bets either heads (Player) or tails (Banker) will win.

In a physical casino, the cards are either hand or machine shuffled, and there is some debate regarding whether the shuffle can be controlled in such a way to allow casinos to present on command a biased shoe, biased toward being choppy (constantly alternating wins between Player and Banker), streaky (long runs of Player or Banker wins), or neutral (even number of alternating and running wins for Player or Banker).

Some online casinos offer live-dealer versions of baccarat, where a live dealer draws from a physical shoe in real time, thus reproducing exactly the same game as one would play in a physical casino (for example, BetPhoenix, Black Orchid, and many others).

Other online casinos offer a computer-based baccarat game, where the outcome of the hand is determined by a random number generator (RNG). Some players believe that RNG-based shoes cannot be beat, since at best, the resulting shoes exhibit no exploitable bias, or at worst, the casino surreptitiously uses a cheating algorithm that effectively bets against the player.

For example, the online casino Bodog uses a 6-card simulated deck and continuous shuffle.

Here is how they word it:

“Bodog Baccarat is played with six decks; cards are reshuffled after each hand.”

(from their webpage here)

In more detail:

Thank you for contacting Bodog Casino Customer Service.

The Bodog Casino uses a cycling Random Number Generator (RNG) in the Flash and Download versions of our software. This generator also initiates results in both the Fun and Real Money areas of our website. Our generator is different from a basic generator in that it cannot become predictable. The cycling occurs when, for a non-predetermined amount of time, the software re-seeds the generator. This eliminates any possibility that the outcome can be determined before the numbers fall and also completely removes any sequence patterns.

Bodog’s cycling RNG is constantly generating numbers, whether the machine is being played or sitting idle; these numbers are being processed at a rate of no less than 100 per second. When an action is initiated by a player, the computer’s processor retrieves the number that is generated at that instant. If a player is idle for 2 seconds, no less than 200 outcomes will already have been generated since their last action.

Our Random Number Generator has undergone testing and has received a TST Certification. Technical Systems Testing (TST) is an internationally recognized Accredited Testing Facility (ATF), which offers a full range of testing and consultation services for Terrestrial (traditional / land-based) and Interactive-based gaming, wagering, lottery, e-Commerce and Information Technology industries, to ensure that gaming operates in a manner that is fair, secure and auditible.

To view this documentation and any other information on our Casino, please visit:

From this information, it seems like Bodog’s RNG controls when or how the simulated shuffle occurs. There are 100 outcomes decided per second, and depending on when the player hits “Deal,” one of those outcomes is presented. This makes the result “random,” which for all practical purposes, just means that the resulting pattern cannot be predicted with any accuracy by a human being.

Assuming Bodog’s RNG presents an entirely fair game and does not cheat, there is an important difference between Bodog’s RNG and a physical shoe that is unrelated to the issue of bias: with Bodog’s RNG, no one ever sees the same shoe.

Why? Because a difference of 1/100 of a second per decision may give a different result. Thus, the chances that two players will ever see the same shoe is practically none. And one cannot go back to re-play a shoe and say, “If I had played it this other way, I would’ve gotten this score instead.” If he had played it another way in real-time, he would’ve probably gotten a completely different shoe, since to exactly reproduce the original shoe, he would’ve needed to play every decision at exactly the same time down to 1/100 of a second.

In contrast, in a physical shoe, the result is the same for all, regardless who is playing, dealing, or the amount of time it took to draw the hands. Then, one could certainly say, “If I had played this shoe differently, I would’ve gotten this score instead,” since he is basing his play on a history that is universally experienced and independently verifiable.

Thus, this reproducibility factor is a very real difference between Bodog’s RNG and a physical shoe.

I think to better simulate a physical shoe, the RNG seed should be used only once at the beginning to predetermine the results of the entire shoe once and for all, and then draw out the hands accordingly to the preset “random” pattern. However, the reason why they can’t do this in practice is because it would open up many avenues for players to cheat. For example, one of the programmers can easily create a back-door access to Bodog’s server, fetch the results of the predetermined shoe, and bet accordingly to always win, or use a pattern of wins and losses to avoid getting caught but still always come out ahead.

Thus, to avoid creating these cheating opportunities, RNG providers like Bodog must make the outcome of every hand unpredictable. However, when they do so, they destroy the universality and reproducibility of the shoe. Unfortunately, they also at the same time destroy a way to objectively demonstrate they are not themselves cheating, since there is no practical way of re-creating the same shoe twice. One has to trust that the third-party verification regulators make sure the game is entirely fair and no “cheating algorithm” is in place.

Of course, in real life, 99% of baccarat players consistently lose, so there’s plenty of blame to go around, whether it be directed at the casino, lady luck, God, fate, or the computer. Curiously, the same happens in trading, where 80% of traders consistently lose, and the same kind of blaming game goes on against brokers, clearing houses, luck, fate, God, or the computer.

It’s just human nature to blame anything and everyone except one’s self.

If RNG-based shoes can be beat using the same methods that beat shoes at physical casinos, then one of the following must be true:

– Those methods are flexible enough to beat shoes without exploitable biases.

– Random shoes can exhibit enough exploitable biases to be consistently beaten.

If the first is true, then those methods should be able to win any even-bet coin-toss.

If the second is true, then there is no need to appeal to the theory that casinos control the shuffle, since even a purely random shuffle can exhibit the same characteristics which makes one believe a non-random, exploitable bias exists.

Either way, if one is able to beat an RNG-based shoe, he can easily beat a shoe exhibiting biases.


5 Responses to “Can We Beat RNG?”

  1. […] generator, and I discuss how it is inherently different from a live-dealer game in this post: Can We Beat RNG?) Unfortunately, I encountered nemesis after nemesis during the game, and the drawdown became […]

  2. By definition no exploitable bias exist in a random outcome… that is what random means. A strategy that beat live baccarat shoe bias will not beat RNG since no practical bias exist… only variance giving the illusion of bias (i.e. skewed distribution in probability distribution).

    So, stick with developing strategy to beat live shoe… (avoid RNG and avoid computer simulated live shoe because later can miss inherent bias).

    • BaccPlayer, you’ve left numerous comments all saying the same thing.

      I’ve addressed this issue fully in:
      1. Shuffle Control: Why It’s Bad For the House and
      2. Beating Random.

      Furthermore, my detailed analysis of the Zumma live shoes show no statistical difference from randomly generated virtual shoes (ref. My Baccarat Shoe Factory).

      The fact is, all the objective evidence indicates live baccarat shoes conform to a random distribution. Temporary apparent biases exist in any random distribution.

      What you keep repeating is Ellis’ theory about artificially manipulated shoe bias, and all objective evidence does not support it. You are entitled to your belief, but it will not help you win in the long run.

      Ellis could not win in the long run, nor could any of his students, and my simulations objectively demonstrate why.

  3. Howdy! This is my first visit to your blog! We are a collection of volunteers and starting
    a new project in a community in the same niche.
    Your blog provided us useful information to work on. You have done a wonderful job!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: