Two events are independent when learning that one event occurred does not change the probability of the other. In casino language, a previous independent result does not make a win, loss, color, number, or symbol more likely on the next trial.
The word is often used too casually. Separate events are not automatically independent, and outcomes that cannot happen together are usually not independent either.
The mathematical test
For events (A) and (B), independence means:
[ P(A\cap B)=P(A)\times P(B) ]
where:
- (P(A\cap B)) is the probability that both events occur;
- (P(A)) is the probability of event (A);
- (P(B)) is the probability of event (B).
When (P(B)>0), the same idea can be written with conditional probability:
[ P(A\mid B)=P(A) ]
This says that after being told (B) happened, the probability of (A) remains unchanged.
Roulette: independent spins, combined sequences
On a fair European roulette wheel, red appears on 18 of 37 pockets. Assuming the wheel, ball, and operating conditions have not changed, the probability of red on the next spin is:
[ P(\text{red})=\frac{18}{37}\approx48.65% ]
Five black results do not alter that next-spin probability. The display records history; it does not create a balancing force.
Independence does not mean sequences are impossible. The probability of three consecutive reds is found by multiplying the same probability three times:
[ P(\text{three reds})=\left(\frac{18}{37}\right)^3\approx11.51% ]
Each individual spin still has a 48.65% red probability. The three-spin sequence is less likely because all three independent events must occur.
This distinction matters: the probability of a sequence is not the probability of the next event.
Dice: events on different rolls and events on the same roll
Successive throws of properly functioning dice are normally modeled as independent. If a six is rolled now, the probability of a six on the next throw remains (1/6).
But two descriptions of the same throw may be dependent. With two dice, let:
- (A): the total is 7;
- (B): the first die is 1.
Before seeing either die:
[ P(A)=\frac{6}{36}=\frac{1}{6} ]
If the first die is known to be 1, only the second die determines whether the total becomes 7. It must be 6, so:
[ P(A\mid B)=\frac{1}{6} ]
In this particular case the probabilities happen to match, so those two events are independent.
Now replace (A) with “the total is 2.” Before seeing a die, the probability is (1/36). Given that the first die is 1, the probability becomes (1/6). Those events are dependent because the new information changes the chance.
Cards dealt without replacement are dependent
Card games provide the clearest casino counterexample.
From a standard 52-card deck, the probability that the first card is an ace is:
[ \frac{4}{52}=\frac{1}{13} ]
If the first card is an ace and it is not replaced, only three aces remain among 51 cards:
[ P(\text{second ace}\mid\text{first ace})=\frac{3}{51}=\frac{1}{17} ]
Because (1/17\neq1/13), the events are dependent.
That changing composition is why removed cards matter in blackjack and why deck penetration can matter to card counters. The next hand is still uncertain, but its probability distribution is not necessarily identical to the distribution at the start of a fresh shoe.
If the first card were returned and the deck fully randomized before the second draw, the ace probability would return to (4/52), making the two draws independent under the model.
Independent is not the same as mutually exclusive
Two events are mutually exclusive when they cannot happen together. On one roulette spin, “red” and “black” are mutually exclusive.
They are not independent. If red occurs, the probability of black on that same spin becomes zero. Knowing one event happened completely changes the probability of the other.
For events with positive probabilities:
- independent events can occur together;
- mutually exclusive events cannot occur together;
- therefore mutually exclusive events are normally dependent.
This is one of the most common terminology errors in elementary gambling math. Penn State’s probability course uses the same product-rule definition in its lesson on independent events.
When the independence assumption needs caution
Casino calculations rely on a defined model. Independence may fail, or the model may need revision, when:
- cards are removed without replacement;
- a drawing removes winning entries;
- a progressive prize changes after a hit;
- game rules or paytables change between trials;
- equipment is biased, damaged, or operated differently;
- wagers refer to overlapping outcomes from the same event;
- a promotion links one result to the next;
- data include several games with different probabilities.
A progressive jackpot illustrates a subtle point. The random outcome of one spin may still be generated independently of the previous spin, while the value of the next wager changes because the jackpot amount changed. Outcome independence and constant expected value are not the same claim.
Why streaks mislead players
A streak is a valid description of completed results. It is not automatically a prediction.
After eight Banker results in baccarat, the ninth result may continue the run or break it. The sequence can look meaningful even when the next-trial probabilities are governed only by the remaining shoe composition and drawing rules. In roulette, treating a long color run as proof that the opposite color is “due” is the gambler’s fallacy.
The practical question is not, “Has this outcome happened too often lately?” It is, “Did the previous result change the probability mechanism for the next event?”
Independence does not prevent streaks. It prevents the streak itself from changing the probability of the next trial.