I started thinking seriously about probability after a conversation about a medical test result. A friend had tested positive for a condition and the doctor had described the test as "95% accurate". She was understandably worried. But what that percentage actually meant for her situation turned out to depend on something the doctor hadn't mentioned: how common the condition was in the population in the first place. A 95% accurate test applied to a condition affecting 1 in 1,000 people will produce far more false positives than true positives — which changes what a positive result actually implies. Probability isn't just arithmetic. It's the right arithmetic applied to the right question, and getting that right changes how you interpret almost any statistical result you encounter.
The Basic Formula
For equally likely outcomes: Probability = (number of favourable outcomes) ÷ (total number of possible outcomes).
Rolling a standard six-sided die: what's the probability of rolling a 4? There's 1 favourable outcome (rolling a 4) and 6 total possible outcomes. P(4) = 1/6 ≈ 0.167, or about 16.7%. The probability of rolling an even number: 3 favourable outcomes (2, 4, 6) out of 6 total. P(even) = 3/6 = 1/2 = 0.5, or 50%.
Our probability calculator handles single events, complement probability, combined events (AND and OR), and binomial probability for repeated trials — enter your values and it calculates the result along with the odds ratio and the "1 in X" expression.
Probability as a Percentage vs as a Decimal
Probability can be expressed as a decimal (0.167), a fraction (1/6), or a percentage (16.7%). These are all the same number. Decimal form is most useful for calculations. Percentage form is most intuitive for communication. When a weather forecast says "30% chance of rain", that's a probability of 0.30.
The Complement Rule
The probability that something does not happen equals 1 minus the probability that it does. If there's a 30% chance of rain, there's a 70% chance of no rain. P(not A) = 1 − P(A). This is useful when it's easier to calculate the opposite of what you want. The probability of at least one heads in 10 coin flips: instead of calculating all the ways to get at least one heads (complex), calculate P(zero heads) = (1/2)¹⁰ = 1/1024 ≈ 0.001, then 1 − 0.001 = 0.999 — a 99.9% chance of at least one heads.
Independent Events: The AND Rule
When two events are independent — the outcome of one doesn't affect the other — the probability of both occurring is the product of their individual probabilities.
P(A and B) = P(A) × P(B)
Flipping a coin twice: P(heads and heads) = 0.5 × 0.5 = 0.25. Rolling two dice and getting a 6 on both: P(6) × P(6) = (1/6) × (1/6) = 1/36 ≈ 2.8%.
Note: this rule only holds for independent events. Drawing two cards from a deck without replacement is not independent — the first draw changes the composition of the deck for the second.
Mutually Exclusive Events: The OR Rule
When two events cannot both occur in a single trial, the probability of either occurring is the sum of their individual probabilities.
P(A or B) = P(A) + P(B) (for mutually exclusive events)
Rolling a die: P(2 or 5) = 1/6 + 1/6 = 2/6 = 1/3. When events are not mutually exclusive — they can both happen — the formula requires subtracting the overlap: P(A or B) = P(A) + P(B) − P(A and B).
Binomial Probability: Repeated Trials
When you're repeating an experiment multiple times (flipping a coin 10 times, sampling customers, testing components), binomial probability tells you the likelihood of getting exactly k successes in n trials, given a probability p of success on each trial.
The formula is P(X = k) = C(n,k) × pᵏ × (1−p)ⁿ⁻ᵏ, where C(n,k) is the binomial coefficient "n choose k". This gets complex to calculate by hand for large n, but the pattern is powerful: it tells you the full distribution of likely outcomes before you run any experiment.
Common Mistakes
The most common error is applying the AND rule (multiplication) to events that aren't independent, or applying the OR rule (addition) to events that aren't mutually exclusive. Always check whether your events can co-occur before choosing which rule to apply. The second common error is confusing "probability of at least one" with "probability of exactly one" — these require different calculations.
The Gambler's Fallacy: Why a Coin Has No Memory
One of the most persistent errors in probabilistic reasoning is the belief that independent events "balance out" over time. If a fair coin lands heads seven times in a row, many people feel that tails is now "due". It isn't. The coin has no memory of its previous results. The probability of heads on the eighth flip is still exactly 0.5, regardless of what came before. Each flip is an independent event, and the probability applies fresh to each one.
This is called the gambler's fallacy, and it causes real harm — it underlies much of the reasoning behind chasing losses in gambling, and it appears in more sophisticated forms in financial markets, sports analysis, and management decisions. The streak of seven heads is genuinely unlikely in advance (probability of (1/2)⁷ = 0.78%), but once the seven have occurred, they're history. The eighth flip doesn't know about them. The correct interpretation of a long streak is not that a reversal is overdue — it's that the streak happened, it was improbable, and the next event has the same probability it always did.
Where this gets subtler: if you don't know whether the coin is fair, seven consecutive heads should update your belief about the coin's fairness. Applying Bayes' theorem, you should assign a higher probability to the possibility that the coin is biased after seven heads than you did before. This is different from the gambler's fallacy — you're not predicting that a fair coin will balance; you're questioning whether the coin is fair at all.