A random variable assigns a numerical value to each outcome of a chance process. Its expected value is the long-run average value predicted by the probability model. It is a weighted average, not necessarily the most likely outcome or an outcome that can occur once.
Build a valid probability distribution
Suppose a game pays 0 dollars with probability 0.5, 2 dollars with probability 0.3, and 5 dollars with probability 0.2. The probabilities are nonnegative and add to 1, so they describe a complete distribution.
If X is the payout, its expected value is
To see why, imagine 1,000 plays with frequencies close to the probabilities. About 500 pay zero, 300 pay 2 dollars, and 200 pay 5 dollars. The total payout would be about 1,600 dollars, averaging 1.60 per play. No single play needs to pay exactly 1.60.
Payouts of 0, 2, and 5 dollars have probabilities 0.5, 0.3, and 0.2 respectively.
The payouts are weighted by their probabilities, not averaged equally.
Describe this figure
Payouts of 0, 2, and 5 dollars have probabilities 0.5, 0.3, and 0.2 respectively.
0: 0.5; 2: 0.3; 5: 0.2.
Example: payout is not profit
If the game costs 2 dollars to enter, the player's net gain is the payout minus 2. Therefore the expected net gain is dollars per play.
Alternatively, use net gains of 0, and 3 directly: Both methods agree. Subtract the entry fee once; do not confuse the prize received with the profit earned.
A fair entry fee in the expected-value sense would be 1.60 dollars, giving expected net gain zero. “Fair” here describes the mathematical average, not a guarantee that any particular player breaks even.
Example: unknown probability or payout
A promotion awards 10 points with probability and 2 points otherwise. If its expected award is 5 points, then
Simplifying gives so The term matters: the two outcomes exhaust the possibilities. Check that the answer lies between 0 and 1 and that the expected value lies between the smallest and largest possible awards.
Spread matters alongside the average
Two games can have the same expected payout but very different variability. One may always pay 5 dollars; another may pay 0 or 10 with equal probability. Both average 5, but only the first is certain.
For a discrete variable, variance measures average squared distance from its mean
The standard deviation is the square root of variance and has the same units as X. For equally likely payouts 0 and 10, the mean is 5, variance is 25, and standard deviation is 5 dollars.
For a transformed value the mean becomes Adding a constant shifts every outcome equally; multiplying scales every outcome.
Add expected values without assuming independence
If X and Y are two numerical outcomes, whenever the expectations exist. Independence is not required for this mean rule. Thus two games with expected payouts of 1.60 and 2.40 dollars have expected combined payout 4 dollars, even if their outcomes are linked.
Variability requires more care. Variances add for independent variables, but generally not for dependent ones. Two quantities can move together or offset each other while keeping the same individual means. Separate what you know about the average from what you know about joint variation.
Try it yourself
A raffle pays 20 dollars with probability 0.1 and zero otherwise. A ticket costs 3 dollars. Find the expected payout and expected net gain.
Answer: Expected payout is dollars. Expected net gain is dollar. A winning ticket is still possible, but the model predicts an average loss over many comparable plays.
What to remember
Check the distribution, multiply each value by its probability, and add. State whether values represent payouts, costs, or net gains, and do not confuse an average with a guaranteed result.