Giveaway Entry Odds Calculator – The Maths Behind the Draw
A giveaway says it will draw three winners. You have ten entries and someone in the comments reckons there are about five thousand in total. What is the chance one of yours comes up? The arithmetic most people reach for is 10 × 3 ÷ 5000, which gives 0.06%, and it is not the answer to that question. This giveaway entry odds calculator computes the correct figure and shows how far the familiar shortcuts sit from it.
The mechanic this models
Winning entries are drawn without replacement: each drawn entry leaves the pool, and one entry can win at most one prize. That is the standard reading of “we will draw three winners from all entries”, and it makes the number of your own entries that get drawn a hypergeometric variable rather than a binomial one. Writing n for the total entries, k for the winners drawn and m for the entries you hold:
P(no win) = C(n − m, k) / C(n, k)
P(at least one) = 1 − C(n − m, k) / C(n, k)
P(exactly j) = C(m, j) · C(n − m, k − j) / C(n, k)
n = 5000, k = 3, m = 10
P(no win) = (4990/5000)(4989/4999)(4988/4998) = 0.994010796
P(at least one) = 0.005989203 → 0.5989% → about 1 in 167The page never builds C(n, k) itself. Entry totals run into the hundreds of thousands, where a factorial overflows long before the ratio does, so the quotient is evaluated as a product of terms each close to 1 — and always by the shorter of its two equivalent forms, so three winners cost three multiplications whatever the pool size. The per-prize distribution goes through log-gamma for the same reason.
Two shortcuts, and what each one really is
The first is m · k / n. It is not a probability; it is the expected number of prizes you win — an average count. It tracks the true odds closely while those odds are tiny, which is why it circulates, but it is unbounded. Put n = 100, k = 50 and m = 10 into it and you get 5, a perfectly sensible average of five prizes and a nonsensical “500% chance”. The correct probability there is 99.94%.
The second is 1 − (1 − m/n)^k. This treats the draws as k independent trials, which describes a draw with replacement: the same entry could be pulled twice and a drawn entry stays in the pool. For the worked example it returns 0.5988% against the true 0.5989% — agreement to three significant figures, divergence at the fourth, and a gap that widens as k grows towards n.
Reading the curve
The chart plots the chance of at least one win against the entries you hold, with the total and the winner count fixed. It is close to a straight line while your holding is a small slice of the pool, then bends over as that slice grows, because each extra entry competes with the ones you already hold for the same fixed number of prizes. The y-axis scales to the data rather than to 0–100%, so a curve living entirely under 1% is still legible.
Where this does not apply
The calculation assumes a single flat pool of identical entries, a genuinely random draw, and one prize per entry. Weighted or bonus entries, per-person caps applied after the draw, tiered prize pools and multi-stage selections all break at least one of those assumptions, and the numbers here do not describe them. A giveaway’s own official rules are the only account of how its draw actually runs, and the law where you live governs who may take part.