Toss a coin until the first head. If it arrives on the k-th toss the bank pays 2^(k−1) ducats — 1, 2, 4, 8 — and the chance of getting that far is 2^(−k). Multiply the two and every term is exactly ½: the tenth round pays 2⁹ = 512 ducats with probability 1/1024, and 512 × 1/1024 = ½. The expected value is therefore ½ + ½ + ½ + … for as long as the game may run, which is why no finite price for a ticket is defensible on the arithmetic alone.
Putting a ceiling on the game is the only way to get a number at all. Cap it at a hundred rounds and the ticket is worth 100 × ½ = 50 ducats, which is the first link; cap it at a thousand and the same rule says 500. The cap does all the work, and that is the paradox. Daniel Bernoulli's 1738 answer was that utility is not money, so the tenth round's 512 ducats — or the twentieth round's 2¹⁹ = 5.24288⋅10+5, on the third link — are worth far less than that many first-round ducats. The modern answer is blunter: no bank on earth can post the prize, so the game with no ceiling does not exist. What it measures is how badly average payoff behaves when the tail is unbounded.
Answer
50
How to read it
A hundred rounds at ½ ducat each. The second link is the tenth round's prize and the third the twentieth round's.
Source
Nicolas Bernoulli set the problem in a 1713 letter to Pierre Rémond de Montmort; it appeared in the St Petersburg Academy volumes of 1738 with Daniel Bernoulli's resolution by logarithmic utility. The bank pays 2^(k−1) ducats on the first head at toss k
Work it in the calculator
These links open the right calculator with this problem's numbers already typed in. The line above is what the screen should show — check it yourself.