expected value – A classic probability puzzle shows why some “unfair” games can still be worth playing—until finite money changes everything.

A coin game can sound like a rigged deal—until math forces a different answer.. The surprise hinges on how probability. stakes. and the size of a player’s bankroll interact. and it starts with a simple question: would you accept a wager you “can’t lose” on. even if you’re asked to pay a high enough price?

Imagine rolling a fair die.. If it lands on 1 or 2, you receive $10; if it lands on 3, you receive $20; otherwise you get nothing.. Because the payoffs are positive and there’s no scenario where you’re charged money by the rules of the game itself. it may feel risk-free.. But the price changes everything: the proposal is that you must stake $10 for each roll.. Under probability theory, there are six equally likely outcomes.. Winning $10 happens in 2 out of 6 cases. which is a one-third chance. while winning $20 happens in 1 out of 6 cases. or one-sixth of the time.. When the winnings are averaged over all outcomes, the expected value per game comes to 40/6, which is 20/3 dollars—about $6.66.

That average is precisely the problem. If you pay $10 per roll, then even though you often win, your average gain becomes $6.66 minus $10, yielding about -$3.33. In plain terms, the deal isn’t actually favorable once the stake is included. The math points to a clear decision: you should decline.

The discussion then shifts to a different wager, one that mathematicians wrestled with long before modern casinos existed.. In 1713. Nicolaus I Bernoulli and Pierre de Montmort exchanged ideas about a coin-toss game played until the first time it lands on heads.. The twist is that the payout grows dramatically with each step the game continues: if tails comes up for the first time. the player wins $1; if tails must appear twice in a row before the first head. the payout becomes $2; with three tails in a row before the first head. it becomes $4; and so on.

This structure makes the game feel strange in a way that’s hard to judge by instinct.. Every extra round before a head doubles the potential reward.. So if you were offered this game and asked to pay an extremely high stake—say $2,000—would you walk away?. Most people would.. Yet the expected value calculation tells a different story.. Each pattern has a halving probability: the chance the coin ends the game immediately on the first toss

is 1/2. the chance it lasts two tails before a head is 1/4. three tails is 1/8. and so forth.. At the same time, the payoff doubles each time the chain extends.. When you match the decreasing probabilities with increasing payouts. the expected value becomes an infinite sum that. in this setup. grows in such a way that there is no finite stake that makes the “average” outcome negative.. Mathematically speaking, no matter how large

the stake is, the recommendation is to always play.

That counterintuitive result became known as the Saint Petersburg paradox. named after a fictional casino setting used to frame the thought experiment.. The “paradox” isn’t that the mathematics is inconsistent; it’s that human decisions often don’t align with the expected-value logic.. The uncomfortable truth is that the expectation comes from adding infinitely many possibilities. and the payoff tail is so heavy that the average is dominated by outcomes that are effectively too extreme to treat realistically.

The oddity grows clearer when you look at what the late rounds would mean.. With a run of tails that lasts long enough. payouts explode: six successful tosses correspond to $32; six more to $2. 048; and if another lucky streak carries the game through tails six more times again. the winnings can reach $131. 072.. The structure is designed so that in theory. the rare events far out in the future would lift the expected value.. But the whole mechanism depends on something most real-world challengers don’t have: infinite resources.. If you have finite capital. there is an upper limit to how long you can keep paying out. and once that limit is reached the “infinite” reasoning no longer describes the actual game.

That’s where the thought experiment starts to look less like a mathematical curiosity and more like a practical warning about real betting.. Suppose someone holds $1,050 in their account and offers to bet anything against you with the coin-toss rule.. Even if they would claim the game is still “the same. ” they cannot honestly allow the payout to grow without bound; after enough rounds. their funds would be exhausted.. If you want to test your own decision at a realistic price. the expected value must be recomputed under a finite-capital assumption.

In that modified situation, the maximum number of rounds the challenger can cover becomes central.. For instance. if the payout could force the challenger beyond their available money partway through the process. they might fail to finance the next toss.. The report works through this by treating the expected value as a finite sum: 1 × 1/2 plus 2 × 1/4 plus 4 × 1/8 and so on. up to 1. 024 × 1/2. 048.. Under this finite limit, the expected value becomes 1/2 × 11, which is $5.5.. With a stake of $6, the recommendation flips again: you should refuse.. But if the stake is lowered to $5. where the price no longer exceeds the corrected expected value. the chances of making a profit become favorable.

Now scale the same idea to a massively wealthy opponent.. If a billionaire challenges you. the question becomes how many coin-toss rounds they can sustain before going broke. given the rule that the amount owed grows rapidly with each additional tails streak.. The report provides a sense of scale: after just 38 rounds. the amount owed would exceed $137 billion. and that magnitude is described as well beyond what typical mega-wealth figures would tolerate in a game like this.. With that capped duration. the expected value at the maximum stake becomes about $19. implying that the amount you’re willing to pay depends on what the challenger can realistically bankroll.

Taken together. the die game and the coin-toss game point to the same core lesson from probability: expected value only tells you what to do when the rules match what will actually happen.. When the stake is too high, even “winning often” doesn’t save the deal, as shown by the $10 die roll.. When payout growth is tied to very long sequences. the infinite expected value can suggest you should always play—but once you impose finite money. the calculation changes. and so does the decision.. It’s a reminder that math can be decisive. yet its verdict depends on whether the assumptions behind the model survive contact with reality.

expected value probability dice game coin toss game Saint Petersburg paradox gambling strategy