Startups as Bets: Expected Value, Bankrolls and Portfolio Thinking

Founders rarely describe their companies as bets. Investors do it all the time. A venture fund expects most of its investments to disappoint and a few to pay for everything. A professional poker player expects to lose plenty of hands and win over thousands. Treating your idea as a bet, rather than a destiny, is one of the most useful mental shifts in validation.

Expected value in one paragraph

The expected value (EV) of a bet is what you would win or lose on average if you could make it many times: each possible outcome multiplied by its probability, added up. A roulette bet of $10 on red at a single-zero wheel wins $10 with probability 18/37 and loses $10 with probability 19/37, so its EV is about minus 27 cents. That is the house edge of 2.70%, and it is why casinos are happy to see you play all night.

EV = Σ (probability of outcome × payoff of outcome)

An idea's expected value

Imagine two ideas. Idea A needs a year and $150,000 of your savings and forgone salary. You estimate a 10% chance it becomes a business worth $3 million to you, a 30% chance of a modest outcome worth $200,000, and a 60% chance of nothing. Its EV is 0.1 × 3,000,000 + 0.3 × 200,000 = $360,000, minus the $150,000 stake: positive. Idea B costs the same but you can only see a 2% chance of a big outcome; the maths turns negative.

The numbers are guesses, of course. The point is not precision; it is to make your assumptions explicit, and to notice which ones dominate the result. Usually it is the probability of success, which is exactly what validation is designed to estimate more honestly.

Validation as buying information

Decision theorists talk about the value of information: how much you should pay to learn something before betting. If a $2,000 smoke test can tell you whether the probability of success is 10% or 2%, and that difference flips your decision, the test is worth far more than it costs. Poker players do this instinctively when they pay a small amount to see the next card. Each validation step in our validation loop is a small bet that buys information about the big one.

Bankrolls and the Kelly criterion

In 1956 John L. Kelly Jr., a researcher at Bell Labs, published a formula for how much of a bankroll to stake on a favourable bet in order to grow wealth fastest over time. For a simple bet paying b to 1 with win probability p (and loss probability q = 1 − p), the Kelly fraction is:

f* = (b × p − q) ÷ b

On an even-money bet (b = 1) with a 55% chance of winning, Kelly says stake 10% of your bankroll. The mathematician Edward Thorp later applied Kelly-style thinking to blackjack card counting and then to investing. Two lessons carry over to founders. First, even with an edge, betting everything is a mistake, because one loss ends the game. Second, many professionals bet a fraction of Kelly, because their estimate of the edge is uncertain. If you are not sure of your odds, bet smaller. For a founder, that means keeping enough runway, savings or income to survive a failed attempt and try again.

The power law: why portfolios win

Venture capital returns are famously skewed. Analyses of fund data have repeatedly found that a small share of deals produces most of the returns, while a large share loses money. Investors therefore think in portfolios: they need enough shots for the rare outlier to appear. A solo founder cannot diversify in the same way, but can think in sequences: several small, cheap tests of different ideas, and a large commitment only to the one that earns it. The mobile-game industry runs this model openly, as described in prototype testing in mobile games.

Thinking in Bets

The former professional poker player Annie Duke, a World Series of Poker bracelet winner, set out this way of thinking for a general audience in Thinking in Bets: Making Smarter Decisions When You Don't Have All the Facts (2018). Her central warning is against "resulting": judging the quality of a decision by its outcome. A good decision can lose and a bad one can win, especially over small samples. Duke argues for separating the two, reviewing decisions on what you knew at the time, and stating beliefs with probabilities ("I'm about 70% sure") to invite challenge.

For founders, this is liberating. An idea that failed after careful validation was not necessarily a bad bet. An idea that succeeded after no validation was not necessarily a good one. Review your process, and keep the bets small until the evidence says otherwise. We take the poker analogy further in poker skills for founders.