Should you take the startup offer? - Better than Random

Better than Random

SubscribeSign in

Should you take the startup offer?<br>The Kelly criterion for your next career bet

Better than Random<br>Aug 03, 2026

Share

Photo by Daria Nepriakhina 🇺🇦 on Unsplash<br>One of the common decisions to make when you think about your career is the risk-reward trade-off between staying within an established, big corporation vs moving to a smaller growing startup or scale-up.<br>Option 1 usually gives you more money now (between base pay and bonus, plus perks like pension contributions and health care plans), and a higher probability of still having a job one year from now, to be on average a small cog in a giant machine.<br>Option 2 gives you less money now but more potential money in the future, and a higher probability of having to look for a new job one year from now, to be on average a big cog in a smaller machine.<br>This decision is very personal and it has many components, but it should be possible to analyze one of them objectively: the financial return.<br>For most professionals under fifty, future earning power is the single largest asset they own. Ibbotson, Milevsky, Chen and Zhu modelled human capital against financial capital across the lifecycle and show human capital dominating until roughly the mid-fifties for typical earners and savers. Take a 40yo earning $72k per year (the median individual income), with 25 working years left. Their future earnings are worth about $1.25M in present value at a 3% real discount rate, assuming no raises or cuts. The median net worth in the 2022 Survey of Consumer Finances was about $135k for households (not individuals) aged 35-44 and about $247k for 45-54. Human capital wins by an order of magnitude<br>One way to model the decision problem of choosing a job is the Kelly criterion. In 1956, John Kelly Jr., a physicist at Bell Labs, published ”A New Interpretation of Information Rate”, stating that a bettor with an edge should wager a fixed fraction of his bankroll to maximize long-term wealth. Funny enough, Kelly framed the whole formula around a gambler using a noisy telephone line to get baseball results tips from a corrupt employee.<br>For a bet that pays b-to-1 and wins with probability p, the optimal fraction (f*) is:<br>\(f^* = p - \frac{1 - p}{b}\)

If f* is positive it means that the bet has an edge. Bet that exact fraction every time, and you maximize the geometric growth rate of your wealth. If f* is negative, do not bet.<br>With the Kelly criterion you never bet the entire bankroll. The size of the bet grows with the edge, defined as the Expected Value (EV) of the bet:<br>\(e = (p \times b) - (1 - p)\)

When f* is positive, underbetting (= less than f*) costs you speed (you grow slower but safer), while overbetting (= more than f*) is too risky. Half Kelly is a common conservative approach, and at double Kelly your long-run growth drops to zero.<br>Let’s map 1950s gambling math to a 2020s career choice. Let’s say that the baseline is a big corporation package, with compensation of $200k, and the alternative is a startup with $100k cash + 0.5% equity with a 4 years runway.<br>Your bankroll is the human capital over 20 years, $200k at 3% discount rate over 20y is $3M. What’s at stake is the difference in cash compensation, $100k x 4 = $400k of foregone big corporation salary over 4 years. So your bet is 13% of your bankroll ($400/$3M).<br>We target 4-to-1 odds (b=4), which means a startup return on the 0.5% equity of 4 x $400k = $1.6M, equivalent to an exit somewhere north of $400M, after dilution and after the investors’ preferences are paid.<br>To find the mathematical justification to go with the startup option, we need to find the probability p where f* = 13.3%.<br>Kelly criterion’s breakeven (f* = 0) for a 4-to-1 bet requires a win probability p of 20%. If your probability is below 20%, f* goes negative: do not bet any amount. To justify betting 13% of your bankroll you need p = 30%<br>Does a seed-stage company clear a $400M exit more than 30% of the time? The base rates of venture capital say no, by an order of magnitude. At an honest baseline probability of p = 3%, this offer is negative Kelly.<br>Two things can rescue a startup bet<br>A bigger p: legitimate private information that increases the edge

A bigger payoff b: more equity or a bigger exit

If that inside view legitimately moves your confidence p to 30%, a $400k stake of a $3M bankroll is justified.<br>In terms of bigger payoff, if the confidence keeps p at 3% and the stake is 13% of the bankroll, the result is negative odds b, which means it is mathematically impossible for a real-world payout to justify this bet. The reason is that there is an ironclad mathematical limit built into the Kelly formula: you can never bet a larger fraction of your bankroll than your actual probability of winning (f* p). No matter if the startup is a 10 bagger, or a 100 bagger, or even an infinite bagger, if you only have a 3% chance of winning, you cannot bet 13% of...