Sure, but the odds aren't 10% or 40%, they're more on the order of 1% to 3%. If you can pick winners at a 40% rate, you're doing orders of magnitude better than

Sure, but the odds aren't 10% or 40%, they're more on the order of 1% to 3%. If you can pick winners at a 40% rate, you're doing orders of magnitude better than even the top VCs in the industry, and your talent is worth billions. Realistically nobody plays at those odds, the actual odds being played are much, much lower. Which also means that, even through a long career, you're not taking on enough jobs to get a reasonable shot at a win - 10 jobs with a 2% chance of winning each aren't good odds - you have a 18% chance of scoring a win. All the while you're giving up a lot of money to do so - BigCo's will pay you easily $100-200k more per year. So even if you switch jobs yearly, 10 jobs = 10 years = $1-2M on the table to play for 18% odds. Which is also why investors who are able to choose well win - YC for example invests in over 200 companies a year. Unless you're switching jobs every week you're not going to be able to raise N high enough to make a difference - at YC's scale a 2% chance of winning for each company means there's a 98% chance at least one company wins. The entire idea here is that nobody has predictive power high enough to the degree of 10% vs. 40%. Realistically when choosing between two competing startups you're choosing success odds in the low single digits. Not to mention the rash of articles in the past week about the low payouts you're likely to get even if the company succeeds. Engineers don't walk away from modern startups with 7-figure payouts, low 6-figures is more likely. So realistically even if you are able to pick companies at a 40% accuracy (again, astronomically higher than reality) your expected payout is lower than the $1-2M you're leaving on the table in order to play.