...in purely financial terms, there is probably at most one company in each YC batch that will have a significant effect on our returns, and the rest are just a

...in purely financial terms, there is probably at most one company in each YC batch that will have a significant effect on our returns, and the rest are just a cost of doing business. Clearly you don't look at YC in purely financial terms, but it would be an interesting thought experiment to consider whether there would be more efficient ways to generate returns if you did. Among other things, there may be more efficient ways of testing variations of ideas around a theme, or testing/ranking skills, or assessing likelihood of success for various experimental team combinations. To use a casino metaphor, it seems like you're taking a 50-100 completely binary and largely independent bets per class and hoping to beat the house on average. As opposed to, say, trying to create the MIT blackjack team to beat the house systemically.