From link “Five startups fail and do $0, three exit at $25 million, one exits at $200 million and our superstar does $1 billion”. The three exits at $25 generate zero for employee shares (liquidation preferences). Discount a lot of the $200MM due to participating shit etcetera.
Make some guesses for dilution (likely severe as you don’t get to participate in further rounds), and make a decision about whether rolling a 1D10 die and getting a 1 is worthwhile.
In my opinion, you need to ensure you have ways to get plenty of value out of the 80% expectation that your equity will be worth nothing. E.g. good $, expected value of what you learn, joy from job, etcetera.
The equations shift somewhat for non-VC equity, where other severe risks and other rewards become relevant.
For ycombinator companies, the numbers are probably skewed different again. Interestingly enough HN should be able to estimate returns for early employees for YC companies, since we know numbers of intake companies and maybe employee counts, and could do an estimate of returns per employee per intake. YC themselves could do population samples per intake and give realistic historical figures.
PS: do not fall for the fallacy that you can pick a future unicorn to be an employee of. VC experts and others with decades of experience fail to pick unicorns, and to believe you can do better than them would be a conceit.