Normal statistics don't really apply when losing puts you out of the game permanently (like dying) or any other one-time risk. By definition statistics are about expected or average results. A company, VC, or casino can take a lot of risks and get close to the average results - an individual can't.
Exactly! The concept of "expected value" is pretty meaningless when it comes to one-time events. For example, when insuring a 1 billion $ item with a loss probability of 1 in 1,000,000 at a premium of 10,000 $, the expected value of the profit is 9,000 $ – which is useless if the insurance case occurs and the insurer is bankrupt afterwards.
Similarly in life: For some events in life, there's just no time to recover afterwards.
