Exponentials drive asymmetry
Per Nassim Taleb, we should make asymmetric bets with long right tails. Even though he is very focused on the idea of (non)ergodicity, these discussions usually implicitly assume a stationary distribution, which is to say they lack a time component. It is assumed that the distribution of outcomes doesn’t change over time – it’s the same bet every single time. But in the real world the distribution of outcomes may shift over time.
It strikes me that one way of achieving asymmetry that may not be present in the current moment is to find intertemporal asymmetries. If you can find a bet where the expected value and the right tail of the distribution are growing fast, continually betting on that thing may yield a form of asymmetry through time. This could justify large expenditures in the present moment that might seem foolish to others but that could have high payoffs in a future, much more positively-skewed distribution. The best sort of thing to bet on along these lines is something growing exponentially in compounding fashion.