Substrat: payoffs - not predictions - are relevant
"Great Pirates think and live comprehensively, in wholes, and they master the entire gameboard—indeed, the world."
focus on activities with remove payoff and no feedback that are ignored by most
geometric avg. > arithmetic avg.: goes to zero with single zero
"the key is efficiency"
safe haven: place of refuge for when things go bad (safety from risk)
goal: increase geometric outcome (in every percentile path, esp. median)
lowering risk: raise 5th percentile ending wealth
essential safe haven features may be emergent and transient
safe haven payoffs
store-of-value payoff: fixed in time and space (e.g. annuity); low/no correlation/sensitivity to systematic exposure
alpha payoff: negative correlation/sensitivity to systematic exposure
insurance payoff: extreme case of alpha payoff (high crash payoff per unit of cost)
minimal (e.g. 2%) explosive insurance payoff (e.g. +1000% at crash, -100% rest of the time) weight vs. large (e.g. 40%) store-of-value payoff (e.g. fixed 5%) may create similar risk mitigation, but very different final wealth
The more explosive the crash payoff, the smaller the safe haven allocation needed.
ergodicity: arithmetic average = geometric average across sample space and time
diversification is "diworsification": most likely not cost-effective risk mitigation (you get less risk, no matter the cost)