Terence Tao on Nostr: nprofile1q…hgug4 I would argue that expected utility is a reasonable target to ...
nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpq4uu69aa64sha0pf9ysyh3kq9kld3j66mjcqxdvhryu9muxnmjf9qhhgug4 (nprofile…gug4) I would argue that expected utility is a reasonable target to optimize either when one has enough trials available that the law of large numbers holds, or if financial markets are sophisticated enough that one can use insurance or other derivatives to manage the risks. But I don't think the entirety of risk aversion can be explained purely by the impact of the concavity of the utility function on the expected utility.
One quirk of your model, by the way, is that the pendulum swings in completely the opposite direction once one goes beyond the logarithmic singularity at $0. If x=100.5, then one now faces certain doom unless one takes the bet, so now any non-infinite odds makes the bet a preferable option to do nothing!
One quirk of your model, by the way, is that the pendulum swings in completely the opposite direction once one goes beyond the logarithmic singularity at $0. If x=100.5, then one now faces certain doom unless one takes the bet, so now any non-infinite odds makes the bet a preferable option to do nothing!