BRIEFLY NOTED: For 2022-03-24 Th: The clever plan was to exhaust the dog early in the morning so that I could work, and get and stay in the Zone until lunchtime without interruption, because the dog would be taking a nap. This plan has a huge flaw…
I always thought that insurance was about limited liquidity and time horizons, rather than risk preferences. People with an income of $200K are fools to insure their cars, since their savings should be able to cover a collision w/o transferring wealth to the insurer. People with an income of $60K are wise to insure their cars. Nothing to do with risk preferences, but just another cost of not having enough money.
I love and respect your friend Robert Waldmann - follow him on Twitter - but framing Kelly Criterion as being about *utility* rather than *maximizing payoff* is unhelpful. Kelly Criterion maximizes returns to a *risk-neutral* agent if the amount bet is a monotone function of current wealth. Admittedly, the Kelly criterion obscures this because it was devised in the context of gambling, where the stakes are plausibly independent of winnings.
But that is not true of your retirement account - by the premise of retirement accounts. You invest for retirement because you assume that your wealth will grow and you plan to reinvest that growth. In that case the 'expected return' is the *instantaneous* expectation. To get the expectation over a finite interval you have apply Ito's Lemma. Seriously, the difference between the 'expected return' and the Kelly criterion is just the difference between the continuous SDE for, say, GBM and the difference equation!
I always thought that insurance was about limited liquidity and time horizons, rather than risk preferences. People with an income of $200K are fools to insure their cars, since their savings should be able to cover a collision w/o transferring wealth to the insurer. People with an income of $60K are wise to insure their cars. Nothing to do with risk preferences, but just another cost of not having enough money.
I love and respect your friend Robert Waldmann - follow him on Twitter - but framing Kelly Criterion as being about *utility* rather than *maximizing payoff* is unhelpful. Kelly Criterion maximizes returns to a *risk-neutral* agent if the amount bet is a monotone function of current wealth. Admittedly, the Kelly criterion obscures this because it was devised in the context of gambling, where the stakes are plausibly independent of winnings.
But that is not true of your retirement account - by the premise of retirement accounts. You invest for retirement because you assume that your wealth will grow and you plan to reinvest that growth. In that case the 'expected return' is the *instantaneous* expectation. To get the expectation over a finite interval you have apply Ito's Lemma. Seriously, the difference between the 'expected return' and the Kelly criterion is just the difference between the continuous SDE for, say, GBM and the difference equation!
The Paul Krugman link is broken