The app for independent voices
Is mean-variance really sufficient? Here are some undeniable facts:
It assumes that return distributions are fully characterized by means and covariances, being symmetric bell-shaped curves.
It assumes that the dependencies are the same in risk-off scenarios as in calm markets.
It minimizes the upside as much as the downside.
Market data clearly shows that 1. and 2. do not hold, and investors are usually not worried about becoming too rich.
Yet some claim that it’s good enough, while I argue that it is a recipe for disaster to proactively ignore the fat left tails and risk-off dependencies.
Over the past couple of years, I have realized that those who continue to defend mean-variance usually haven’t tried new and better methods.
The mean-variance justification is mostly based on illogical dogma spread by people who want to continue selling their old work. Don’t fall for it.
For a presentation of better modern methods, see the Portfolio Construction and Risk Management book:
