david merkelDavid Merkel submits:

Longtime readers know that I am not a fan of modern portfolio theory. It is a failure for many reasons:

  • It assumes there is one type of risk, the occurence of which is random.
  • It assumes that this risk can be approximated by volatility (variance of returns), rather than probability of loss, and the likely severity thereof.
  • Mean return estimates, volatility estimates, and correlation coefficient estimates aren’t stable.
  • In crises, correlations head to 1 or -1. Assets divide into safe and “not safe.”
  • Problem: some assets always fall into the “not safe” bucket, but what falls into the safe bucket can vary. Long Treasuries and commodities could be examples of assets that vary during a crisis, depending on the type of crisis.
  • It does not recognize multiple time horizons easily. Bonds held to maturity have a different risk profile than a constantly rebalanced portfolio.
  • Risk is the same for all people, and their decision-making time horizons are the same as well.
  • And more…

I’m still playing around with the elements of what would make up a new asset allocation model, but a new model has to disaggregate risk into risks, and ask some basic questions:

Complete Story »