Asset allocation models have been dominated by Markowitz mean-variance optimization. Even today, commercially available asset allocation software generally ignores estimation error. For the past 50 years, the investment and actuarial community has overlooked the fact that mean-variance optimization and its variations have little, if any, inherent investment value. Recent research shows that a better optimizer often dominates traditional mean-variance optimization even with refined risk-return estimates.

This session presents a condensed review of the classical material and recent research including:
• Problems with optimizers: how severe
• Ignoring estimate uncertainty
• A double patented procedure: “Resampled Efficiency
  optimization
• Why good inputs are not enough
• Implications for practice