AQR Shows How to Construct Portfolios by Going Beyond MVO

AQR Shows How to Construct Portfolios by Going Beyond MVOPortfolio construction is hard enough when comparing assets such as long-only stocks and bonds – add in alternatives and the challenges are multiplied. How does one “put it all together?” That’s precisely the question explored by AQR’s Q2-2015 edition of Alternative Thinking, “Strategic Portfolio Construction: How to Put it All Together.”

“Many investors ask for our thoughts on putting it all together,” writes AQR in the concluding thoughts of the white paper. “This article does not give definitive answers, or even recommend a particular framework for making such decisions.” Instead, the paper starts with AQR’s “quantitative home territory” – expected returns, volatilities, and correlations – and then examines how investor-specific beliefs and constraints can inform and interact with formal optimization methods. The most prominent of these methods is Mean Variance Optimization, or MVO.

Mean Variance Optimization

The authors of the AQR paper believe MVO is a useful tool for portfolio construction, but warn that it has “many pitfalls.” Investment decisions such as how much leverage and shorting to allow, or how much to allocate to alternatives, are often impossible to make through MVO alone, as are decisions on how to weigh across alternative risk premia versus illiquid assets or manager-specific alpha – these decisions must almost inevitably be made according to investor-specific beliefs and constraints.

But MVO does help with one very important top-down investment decision: how to deal with portfolio volatility. MVO allocates across investments on the basis of volatility, but then tilts toward investments with higher Sharpe ratios and/or lower correlations. It works best with a small number of comparable assets, but less well when combining traditional and alternative investments. In such cases – and without investor-specific constraints – MVO tends to load up on leverage and illiquidity, creating what many would consider unbalanced portfolios.

Constraining MVO

Constraints can be placed on MVO, but MVO itself offers no guidance as to what those constraints should be. AQR notes that sometimes laws, regulations, and liabilities force constraints on MVO, but more often than not, constraints are self-imposed. This might sound like a good thing, but the constraints are frequently guided by peer behavior, which can create “gradual herding among institutions,” according to AQR.

Instead, investors should carefully consider their own beliefs and preferences, and then make choices that combine the broad spirit of optimization, by which AQR means wide diversification with a tilt toward higher risk-adjusted returns. The five key choices investors must make are:

  1. Risk tolerance
  2. Leverage vs. concentration
  3. Liquid vs. illiquid
  4. Long only vs. long/short
  5. Systemic vs. idiosyncratic returns

Model Portfolio Examples

Although AQR doesn’t give “definitive answers,” it does offer five examples of “putting it all together.” The five model portfolios listed below are consistent with AQR’s core belief in more aggressive diversification than traditional portfolios – using leverage and shorting – but they’re all different, and designed to appeal to different investor-preferences. Investors can judge which portfolio characteristics fit them best.

Five Candidate Portfolio Allocations with Performance and Risk Statistics

As you can see from the image above, each of the five portfolios are tailored to different risk tolerances and investment outlooks. Allocations for each according to capital and risk are shown, across “60/40” smart beta, risk parity, alternative risk premia, and endowment alternatives. The 3-Layer, Reduce Leverage, Add More Liquidity, and Increase Diversification portfolios are all based off the Base Case portfolio, with “tweaks” made in order to satisfy different preferences.

For more information, visit aqr.com to download a pdf copy of the white paper.

Add a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.