Maximum Diversification Portfolio
The maximum diversification portfolio (MDP) is the portfolio that maximizes the diversification ratio — defined as the ratio of the weighted-average volatility of individual assets to the portfolio volatility — thereby achieving the greatest possible benefit from diversification across all available assets. Unlike mean-variance optimization, the MDP requires no expected return inputs, relying solely on risk estimates.
Key takeaways
- The diversification ratio equals the portfolio's weighted-average component volatility divided by the portfolio's total volatility; the MDP maximizes this ratio.
- A diversification ratio greater than 1.0 indicates positive diversification benefits; higher ratios imply greater risk reduction from combining assets.
- The MDP overweights assets with high idiosyncratic risk relative to their systematic risk — assets whose risk is most distinct from the rest of the portfolio.
- Unlike minimum variance optimization, the MDP is not corner-solution prone and tends to produce more balanced allocations across the asset universe.
- The MDP was formalized by Yves Choueifaty and Yann Coignard in their 2008 paper 'Toward Maximum Diversification,' published in the Journal of Portfolio Management.
Explanation
The maximum diversification portfolio addresses one of the fundamental tensions in portfolio construction: the desire to reduce risk through diversification versus the need to specify expected returns, which are notoriously difficult to estimate accurately. By eliminating expected return inputs from the optimization entirely and focusing solely on the covariance structure of asset returns, the MDP offers a more robust alternative to classical mean-variance optimization for risk-oriented portfolio construction.
The diversification ratio, the objective function of the MDP, has an intuitive interpretation. When assets are perfectly correlated (correlation = 1), the portfolio volatility equals the weighted average of individual volatilities, and the diversification ratio equals exactly 1 — no diversification benefit exists. As correlations decrease below 1, portfolio volatility falls relative to the weighted-average component volatility, and the diversification ratio rises above 1. The MDP finds the allocation that maximizes this ratio — the portfolio that extracts the maximum possible benefit from the diversification available in the asset universe given the current correlation structure.
Mathematically, the MDP solution concentrates allocation in assets that have high stand-alone volatility but low correlation with the rest of the portfolio. These are assets whose risks are most 'idiosyncratic' relative to the broader portfolio — they add significant diversification benefit because they don't move in lockstep with other holdings. In practice, this often leads the MDP to overweight asset classes with low cross-asset correlations, such as commodities, real assets, or alternative risk premia.
One important property of the MDP is that it coincides with the maximum Sharpe ratio portfolio when all assets have identical Sharpe ratios. This equivalence provides a theoretical justification for the MDP as an efficient choice when expected returns are difficult to estimate reliably — it effectively assumes that the expected risk premium is proportional to volatility across all assets, which is a weaker assumption than requiring precise return forecasts. Research has shown that the MDP tends to outperform equal-weight and cap-weight portfolios on a risk-adjusted basis over long historical periods.
Formula
DR = (Σ w_i σ_i) / σ_p; Maximize DR subject to Σw_i = 1, w_i ≥ 0
Example
An asset allocator constructs a MDP across five asset classes: U.S. equities (σ=15%), international equities (σ=18%), corporate bonds (σ=7%), commodities (σ=20%), and real estate (σ=14%). Using the estimated correlation matrix, the optimizer finds that commodities and real estate have the lowest correlations with equities and bonds. The MDP solution overweights these two asset classes significantly relative to a cap-weighted or equal-weight baseline. If the weighted-average component volatility is 15.2% and the portfolio volatility is 10.4%, the diversification ratio is 15.2/10.4 = 1.46, compared to an equal-weight portfolio that might have a diversification ratio of 1.28.
Related terms
Basis Calmar Ratio Cap Correlation Correlation Matrix Covariance Diversification Dynamic Asset Allocation Equal Weight Portfolio Mean Variance Optimization Portfolio Rebalancing Premium