optimizer
Maximum Diversification
An optimization that maximizes the ratio of weighted average volatility to portfolio volatility, pushing capital toward less correlated assets.
Maximum Diversification is an optimization that picks the weights with the highest diversification ratio: the weighted average volatility of the holdings divided by the volatility of the portfolio as a whole. It pushes capital toward assets that move independently of one another, so their individual swings offset as much as possible. Yves Choueifaty and Yves Coignard introduced it in 2008.
The diversification ratio is the sum of each weight multiplied by that asset's volatility, divided by portfolio volatility. The numerator is what portfolio volatility would be if every pair of holdings were perfectly correlated; the denominator is the actual volatility once correlations are counted. For a long-only portfolio the ratio is at least 1, equals 1 for a single asset or perfectly correlated holdings, and rises as correlations fall. For an equal-weight portfolio of N independent assets with the same volatility it equals the square root of N, which is why its square can be read as an effective number of independent bets.
Higher ratios mean more of the holdings' individual risk cancels out. For instance, an equal-weight portfolio of many stocks with similar volatility and an average pairwise correlation of 0.3 has a ratio near 1.8. The method ignores expected returns, so it can overweight a low-correlation asset with poor prospects, and it relies on correlation estimates that can rise in crises, shrinking the benefit when it matters most.
Example with illustrative numbers: two assets each have 20% volatility and a 0.25 correlation, held 50/50. Their weighted average volatility is 20%, and portfolio volatility is about 15.8%, so the diversification ratio is about 1.26x. Squared, that is about 1.6 effective independent bets. If the correlation were 1, the ratio would drop to 1.0.
On GNG, the Portfolio Optimizer's Max Diversification engine maximizes this ratio exactly when no limits get in the way, using the covariance estimate from your lookback window. When the unconstrained answer breaks your single-name cap or sector caps, or when you set a yield floor, a dividend-growth floor or a turnover limit, it switches to an approximation that minimizes variance while rewarding weighted asset volatility, with those limits built in, so the result may not reach the highest ratio your limits allow. It is also the engine behind the Maximally diversified build goal.
Correlation drives the ratio, and Risk Parity is another way to spread risk evenly across holdings.
Related terms
More optimizer terms
Get full access to GNG Research
Create a free account to access portfolio tracking, advanced tools, and more.