Modern portfolio theory relies on a number of assumptions that do not necessarily hold with illiquid assets, namely that sound historical data on returns is available, that liquidity is roughly the same across the asset classes and that the risk characteristics of the assets are similar and well understood. “It assumes that returns and their correlations are observable, and, more importantly, that they are ‘normal,’ but that is not necessarily the case with illiquid assets,” explains Mary Duke, New York-based head of wealth advisory services for the Americas at HSBC Private Bank. By normal, Duke means that the probabilities of an asset making or losing money, when plotted on a graph, are distributed around the mean and form a so-called normal distribution (commonly called a bell curve). The normal distribution allows an analyst to use the full power of statistical modeling on an asset. But if the asset’s behavior is not normal—and few assets really are—statistical model results can be profoundly wrong.

Nearly all the factors MPT needs are absent in illiquid investments. Take hedge funds, for example, which have only emerged in relatively recent times as important investment vehicles. Unlike equities, which have been traded on established markets for more than a century, reliable recorded data on hedge fund performance is nearly impossible to find. In addition, hedge fund investments are far less liquid than equity investments because they typically lock investors in for a period of time. Similarly, it is impossible to derive risk, return and correlation figures to private equity or private real estate in a truly accurate way.

Experts who have attempted to apply MPT to illiquid assets have found that if this process is done naively, it can lead to outrageous assumptions. “The underlying asset would appear less risky than it is in reality,” says Geneva-based Jacques Roulet, head of quantitative analysis and risk management at ABN Amro Private Clients.