When is Volatility a Good Thing?

(June 18, 2013) — A new framework for portfolio optimization has been developed in which the direction of returns is used to determine asset allocation strategies. 

Designed by Professor Joonas Hamalainen, a Finnish professor working with the University of Turku, the new framework uses developed equations to produce an improved portfolio composition by better predicting the direction of returns.

Forecasting how much an asset’s value is going to rise or fall is a near impossible task, Professor Hamalainen argued. Therefore, the most prominent forecasts of asset returns are often the directions, or signs, of future returns.

The paper was constructed on two major premises:  evidence that return signs are predictable, and investor behavior in estimating future performance of assets.

By using asset return predictions to construct a portfolio, instead of the traditional route of using a mean-variance framework, Professor Hamalainen claimed there was no large-scale negative correlation between asset returns, because it presents the possibility of the investor being either right or wrong on both assets’ signs simultaneously, therefore actually increasing the variance of the entire portfolio.

Another major difference from the traditional mean variance method is that under Professor Hamalainen’s guidance, investors seeking to maximize returns should pick assets with higher volatility. 

His study found the framework performed better when put through a trading simulation, compared to the performance of a traditional mean-variance optimization portfolio.

The paper also found that correlation between absolute returns of the assets plays a role in portfolio selection.

Is this a weird quirk that only works with the S&P100? It appears not. Concerned that the results may differ when using other indices, Professor Hamalainen repeated the study with two different datasets of stocks.

“It would be natural to expect that if the forecasting accuracy drops, the advantage of the new framework is also lowered. However, it is still expected to be significant in a wide variety of setups,” he said, with a promise to report the results of the robustness tests at a later date.

In addition, more details about the turnover of wealth (to examine what kind of a role trading costs would play), and the amount of wealth allocated to bets that are against the investor’s views will also be analyzed and reported.

The full paper can be read here.