June 14, 2011

How to Arrive at Better Capital Markets Forecasts

 

Capital markets forecasts are fundamental to building effective portfolios. In our last few articles, we’ve focused on the pitfalls that advisors often encounter in trying to determine forecasts. Now let’s look at the other side of the coin: Just what is an optimal process? How can we develop forecasts that actually make a difference in portfolio construction?

As an advisor, your goal is to estimate, as accurately as possible, expected returns and risk levels across multiple asset classes. It is our view that over the long-term the markets are efficient. Although over the short term bubbles and mispricing do occur, current market prices are ultimately our best guide to deciphering the long-term, equilibrium value of an asset.

Advisors who seek to improve their forecasting must find a method that is statistically robust —portfolios are only as good as the forecasts they are built on. Following is a guide to the most optimal approach we use.

Step One: Establish a Baseline

In order to estimate what assets will be worth, advisors must first determine the relative size of their present value. Calculating the world market portfolio by viewing various indexes and determining the total value as represented by the market prices of each provides an overview of the market, as well as how various asset classes are valued within it. 

Step Two: Refine Input

An optimal portfolio is one that yields the best combination of risk and reward – in other words, investors get the most returns while taking on as little risk as possible. In order to determine what this portfolio looks like, advisors rely on mean variance optimization, which crunches any given set of returns, standard deviations and correlations among assets and produces a set of optimal portfolios.

Of course, mean variance optimization is only as good as the data that goes into it. To calculate estimates of standard deviations and correlations, we utilize a method developed by Robert Stambaugh, an economist who is a professor of finance at the University of Pennsylvania’s Wharton School and a leading researcher of empirical asset pricing.

This sophisticated method sidesteps the problems associated with using short-term common historical data to derive expected returns. Whereas other methods might eliminate data based on market discrepancies, the Stambaugh method uses all available data to come up with estimation error for its risk-return estimates. His method also uses longer historical periods for each asset class to infer the volatility and co-volatility of shorter asset classes; the method can be improved by selecting a model for each individual asset class.

The result: a higher estimate of risk than historical data alone can provide and, therefore, more accurate capital markets forecasts.

Step Three: Extract Expected Returns

Since we have estimates for standard deviations and one portfolio we would like to use these to extract the

 

returns expected by the market. To do this, many use a method known as Black-Litterman.

Asset values from the world market are used to determine the value of every asset in an optimized portfolio. Reverse optimization and Black-Litterman are then used to extract the expected returns for each asset class given the standard deviation and correlation from the world market portfolio and a few other inputs.

Black-Litterman has emerged as the optimum method for portfolio allocation because it does not require advisors to guess at expected returns. Instead, it assumes the asset allocation of a portfolio should be proportional to the market value of available assets. Further, Black-Litterman allows advisors to impose their own views on asset returns to the allocation.

Determining capital markets estimates is both an art and a science. No one mathematical formula will produce an optimal portfolio – advisors must be able to utilize these tools while also inputting their own knowledge of assets and the markets. For advisors to improve their forecasts, they must understand the data that goes into these equations, as well as the limitations of these systems.

Disclaimer: Past performance is not indicative of future results. The opinions expressed herein reflect our judgment as of the date of writing and are subject to change at any time without notice. They are not intended to constitute legal, tax, securities or investment advice or a recommended course of action in any given situation. Information obtained from third party resources are believed to be reliable but not guaranteed.

Page 2 of 2
Single page view Reprints Discuss this story
This is where the comments go.