If you’ve ever tried to build an investment portfolio for a client using a rigorous, logical methodology, then you’ve encountered a lot of problems. If the client has a long-term investment horizon the problem gets worse. If you attempt to quantify your methodology – good luck. Well, in the fall of 1997, we decided to do just that. The following series of articles have been written to provide the sophisticated practitioner with a guide to how we solved some of the problems we encountered.

At the end, we came to a number of conclusions and made some observations of current investment practices:

• The main benefit to developing a quantitative solution is that it generates a great deal of discipline in understanding what the key issues are and how to build an optimal portfolio.
• It is extremely difficult to use only quantitative methods. There are too many statistical and theoretical pitfalls, some of them quite subtle. This will frequently result in erroneous and/or unstable methodologies. Therefore, the best approach is to adopt quantitative methods to a point and then add qualitative experience to refine the solution.
• When we applied our model in the spring of 2000 to a large number of pension funds, we found the vast majority were overweight in risky assets.
• A major problem we discovered is that risk is improperly defined from the investor’s perspective. Frequently, portfolio manager performance and their risk measures are used to define investor risk. In effect the liability component of Asset Liability Management is naively applied.
• One byproduct of our work was the development of an option pricing model for longer term options. We found that the Black-Scholes Model undervalues 5 year employee options by approximately 30%.