Forecasting Fund Manager Alphas: The Impossible Just Takes Longer

Expected alpha from active fund managers can be forecasted—as long as one is mindful of the rules of the zero-sum game of investing. Explicit forecasts are preferred over implicit forecasts because sponsors can use explicit forecasts to build optimized portfolios of managers with improved manager weighting. To make explicit alpha forecasts, the investor combines two equations derived from the fundamental law of active management. The elemental variables for the equations are the sponsor’s estimate of the manager’s “goodness” at beating the manager’s benchmark, the sponsor’s assessment of the sponsor’s skill in estimating manager ability, the cross-sectional standard deviation of manager skill, portfolio breadth, implementation efficiency, expected active risk of the portfolio, and fees. Most plan sponsors (by which, we also mean foundations, endowments, individual investors, and anyone else facing the task of evaluating professional investment fund managers) chafe at and resist as undoable the suggestion that they should make explicit alpha forecasts for the active fund managers they hire or consider hiring. But fund manager alphas can and should be forecasted.The game of active management is difficult to win. The zero-sum nature of the game makes winning impossible for most fund managers and plan sponsors. Not everyone can be above average. To win this zero-sum (negative-sum after fees and costs) game, a fund manager must have exceptional skill. And if one is a sponsor considering hiring active fund managers, winning is even more difficult: To justify hiring active managers at all, a sponsor must also have a special skill—a skill for identifying the skillful active managers. So, two kinds of skill are necessary if active management is to be successful for a sponsor—one set of skills at the manager level and one at the sponsor level.Both sponsor skill and fund manager skill can be estimated, quantified, and incorporated into an explicit alpha forecast by using the framework developed in this article. By combining equations for the fundamental law of active management and a new version of the forecasting equation, one can derive a formula for building an alpha forecast out of elemental parts. The formula contains seven variables to be estimated:the sponsor’s forecast of the manager’s “goodness” at beating the manager’s benchmark, expressed as a z-score;the sponsor’s self-assessment of its own ability to select good managers, expressed as an information coefficient for the sponsor;the cross-sectional standard deviation of information coefficients for the manager population (this variable cannot be observed directly, but a reasonable value for it can be inferred);the breadth of the portfolio, where breadth is the number of independent active management decisions made per year by the portfolio manager;the transfer coefficient, or implementation efficiency, of the portfolio, which accounts for the performance drag from constraints (principally the no-shorting constraint);the expected active risk of the portfolio; andfees.The first two variables are difficult to estimate because they involve forming (and then perhaps defending) a belief that one can beat the market. Investors who deeply understand the difficulty involved in beating the market are likely to be reluctant to make forecasts of these two variables; yet, these investors are, paradoxically, the best placed to try to make these forecasts. And the fact is that investors cannot approach the task with an expectation of success if they do not think they have the skill to win.Alpha forecasts made by using this approach will be properly adjusted for differences in risk, differences in fees, and other differences, so the various funds will have alpha estimates that bear reasonable relationships to one another in light of such differences. Ultimately, however, the estimates will be only as good as the sponsor’s estimate of its own skill—which is as it should be.Investors should be encouraged to use the method described in the article for explicit alpha forecasting because it brings discipline and structure to the process of building portfolios of managers and because the forecasts are required for optimizing the structure of the portfolio of managers. An optimized structure gives the most weight to managers who have the highest alphas and the lowest risks, a result that almost never happens by using intuition alone.