When an investor hires a manager, it has beliefs about the future performance that the manager will generate. Typically, the manager is given three years to do its job. The investor then reviews the manager in light of the performance that it generated. In this paper, we formulate a Bayesian model that enables an investor to perform such a review.

In doing so, we reveal some profound issues with many of the prevailing conventions of performance evaluation.

It becomes clear, for example, why clients often unwittingly and erroneously inflate their expectations about the impact that a manager's performance will have on their portfolio. We also learn that it is crucial for investors to understand the mean-reversion of their manager's excess returns.

Once investors have this understanding, they can use this model, alongside other approaches, as a form of expert system. That way, if the manager’s performance is unexpectedly good or bad, the investor can use the model to gauge its likely response to this performance. This is particularly helpful, and value enhancing, as the model often recommends the opposite action to what the investor would prefer from a behavioural perspective.

(The full paper is available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2101771. A ten-minute version of the presentation is also available as a screencast, should you wish.)

*Awarded Consulting Track Prize