Risk-Adjusted Performance Measurement
Investment results can be calculated, reported, and to a considerable extent analyzed without recourse to ex post risk measures. For example, the arithmetic difference between portfolio and benchmark rates of return conveys some information about the worth of active management; attribution analysis on the Brinson model isolates the value added by the manager’s sector weighting and security selection decisions; triangle charts display benchmark-relative results over historical periods that may include up and down markets; and quality control charts test active returns against the null hypothesis that the manager has no investment skill.
But investment results cannot be well and truly evaluated without explicitly taking ex post risk into account. As Carl Bacon remarks, “If a manager achieved a reasonable but unexciting return, but took very high risk to achieve this return, then the client should be very disappointed.” Ex post risk analysis can also provide risk managers with valuable feedback on the reasonability of ex ante projections. Bacon says, “Risk controllers in particular need to be grounded in reality; it is essential that they compare realized risk with their previous forecasts of risk.”
Bacon’s new book, Practical Risk-Adjusted Performance Measurement (Wiley, 2012), is an exhaustive compendium of ex post risk measures written for risk and performance practitioners from a buy-side asset management perspective. One of the author’s stated purposes was to document, in a structured format, as many discrete measures as possible, and he succeeded admirably. For instance, in addition to excellent, illustrated explanations of commonplace measures such as skewness and kurtosis, an early chapter on descriptive statistics introduces less familiar indicators, including the Bera-Jarque statistic, which tests for normality; the Hurst index, which tests for mean reversion; and Abdulali’s trade-marked bias ratio, which signals possible return smoothing. Despite its brevity (some 200 pages), Practical Risk-Adjusted Performance Measurement is a hefty book, and it will undoubtedly stand as the reference manual for many years to come. For that reason alone it merits a place in every practitioner’s and system developer’s professional library.
However, this work is not ‘merely’ an encyclopedic treatment of the subject. Like its predecessor, Practical Portfolio Performance Measurement and Attribution (2nd ed., Wiley, 2008), it presents the mathematics with no more complexity than strictly necessary, explains the concepts and calculations in plain language, and provides many worked examples. It is, accordingly, an incomparable instructional resource, one made all the more readable by Bacon’s personal observations. For example, he writes, “Some commentators would suggest that negative Sharpe and information ratios have no value, pointing out that if performance is negative the respective ratios actually reward higher variability and higher tracking error. I disagree—it seems self-evident to me at least, that if you are going to underperform it is better to underperform inconsistently rather than underperform consistently.” Practical Risk-Adjusted Performance Measurement is a technical work written in the author’s own voice, and that’s a rare achievement.
In short, I strongly recommend this book to performance and risk professionals seeking to extend their mastery and improve their practice of both disciplines.