Risk-Adjusted Performance and Informed Decision Making Dr. Arun Muralidhar
2 Arun Muralidhar - Bio Chairman of Mcube Investment Technologies, LLC and Managing Director at FX Concepts, Inc. Head of Investment Research and Member of Investment Management Committee, World Bank Investment Department, Derivatives and Liability Management, World Bank Funding Department, Managing Director and Head of Currency Research, JPMIM, BA, Wabash College (1988); PhD, MIT Sloan (1992)
3 Agenda Background Why the information ratio is wrong – M 2 Risk budgeting – connecting returns to risk – M 3 Confidence in skill: History matters - SHARAD Optimal portfolio construction using these measures
4 Relative Risk – Tracking Error Most commonly used measure Ann. standard deviation of excess returns Depends on the standard deviation of the benchmark, strategy and correlation Few test ex-ante forecast with ex-post outcomes Papers that recommend that managers stay within tracking error ranges are WRONG
5 Performance Measures Returns – Absolute and Relative Annualized versus Cumulative Ratios – Risk-Adjusted Sharpe, Information Ratio, Sortino Ratio Risk-Adjusted Returns M2, M3, SHARAD Skill Measures
6 How to Calculate Standard Measures? Unadjusted measures Excess return = Portfolio return - Benchmark return Risk-adjusted ratio measures Sharpe ratio = Excess over risk free rate/standard deviation of portfolio –higher the ratio, the better the investment opportunity Information ratio = Excess over benchmark/standard deviation of excess returns –higher the ratio, better the manager
7 Some Advanced Risk-Adjusted Measures Sortino ratio = Excess/Downside risk measure Risk-adjusted measures (in return terms) M 2 = extended Sharpe ratio (in basis points) M 3 = extended M 2 ratio; corrects for correlation SHARAD = Normalizes for different length of history Confidence in Skill – Tells how confident one can be that there is skill (as opposed to noise) in a given track record
8 Riskless asset Return Active portfolio A Standard deviation of unlevered portfolio Benchmark Standard deviation Market risk Active portfolio B Information Ratio is Wrong - M 2 Need to normalize for different Std. deviations B has a negative Information Ratio!
9 Riskless asset Benchmark Return Benchmark Standard deviation Market risk Portfolio B has a Higher M 2 Return Problem: Need to normalize for different correlations i.e., have different tracking error M 2 Return for B M 2 Return for A
10 Correlation-Adjustment: M 3 Measure FundReturn (%) Standard deviation (%) M2M2 TE(basic) (%) TE(M 2 ) (%) M3M3 (1)(2)(3)(4)(6)(7)(8)(12) F B Correcting for tracking error – different rankings F = Risk-free asset; B = Benchmark (S&P500)
11 Ranking Portfolios: Different Methods RankingUnadjusted Skill using raw returns M2 or Sharpe Skill using M2M3 Skill using M3 Information ratio (1)(2)(3)(4)(5)(6)(7)(8) First Second Third Fourth Fifth Sixth Seventh Eighth Ninth Tenth M 3 is the only one consistent with Skill Information ratio, Sharpe or M 2 say little about Skill Skill = Confidence in Skill Measure M 3
12 Which Measure Should you Use? Manage portfolio yourself – Sharpe, Information Ratio or M 2 External manager and with a tracking error budget – M 3 (Previous papers ignore possible actions by client) Worried about skill – M 3 M 2 provides valuable advice on leverage/ deleverage; M 3 provides valuable advice on (a) active versus passive (“beta”) and leverage/deleverage Important to have a risk budget
13 How Do You Compare 2 Strategies with Different Data Histories? In the past – drop non-overlapping data Lose valuable information on strategy with longer history SHARAD Measure – Normalizes for different risk and different data history SHARAD = {M 3 return}*{Confidence in skill (of M 3 portfolio)} Confidence in skill acts as a probability measure and explicitly captures the length of data history – more history, greater the confidence in skill
14 Risk Budgeting is only First Step Cannot force tracking error ranges on managers – client needs to be evaluate how good manager is in dynamically managing risk Risk-adjusted performance measures are the right way to go, but many are paid on the basis of excess returns Measures can tell you how good manager is at managing risk Use creative measures along with risk budgets to achieve optimal portfolio performance