Detecting Abnormal Return Patterns
One of the main reasons why performance and risk measures fail to adequately expose certain risks such as fraud in hedge funds is the uniqueness of hedge fund strategies and related operational issues in executing these strategies. Hedge fund returns are not as easily decomposed and attributable to explicit risk factors as long-only funds.
That said, quantitative analysis can provide a clue to potential return manipulation and fraud. A number of authors, including Getmansky et al., have argued that the presence of autocorrelations in hedge fund returns is an indication of fraud, while others have argued that it is simply an indication of the illiquidity of the underlying assets held by hedge funds.20 Bollen and Krepely have claimed that the presence of positive autocorrelation can be an indication of return-smoothing, though Asness, Krail and Liew consider it evidence of stale pricing which can be benignly incidental or malignantly intentional.21
Any quantitative assessment of an individual fund's returns must be evaluated with respect to its strategy peers. This is because certain strategies display higher positive autocorrelation than others. Autocorrelation estimates are higher for convertible arbitrage, distressed securities and fixed income arbitrage and very low for equity and global-macro strategies mainly because of the differentials in liquidity of the underlying securities used in each strategy.
As autocorrelation alone is not evidence ...