Quantitative risk analysis would have raised red flags on Madoff - Investors crushed by Madoff's Wall Street fraud

Quantitative risk analysis would have raised red flags on Madoff - Riskdata
The Madoff situation raises two important issues: the place of risk management in the investment process, and the quantitative techniques used to assess the risk.

Although Madoff's fraud falls into the category of "operational risk", Riskdata's analysis shows that quantitative risk analysis can help detect this type of fraud by statistical analysis of returns. This analysis should be used to wave red flags on funds that clearly manifest statistical inconsistencies and contradictions between declared strategies and posted results.

Riskdata has identified two main Red Flag cases, based either on returns (abnormal bias ratio) or on risk profile inconsistencies.

Red Flag #1: The Bias Ratio
The Bias Ratio - invented by Adil Abdulali of Protégé Partners and available in Riskdata's analytics suite - can detect return smoothing practices and possible performance manipulation. While Madoff's funds did not report to the HFR database, his series of returns is identical to that of one of his feeders, Fairfield Sentry Ltd., managed by Fairfield Greenwich Group.

In the HFR database, Fairfield Sentry is classified as "Equity Hedge", a category which contains 2290 funds. Since 2005, Fairfield Sentry's Bias Ratio ranges between 6 and 7, when Equity Hedge funds are mostly are 1-3. The acceptable range for the Bias Ratio depends on the liquidity of traded securities: up to 3 for very liquid funds, up to 20 or 30 for the most illiquid ones. Madoff claimed that he traded stocks from the S&P 100 index and index options, all being quite liquid securities, another proof that his Bias ratio should not be above 3.

Red Flag #2: Inconsistent Risk Profiles
Madoff has given a very precise description of his strategy, which he called "split strike conversion", making it possible to simulate his returns. Several investors asked to invest in his funds did the exercise of trying to replicate his performances and concluded with a "no go" because they couldn't come even close to his advertised returns.

Madoff's description of his strategy is not complete enough to allow its fulull replication, so Riskdata simulated a range of parameters of the "split strike conversion" strategy, and discovered that whatever the parameters, one cannot reach his extraordinary posted performance.

In addition, even when performances themselves cannot be replicated, the risk profile generally stays the same, i.e. relevant risk factors are the same between the original fund returns and the replication. Madoff's (weak) exposure to interest rate factors, and lack of exposure to equity and volatility factors, is inconsistent with his advertised strategy, which is mostly sensitive to US equity and equity volatility factors.

Long-short Equity Hedge funds can be divided into two very distinct buckets: funds whose risk profile is close to that of Madoff, and all the others. Most of funds close to Madoff's risk profile were either directly managed by Madoff, or feeders of his funds. Moreover this group are all extremely highly correlated with each other - typically above 95% correlation. Suspiciously, the only way to replicate Mr Madoff's performance is to invest in his funds!

Digging further into the risk analysis of his performance, the factors to which it is sensitive tend to change from month to month, which is surprising for such a stable strategy. The cluster of funds that were essentially invested in Madoff followed an improbable path across various risk profiles, one that was followed by no other fund, even temporarily.

Conclusion

Properly used, quantitative risk systems provide an unprecedented insight into funds actual behaviour and their consistency with declared strategy - this is the lesson fund managers must learn. The Madoff case also shows that quantitative risk assessment must become central not only to investors, but to the SEC's approach to risk management. While the Bias Ratio may not detect all types of fraud, along with Risk Profiling it is a very accurate instrument to check the likeliness of posted returns, and verify the consistency of actual performances vs. managers' statements. The Bias Ratio can easily be applied by the SEC in the same manner as the Benford's Law is used by the IRS - a selective tool to spot a possible fraud. Combined with Risk Profiling and Strategy Description Analysis, the Bias Ratio may well be the most effective instrument of fraud detection.

The affair however also demonstrates the fundamental need for independent risk management, preferably with a third party risk management tools (as opposed to the in-house). Independent risk management is the key to successful investment and it needs to be independently assessed on all levels. Risk management should also go beyond basic risk indicators because they may not adequately capture the true portfolio situation. Lip service to risk control and colorful charts in the annual report will never replace hands-on approach to risk. A minimum check-list of risk indicators should be required by investors, who in turn should be equipped with necessary tools to verify and challenge managers' reports. To prevent the next Madoff scandal, the industry standard should not settle for less than an independent truly decision-supporting risk management that relies on adequate quantitative instruments at all stages of the investment process.