 |
BMFR 3nd Qtr 1999
How to Avoid Bad Managers Even When They Have Good Performance
Every year, hundreds of billions of dollars in new capital are allocated to money managers. Until recently, manager selection methods were non-quantitative and ad hoc. Given the large and increasing importance of this allocation process, there is an acute need for objective and quantifiable manager selection methods.
From March 1996 to February 1997, the Russian stock market, as measured by the Russian Trading Systems Index, rose approximately 600%. At least one mutual fund and several hedge funds benefited greatly from one-sided positions in this market. When Russia subsequently collapsed, many, if not most, of these funds suffered significant declines.
Fortunately, in the 1990s a small but informative literature has appeared on the persistence of money manager returns. One characteristic that all studies of manager performance persistence have in common is that explanations of manager performance are provided ex post facto, after certain regression models and correlations have been analyzed. We believe that creating causal hypotheses prior to an examination of the empirical data can lead to a qualitatively different understanding of manager performance. In this section we present two causal factors, which we term the Correlation Effect and the Skewness Effect, which we believe play an important role in manager performance. The effect that these two factors have on manager performance is discussed below.
The Correlation Effect refers to the effect that one-sided positions in markets with extended trends have on manager performance. By inflating manager performance, the Correlation Effect leads to overestimates of the abilities of the manager. The Correlation Effect also applies to market characteristics, such as volatility, and to market relationships, such as yield spreads. The intuitive claim is that managers who can benefit from increases and decreases in markets, market characteristics, and market relationships are superior to those who can only trade one side of these changes.
Eliminating the Correlation Effect by applying traditional performance metrics is quite difficult. One suggestion is to compute various benchmark-relative statistics such as beta and select managers who have strong benchmark-relative performance. However, there is evidence that benchmarks are somewhat arbitrary; and even when they can be objectively determined, benchmarks are often not fine-grained enough. For example, there is no standard widely used Russian benchmark.
Another way to avoid investing in managers who benefit from the Correlation Effect is to examine their investment records on a trade-by-trade basis. If a substantial fraction of the profits come from trading one idea or from trading on one side of the market, then superior track records must be treated with caution.
However, trade-by-trade data are not easily available. Many managers consider their trades to be proprietary data and are loathe to provide it even to potential investors. In this paper we detect managers who benefit from the Correlation Effect, and are thus to be avoided, by computing the correlation of a manager's returns against major indices to infer a manager's positions. We call this metric correlation rank. Managers who have high correlation ranks are relatively uncorrelated with indices and are to be preferred to those that have low correlation ranks.
From the end of 1993 to the end of 1996, Victor Niederhoffer's firm had a compound annual rate of return of approximately 40% and an annualized Sharpe ratio of approximately 3. This ranks as one of the best-ever performances for directional CTAs over a three-year period. In 1997, however, Niederhoffer experienced a 100% loss and closed his firm. In a letter that Niederhoffer sent to clients shortly after the October collapse of his firm, he noted that a short out-of-the-money put position that his firm held on the day of the October 27 crash was one that his firm had often traded profitably in the past. In 1994 the S&P 500 traded in a narrow range of approximately 10%. In 1995 and 1996, the S&P 500 moved steadily upward. In this environment, a short out-of-the-money put option strategy would have been very profitable indeed."Selling deep out-of-the-money options only leads to consistent profits until a catastrophe arises. Then you lose it all, plus some."
- Blair Hull (Schwager, 1997)
However, as Blair Hull noted in the opening quotation, shorting options only gives the impression of providing consistent profits. Assuming that the options market is efficient, then indiscriminately selling out-of-the-money options will lead to steady gains until the event that Hull refers to as a catastrophe occurs and then the manager will lose all of the prior gains plus transaction costs. This occurred in 1995 when managers who had extensive covered call programs found that their risk-adjusted gains over the prior decade disappeared.
We call such strategies high probability low return investments. The probability that a particular trade will be profitable is very high, so the manager exhibits month after month of outstanding performance. However, when the catastrophe does occur, the investment strategy is shown to have a low or even negative return.
This phenomenon is very difficult to analyze and capture using standard performance statistics. Because the period of outperformance can last for several years, the fund of fund manager or other asset allocator will find it difficult to distinguish between a genuine edge and a high probability low return strategy. Indeed, in Victor Niederhoffer's case, his assets rose sharply during the three-year period between 1993 and 1996.
Fortunately, however, these strategies do have distinguishing characteristics that can be quantified. In particular, the returns of high probability low return strategies have negative skewness. For more information about the skewness of CTA returns consult "Barclay Research Report: The Positive Skewness of Managed Futures Volatility" by Sol Waksman and Randy Warsager (Barclay Managed Futures Report, 1st Quarter, 1998).
Based on the qualitative intuitions expressed in the previous section, we have two hypotheses about performance persistence:
Each of these hypotheses has a corresponding decision rule:
- Correlation Effect hypothesis: Managers who have relatively low correlations with certain indices are likely to perform better than their counterparts.
- Skewness Effect hypothesis: Managers who have relatively high skewness are likely to perform better than their counterparts.
- Correlation Effect decision rule: Rank managers by the absolute value of their correlation, from least to greatest, with each of the selected indices. Then rank the managers again, from least to greatest, according to the sum of the index correlation ranks, and select the managers in the top 10%. Note that because the ranking is done according to the absolute value of the correlation, the sign of the correlation, whether positive or negative, is irrelevant.
- Skewness Effect decision rule: Select managers with positive skewness.
We hypothesize that the set of managers selected by each rule will outperform the managers that are not selected.
Table 1 - Database Descriptive Statistics 4/91-3/95 4/95-3/99 4/91-3/99 Number of managers 137 137 137 Mean monthly return 1.38% 1.07% 1.23% Mean minimum monthly return -13.27% -11.48% -14.77% Mean maximum monthly return 23.88% 18.24% 25.96% Mean Sharpe Ratio 0.48 0.38 0.43 Lowest Sharpe Ratio -1.53 -1.35 -0.66 Highest Sharpe Ratio 2.89 2.28 1.78 Mean Standard Deviation 7.29% 6.01% 6.79% Lowest Standard Deviation 0.36% 0.15% 0.27% Highest Standard Deviation 19.64% 28.18% 22.40% The first column is for the 48 month in-sample period, the second column is for the 48 month out-of-sample period and the third column is for the entire 96 month sample period.
Beginning with performance histories of 450 managers provided by Barclay, we selected the 137 managers that had eight-year performance records. We limited our study to these 137 managers because short performance records would not accurately reflect the regularities that we were attempting to locate. Descriptive data for these managers is shown in Table 1.
To test the Correlation Effect decision rule, we computed correlations against the following five indices:
For the first four indices, correlations are computed against the monthly percentage change. For the volatility index, the correlation is computed against the absolute value of the index.
- S&P 500 index.
- Thirty-year yield index. This Chicago Board Options Exchange Index is the yield-to-maturity of the most recently auctioned thirty-year Treasury bond multiplied by 10. Note that data were available starting in November 1993.
- Commodity Research Bureau Index. An unweighted geometric average of seventeen commodity prices.
- Dollar index. The geometric average of the changes in a basket of ten major world currencies against the US dollar.
- Volatility index. An index based on the implied volatility of nearby options on the S&P 100 index.
The top 10% of the managers which have the best correlation rank using the algorithm described in the Hypotheses section above have Sharpe ratios on the out-of-sample data set ranging from a high of 1.32 to a low of -0.20. The average Sharpe ratio for these selected managers is 0.50 versus an average Sharpe ratio of 0.37 for the other 90% of the managers. Assuming a monthly standard deviation of 6.01% (which is the average standard deviation on the out-of-sample data, as shown in Table 1), this improvement in the Sharpe ratio amounts to a monthly excess return of 24 risk-adjusted basis points, an economically significant amount by any measure.
Note that the results are critically dependent on the choice of indices. We believe that the five indices used in this study will prove to be most valuable in analyzing the performance of directional commodity traders.
To test the Skewness Effect metric, we selected only the managers that had positive skewness on the in-sample data. Out of the 137 managers in the database, 124 had positive skewness while 13 had negative skewness. The managers with positive skewness had average Sharpe ratios of 0.41 on the out-of-sample data while the managers with negative skewness had average Sharpe ratios of 0.12 on the out-of-sample data. Assuming a monthly standard deviation of 6.01%, this amounts to a monthly excess return of 51 risk-adjusted basis points, an economically and statistically significant difference.
A manager who had a Sharpe ratio of 0.32 on the in-sample data illustrates the effectiveness of the skewness decision rule. This manager' s returns had a skewness of -0.11 and on the out-of-sample data the Sharpe ratio fell to -1.35, the lowest Sharpe ratio on the out-of-sample data in the entire database. Hence, the application of the simple skewness rule could have alerted an asset allocator to serious performance problems that were not readily apparent.
In this paper, we tested two ways of selecting managers. The Correlation Effect metric selects managers based on low index correlations while the Skewness Effect metric chooses managers with positive skewness. We found statistically significant evidence that the Correlation Effect metric chooses managers with above-average risk adjusted returns. Managers selected by this metric outperform those that are not selected by an average of 24 basis points per month, an economically significant amount. Similarly, we found that managers selected by the skewness decision rule outperform those that are not by 51 risk-adjusted basis points per month.
Michael de la Maza is a co-founder and principal of DaVinci Sentient Technology, L.P., a futures fund that uses proprietary adaptive software to trade futures.
RISK DISCLOSURE
WHEN CONSIDERING ALTERNATIVE INVESTMENTS YOU SHOULD CONSIDER VARIOUS RISKS INCLUDING THE FACT THAT SOME PRODUCTS USE LEVERAGE AND OTHER SPECULATIVE INVESTMENT PRACTICES THAT MAY INCREASE THE RISK OF INVESTMENT LOSS, CAN BE ILLIQUID, ARE NOT REQUIRED TO PROVIDE PERIODIC PRICING OR VALUATION INFORMATION TO INVESTORS, MAY INVOLVE COMPLEX TAX STRUCTURES AND DELAYS IN DISTRIBUTING IMPORTANT TAX INFORMATION, ARE NOT SUBJECT TO THE SAME REGULATORY REQUIREMENTS AS MUTUAL FUNDS, OFTEN CHARGE HIGH FEES, AND IN MANY CASES THE UNDERLYING INVESTMENTS ARE NOT TRANSPARENT AND ARE KNOWN ONLY TO THE INVESTMENT MANAGER.
WITH RESPECT TO ALTERNATIVE INVESTMENTS IN GENERAL, YOU SHOULD BE AWARE THAT:WITH RESPECT TO AN INVESTMENT IN A FUND, YOU SHOULD BE AWARE THAT:
- RETURNS FROM SOME ALTERNATIVE INVESTMENTS CAN BE VOLATILE.
- YOU MAY LOSE ALL OR PORTION OF YOUR INVESTMENT.
- WITH RESPECT TO SINGLE MANAGER PRODUCTS THE MANAGER HAS TOTAL TRADING AUTHORITY. THE USE OF A SINGLE MANAGER COULD MEAN A LACK OF DIVERSIFICATION AND HIGHER RISK.
- MANY ALTERNATIVE INVESTMENTS ARE SUBJECT TO SUBSTANTIAL EXPENSES THAT MUST BE OFFSET BY TRADING PROFITS AND OTHER INCOME. A PORTION OF THOSE FEES IS PAID TO CAPITAL MANAGEMENT PARTNERS, INC.
- TRADING MAY TAKE PLACE ON FOREIGN EXCHANGES THAT MAY NOT OFFER THE SAME REGULATORY PROTECTION AS US EXCHANGES.
A FUND'S OFFERING MATERIALS OR A MANAGER'S DISCLOSURE DOCUMENT DESCRIBES THE VARIOUS RISKS AND CONFLICTS OF INTEREST RELATING TO AN INVESTMENT AND TO ITS OPERATIONS. YOU SHOULD READ THOSE DOCUMENTS CAREFULLY TO DETERMINE WHETHER AN INVESTMENT IS SUITABLE FOR YOU IN LIGHT OF, AMONG OTHER THINGS, YOUR FINANCIAL SITUATION, NEED FOR LIQUIDITY, TAX SITUATION, AND OTHER INVESTMENTS.
- THERE IS OFTEN A LACK OF TRANSPARENCY AS TO THE FUND'S UNDERLYING INVESTMENTS. AS TO FUND OF FUNDS, THE FUND'S MANAGER HAS COMPLETE DISCRETION TO INVEST IN VARIOUS SUB-FUNDS WITHOUT DISCLOSURE THEREOF TO YOU OR TO US. BECAUSE OF THIS LACK OF TRANSPARENCY, THERE IS NO WAY FOR YOU TO MONITOR THE SPECIFIC INVESTMENTS MADE BY THE FUND OR TO KNOW WHETHER THE SUB-FUND INVESTMENTS ARE CONSISTENT WITH THE FUND'S HISTORIC INVESTMENT PHILOSOPHY OR RISK LEVELS.
- A FUND OF FUNDS INVESTS IN OTHER FUNDS AND FEES ARE CHARGED AT BOTH THE FUND AND SUB-FUND LEVEL. THUS THE OVERALL FEES YOU WILL PAY WILL BE HIGHER THAT YOU WOULD PAY BY INVESTING DIRECTLY IN THE SUB-FUNDS. IN ADDITION, EACH SUB-FUND CHARGES AN INCENTIVE FEE ON NEW PROFITS REGARDLESS OF WHETHER THE OVERALL OPERATIONS OF THE FUND ARE PROFITABLE.
- THERE IS NO SECONDARY MARKET FOR FUND INTERESTS. TRANSFERS OF INTERESTS ARE SUBJECT TO LIMITATIONS. THE FUND'S MANAGER MAY DENY A REQUEST TO TRANSFER IF IT DETERMINES THAT THE TRANSFER MAY RESULT IN ADVERSE LEGAL OR TAX CONSEQUENCES FOR THE FUND.
KEEP IN MIND THAT THE PAST PERFORMANCE OF ANY INVESTMENT IS NOT NECESSARILY INDICATIVE OF FUTURE RESULTS. YOU SHOULD ONLY COMMIT RISK CAPITAL TO A FUND INVESTMENT. ALTERNATIVE INVESTMENT PRODUCTS, INCLUDING HEDGE FUNDS, ARE NOT FOR EVERYONE AND ENTAIL RISKS THAT ARE DIFFERENT FROM MORE TRADITIONAL INVESTMENTS. YOU SHOULD OBTAIN INVESTMENT AND TAX ADVICE FROM YOUR ADVISERS BEFORE DECIDING TO INVEST.
CAPITAL MANAGEMENT PARTNERS, INC. (CMP) HAS ENTERED INTO SELLING AGREEMENTS WITH SOME OF THE FUNDS DESCRIBED ON THIS WEBSITE, PURSUANT TO WHICH IT IS PAID FEES BY THE FUND OR ITS MANAGER IN CONNECTION WITH FUND SALES MADE BY CMP. IN ADDITION, CMP MAY ACT AS AN INTRODUCING BROKER FOR INDIVIDUALLY MANAGED FUTURES ACCOUNTS AND TO SOME OF THE FUNDS AND AS SUCH, MAY RECEIVE A PORTION OF THE COMMODITY BROKERAGE COMMISSIONS THEY PAY IN CONNECTION WITH THEIR FUTURES TRADING OR RECEIVE A PORTION OF THE INTEREST INCOME (IF ANY) EARNED ON AN ACCOUNT'S ASSETS. CERTAIN CTAS MAY ALSO PAY CMP A PORTION OF THE FEES THEY RECEIVE FROM ACCOUNTS INTRODUCED TO THEM BY CMP. BEFORE SEEKING ANY ADVISOR'S SERVICES OR MAKING AN INVESTMENT IN A FUND, INVESTORS MUST READ AND EXAMINE THOROUGHLY THE RESPECTIVE DISCLOSURE DOCUMENT OR OFFERING MEMORANDUM.
Information presented on this site has been obtained from sources that CMP believes to be reliable. However, CMP has not and cannot verify the accuracy of such information, and potential investors should be aware that such information is subject to change without notice. Not all products and services described on this website are available in all jurisdictions, and certain investments may not be suitable for all investors. Certain investment products that are available to U.S. persons have not been registered with any federal or state regulator, and as a result are available only to certain qualified investors. Qualifications vary from product to product.
Any information on this website referring specifically to investment products offered by CMP is only available to view by qualified persons with a username and password, which can be obtained by registering and qualifying. By registering you understand that CMP may, at its sole discretion, send you information (electronically or otherwise) on various products it offers that it believes are suitable for you and you expressly consent thereto. You may notify CMP in writing, at any time to remove your name from its product distribution list. The information on this website should not be construed as investment advice. You should obtain independent investment and tax advice before deciding to invest.
Past results are not indicative of future results. There is risk of loss when investing in managed futures or funds.
The information contained in this website may not be reproduced, reorganized or used to create a financial product including an index. This information is for informational purposes only and is provided on an "as is" basis. The user assumes the entire risk of its use. CMP expressly disclaims all warranties of originality, merchantability, of fitness for any particular purpose. Without limiting the foregoing, in no event will CMP or employees have any liability in connection with the use by any person of the information or data presented in this website.
CMP is a broker/dealer firm registered with the Securities Exchange Commission (SEC) and is a member of the Financial Industry Regulatory Authority (FINRA). CMP is also an introducing broker and a commodity trading advisor registered with the Commodity Futures Trading Commission (CFTC) and is a member of the National Futures Association and Managed Futures Association.
1100 North Fourth Street
Suite 141
Fairfield, Iowa 52556
Phone: 800-621-2520 / 641-472-8800
Fax: 641-472-7320
Contact CMP via emailHome | Investment Database | Funds Report | Hedge Fund | Fund Strategies | Fund Diversification | Futures and Hedge Funds | Futures Facts | Risk Management | Manage Selection | CTA Performance | Resources