Hypothetical Backtested Performance

Yesterday’s post on faking returns made me think about the use of theoretical models and the ability of an investment adviser or fund manager to use hypothetical performance instead of actual performance. The real use of performance figures is in advertising, so the SEC’s rules on advertising are the key focus. (I don’t see how hypothetical performance works on reports to investors, unless you’re Bernie Madoff.)

Backtesting involves the use of theoretical performance developed by applying a particular investment strategy to historical financial data. You’re more likely to see it for a quantitative or formula-based strategy than anything else. The backtested results show investment decisions that theoretically would have been made had the given strategy been employed during the particular past period of time. However, backtesting does not involve actual market risk or client money.

The SEC rules do not explicitly address model performance. You would have to look at IA Rule 206(4)-1 (a)(5) which prohibits any advertisement that “contains any untrue statement of a material fact, or which is otherwise false or misleading.” Backtesting is going to start from a position as being highly suspect since it’s not based on actual events.

The adviser will need to disclose that there are inherent limitations on the data derived from the retroactive application of a model developed with the benefit of hindsight. The adviser needs to disclose the reasons why actual results may differ. See In re Market Timing Systems, Inc. et al., SEC Release No. IA-2047

One of the problems with backtesting is whether the securities and trades that would be used going forward were available in past. This is a particular problem when using synthetic products or derivatives. Of course, the advertised performance must reflect the deduction of advisory fees, brokerage or other commissions, and any other expenses that a client would have paid.

The most obvious need in using a theoretical model is that adviser needs to disclose that the performance was derived from the retroactive application of a model developed with the benefit of hindsight and not with real money at stake. See In re Schield Management Company et al., SEC Release No. IA-1872

The SEC has indicated that labeling backtested returns as “hypothetical”, “pro-forma”, or that “actual results were available upon request” in and of itself, is insufficient to satisfy the disclosure requirement. (see In re Schield Management Company et al., SEC Release No. IA-1872  and In re LBS Capital Management, Inc., SEC Release No. IA-1644) It  fails “to convey fully the inherent limitations of the data derived from the retroactive application of a model developed with the benefit of hindsight.” The disclosures need to be robust enough to dispel the misleading suggestion that the advertised performance represented actual trading.

There are no set rules, so you need to look toward enforcement actions to see what actions the SEC found to be egregious.

In re Patricia Owen-Michel, SEC Release No. IA-1584 (Sept. 27, 1996)

In the enforcement case against Patricia Owen-Michael, the SEC sanctioned the president of an investment adviser for allegedly circulating misleading advertisements that used a computer-based statistical model to select stocks and mutual funds and to generate trading signals. The adviser’s advertisements included charts and graphs depicting hypothetical performance of an investment model applied retroactively. The SEC alleged that the various charts and graphs depicting hypothetical performance of the model failed to disclose:

  • That the adviser only began offering the given service after the performance period depicted by the advertisement;
  • That the advertised performance results do not represent the results of actual trading but were achieved by means of the retroactive application of a model designed with the benefit of hindsight;
  • All material economic and market factors that might have had an impact on the adviser’s decision making when using the model to manage actual client accounts;

In re Schield Management Company et al., SEC Release No. IA-1872 (May 31, 2000)

In the 2000 case against Schield Management, the SEC alleged that the firm distributed materially false and misleading advertisements because it combined the pre-implementation data with performance data from periods following Schield’s implementation of the relevant trading strategies. One chart showed that the Schield’s model consistently outperformed the S&P 500 index without disclosing that Schield’s actual implementation of the strategy actually underperformed the S&P 500 index. The advertisements also failed to disclose that it applied materially different trading rules in calculating the performance of the strategy before and after the actual implementation of the strategy.

According to the SEC, Schield published advertisements that materially overstated their performance because they failed to deduct the full management fee and other fees earned by the firm from the performance results. On a cumulative basis, this had the effect of overstating the performance of the strategy by more than thirteen percent. The firm also included performance numbers that were calculated erroneously.

In re LBS Capital Management, Inc., SEC Release No. IA-1644 (July 18, 1997)

In the case against LBS Capital Management, Inc., the SEC sanctioned an investment adviser who had developed a mutual fund timing and selection service by using historical financial data, but failed to “disclose with sufficeint prominence or detail that the advertised results … did not represent the results of actual trading using client assets”.

The advertisement disclosed in a footnote that the performance results were “pro-forma,” that “model” performance was “no guarantee of future results,” that the timing service “ went live” in January 1994, and that “actual results” were “available upon request.”

The SEC found that the footnote disclosure was inadequate under the  facts and circumstances.  Citing In the Matter of Jesse Rosenbaum (IA Release No. 913, May 17, 1984), the SEC pounded on the table and stated that a misleading statement in an advertisement cannot be “cured by the disclaimers buried in the [smaller print] text [of the advertisement].”

The SEC also noted that the advertisement was distributed to the adviser’s existing and prospective retail clients “without regard for their investment sophistication or acumen.” Using the facts and circumstances test, the SEC used the standard of an unskilled and unsophisticated investor.

In re Market Timing Systems, Inc. et al., SEC Release No. IA-2047 (Aug. 28, 2002)

In the case against Market Timing Systems, the SEC alleged that Market Timing Systems, Inc. promoted returns for its model of over 70% for a 13 year time period. However, the advertisements did not disclose that the performance results were hypothetical and were generated by the retroactive application of the mode. The advertisements with 13 years of performance were distributed in 1999 and Market Timing did not begin business until 1998.

One point in this case clarifies the problem with using hypothetical models. The actual performance of client accounts during its first quarter of operations was materially less than the model’s hypothetical results for the same period.