Lies, damn lies and statistics.
Some popular hedge fund strategies, widely touted as having low volatility and high Sharpe ratios, actually might be a disaster waiting to happen, according to new research by Andrew Weisman, chief investment officer of Nikko Securities Co. International, New York.
And the problem is exacerbated by traditional institutional investment tools, such as the mean-variance optimizer, that drive investors into higher allocations of strategies that might blow up in their faces. "In other words, you are building yourself a big bomb," Mr. Weisman said in an interview.
In a paper presented at a recent meeting of the Institute for Quantitative Research in Finance, known as the Q Group, Mr. Weisman explained how statistics can provide certain hedge fund strategies - including some market-neutral, merger arbitrage and fixed-income arbitrage techniques - the illusion of stellar risk-adjusted returns, but, in reality, hide some deep, dark truths.
The bottom line: when these strategies work well, they are great. But when markets go against them, investors stand to lose their shirts - as the Long-Term Capital Management LLC and other strategies that relied on credit spreads have in recent years.
Short histories problematic
A key issue, Mr. Weisman said in an interview, is that most hedge funds have relatively short histories and thus have not survived an entire business cycle. In fact, academic research shows hedge fund managers have a 20% annual attrition rate - four times the level experienced by mutual-fund managers. Looking at the TASS Database, published by a unit of Tremont Advisers Inc., Mr. Weisman found that 55% of managers have track records of less than five years, and 90% have less than 10 years.
With such short records, hedge fund Sharpe ratios - a measure of manager skill that divides a fund's return by its standard deviation - can give the appearance of superior risk-adjusted returns when that might not be the case.
This is precisely where institutional investors, their consultants and many hedge fund-of-funds managers get it wrong, he said. Most do not know how to evaluate hedge fund strategies and thus tend to rely upon statistical tools that give a totally inaccurate picture of the riskiness of a given manager.
"You have to ask yourself, if you're flying an airplane, you'd almost rather have no altimeter than a faulty altimeter," said Kevin Johnson, partner at Aronson + Partners, Philadelphia, which offers market-neutral strategies.
Mr. Weisman, who creates hedge funds and fund-of-funds strategies that are marketed in Japan and consults to American hedge fund-of-funds managers, hastens to add that not all hedge funds pursue "informationless" trading strategies. Of those that do, "I don't think they are engaged in a conscious desire to deceive investors." In reality, they likely are deceiving themselves, he said.
George Oberhofer, senior practice consultant at Frank Russell Co., Tacoma, Wash., agreed. "That's the interesting problem, when you hear the investment process and see the return patterns, it (their use of informationless strategies) doesn't jump up and scream at you. Andrew points out there's a lot of detective work to be done, but it's hard work."
Three types of informationless strategies - ones that do not depend on manager skill and are expected to yield no theoretical long-term return - are involved, Mr. Weisman wrote.
Short-volatility strategies, such as merger arbitrage and pairs-trading (which includes some market-neutral approaches), allow traders to collect premiums for taking the risk for an unlikely event. Investors are basically writing insurance policies against catastrophic events, thus generating steady income streams.
By setting up a series of "strangles" - writing puts and calls at 2.5 standard deviations from the prevailing market volatility - the manager easily can earn enough in premiums to double the risk-free return offered by Treasury bills, Mr. Weisman wrote. In any year, the manager has an 86% chance of doubling the risk-free rate, and a 46% chance of doing so over a five-year period, he found. And, with a less than 0.25% chance of the strategy going bad in any given month, the manager can continue collecting premiums for a long time before a crisis strikes.
But underwriting insurance essentially is a zero-sum game. At some point, a major loss will occur that will take back all of the premiums the manager has raked in. And assuming the manager does not realize the market is about to go against him and continues setting up strangles, that giveback will be huge, nearly one-third of invested capital, Mr. Weisman wrote.
But investors fail to recognize these hazards. "Not many people are aware" of the risk, said Mark Anson, senior investment officer, global equities, at the $160 billion California Public Employees' Retirement System, Sacramento, which is expected to propose a strategic partner to invest $1 billion in hedge-fund strategies in May.
In fact, as long as market volatility does not exceed the collar during the period in which performance is measured - such as in 1987, 1990, 1994 and 1998 - the manager's returns will look positive regardless of whether the market goes up or down, its Sharpe ratio will look stunning and regression analysis will show the manager is highly uncorrelated with other asset classes.
Using these risk-adjusted returns, the mean-variance optimizer could serve to "maximize a future disaster," Mr. Weisman said.
Buying `a cat in the bag'
Illiquid securities investing creates a similar statistical nightmare. Managers who invest in illiquid securities - for example, buying a subordinated tranche of a small mortgage-backed security issue - tend to systematically underreport both ups and downs in the portfolio's valuation. The managers tend to go part-way in their valuations - as do managers of real estate and private equities - because of the difficulties in getting current valuations.
Fudging valuations smoothes out volatility, which in turn dramatically improves the fund's Sharpe ratio. Hedge fund managers that invest in mortgage securities, Mr. Weisman said, typically reflect 15% to 20% of true volatility, which overstates their Sharpe ratios "by a factor of five."
Again, the traditional mean-variance optimizer will cause pension funds and hedge funds-of-funds to load up on these strategies. "You will be maximizing for illiquidity in the portfolio," he said. "You will be buying a cat in a bag - you think you will own something when you own something entirely different."
Mr. Weisman said one can predict how long it will take for a hedge fund strategy to blow up, which usually occurs when its prime broker becomes concerned about a possible spread between a fund's reported and actual net asset value.
For example, his "AT Deathometer 9000" predicts it will take 49 months for disaster to strike at a fund whose valuations place the volatility level at 15% when the true measure is twice that level. In fact, major volatility events tend to occur once every four years.
The St. Petersburg Paradox
Finally, Mr. Weisman examines the St. Petersburg Paradox, an informationless trading strategy under which a hedge fund manager risks 0.5% of capital on a strategy with only two possible outcomes. If he loses, he doubles his bet the following week, and continues doubling up until he succeeds. After that, he reverts to his initial 0.5% bet.
Investors don't see this activity, since hedge funds generally report returns only at the end of the month. Using a Monte Carlo simulation, Mr. Weisman found this strategy works well for 400 weeks, earning about 15% annual return with a 12% standard deviation. But when the good times end, the investor can expect to lose half of his capital.