Separate studies conduct returns-based analyses of Renaissance and Amaranth

Michael Markov, CEO of Markov Processes International says that the Law of Large Numbers was the “last great gift of the Renaissance”.   In a twist of irony, says Markov, the Law of Large Numbers also explains why “a simple combination of factors can mimic the performance of a large and well-known hedge fund”. That fund?  Renaissance Technology’s $25 billion “Renaissance Institutional Equities Fund” (RIEF).

Markov studied RIEF’s August performance to understand the types of factor exposures undertaken by RIEF.  He found that the need to liquidate positions at the worst possible time (an oft-cited reason for August’s mayhem) “may only be a part of the story.”

But how useful is the examination of factor exposures when a fund strays unexpectedly from its typical exposures?  In a separate study, Bhaswar Gupta and Hussein Kazemi fund out what, if any, information might have been gleaned from the historical return stream of Amaranth in order to predict that fund’s eventual demise.

Renaissance: Over-diversified?

Renaissance’s RIEF fund was undoubtedly caught up in the August run for the exit.  But “the rest of the story”, according to Markov, was uncovered using his firm’s proprietary returns-based factor model.  In a nutshell, he plugged RIEF’s returns into his model “in an attempt to see if some of the losses could (or should) have been anticipated.”

You may recall a similar analysis conducted by Professor Ross Miller of the State University of New York at Albany on another monolithic fund, Fidelity Magellan (see posting “Magellan a Frankenfund: Professor“).

Markov’s analysis of returns since the fund’s inception in June 2005 reveals exposures to several factors including: EAFE, large cap growth, large cap value, and small cap value.  In addition, he finds a rather large short position in mid-cap growth stocks.  (Note that net exposure is consistently 100% due to the fund’s mandate.)

According to Markov, these weights yield the following cumulative return stream…

Markov describes this result as, “pretty remarkable, especially as the factor exposures haven’t changed at all over the two-year period.” He continues, “This adds a lot of credibility to the analysis, which otherwise could be considered as a fitting exercise.”

With R-squared in the neighbourhood of 0.7 to 0.8, the fit is pretty tight.  In fact, Markov says that such values are “more common to the analysis of diversified long-only mutual funds”.

Acknowledging that this tight fit was a result of “in sample” tests (where one would expect a tighter fit), Markov also conducts an “out of sample test”.   Here’s how the replica performed against the real thing over the summer:

As you can see, the replica performed admirably over the summer.  But when Markov runs the model back to 1987, he finds that the replica portfolio also underperforms in the recession of the early 1990’s and in the tech bubble of 1999-2000.

Markov concludes that:

“Proper hedge fund due diligence should go beyond ratios and drawdown statistics which have little predictive power. At the same time, if estimated accurately, factor and/or index exposures of a fund could provide sufficient guidance of what to expect from the strategy in various future market environments. When it comes to the replication of hedge funds, dynamic multi-factor analysis of hedge fund returns provides both the means of replication and sufficient information to decide whether a given strategy should be a replication target in the first place.”

Amaranth: Careful analysis revealed certain risks beforehand

Bhaswar and Kazemi find that Amaranth’s factor exposures (to things like high yield, commodities, Fama/French factors etc.) went off-piste around May 2006, several months before the fund blew up.   

The duo concludes that:

“…knowledge of the Amaranth’s returns did not enable investors to assess the fund’s unique risks correctly.  (However) careful quantitative due diligence would have revealed some interesting facts about the Amaranth portfolio…(There were) structural breaks in its return patterns in mid 2005 as well as the stellar returns generated in late 2005 and early 2006 from a clear change in strategy.”

The Bottom Line

After bending and stretching the numbers every which way, both studies eventually seem to reach the same conclusion – that their results must be taken with a grain of salt since returns-based analysis is only one way of looking at a hedge fund.  Bhaswar and Kazemi echo Markov’s view that proper due diligence must go “beyond ratios” when they conclude:

“…risk-return measures that are solely based on historical returns very often fail to provide the tools that investors need to protect themselves against serious losses.”

Be Sociable, Share!

Leave A Reply

← New paper claims humans required to manage hedge funds Assessing Hedge Fund Performance: Does the Choice of Measures Matter? →