Hedge Fund Replication: A Blended Approach

Combining single managers with replication products

Originally published in the April 2010 issue

It has become more widely accepted that a significant portion of hedge fund industry returns are generated through exposures to common factors including both beta factors (equities, bonds, etc) and alternative beta factors (spreads such as credit, default etc). A growing body of academic research has confirmed this, but it took the generally disappointing performance of hedge funds in 2008 for investors to understand that much of the returns they were paying high fees to obtain could be purchased directly at much lower cost. This, combined with concerns over the liquidity and transparency of traditional hedge fund investments, has fuelled the rapid development and growth of hedge fund index products, in particular, synthetic replication products sometimes referred to as hedge fund clones. These products are increasingly being seen as one possible method to “index”the hedge fund industry with low fees and high liquidity. It is, however, very difficult to accurately index an industry whose true composition is blurred by reporting biases, and where alpha still contributes to the benefits of hedge fund investing. Here, we describe the background to, and construction of, a unique approach to indexing hedge funds that combines a core holding of single managers and combines this with a secondary holding of a blend of synthetic replication products to reduce cost and increase liquidity. The result is a low cost and liquid method of capturing returns from the hedge fund industry while at the same time minimising the impact of a single manager event (blowup).


Sources of hedge fund returns
Academic research and empirical evidence indicate that hedge fund returns are broadly separable into four parts: traditional beta, arising directly from the returns of equities, bonds, commodities, etc; alternative beta from factors such as spreads, illiquidity premium, volatility and momentum; exotic beta, from investments such as life settlements, ground rents, shipping, carbon emissions, micro finance etc; and alpha which, in addition to the common definition of manager skill in market timing and security selection, also includes pure arbitrage, and structural alpha from activities like block trading, market making and other information arbitrage activities.

Alpha, once deemed to be the driving force behind the hedge fund industry, now appears to contribute the least to hedge fund returns, and is diminishing due to the increased number of hedge funds and investor capital entering the space. Professor Narayan Naik’s 2006 research is consistent with other findings: on average, 74% of fund of hedge fund returns could be explained by beta factors, with the returns of funds considered to be high-alpha suffering from the rapid expansion of their funds. Recent alpha levels are approximately 40% of pre-2000 levels. Arguably, the industry shakeout of the last 18 months will mean a reduction in risk capital by both hedge funds and banks and therefore the possibility for alpha to become a larger share of total return drivers but it is unlikely to reach levels that invalidate the existing research on hedge fund beta.

If a significant portion of hedge fund industry returns can be attributed to beta/alternative beta factors, then it follows that these can be replicated without the need for investments into hedge fund managers. Of course, not all hedge funds are created equal and not all can be divisible into simple beta factors. Equity long/short returns have proved to be the most easily replicated, with around 80% of returns explained by two alternative beta factors – the market excess return, as defined by the S&P 500 minus treasuries, and a small cap minus large cap spread defined by the Russell 2000 minus the S&P 500. Other strategies, with less market exposure or a more niche proposition, have seen less success with replication. Some very large hedge fund investors have been monitoring the performance of hedge fund clones with great interest and several have already invested sizable proportions of their hedge fund allocations in these products.

Mapping the Industry
Despite the amount of investor capital that has flowed into the hedge fund industry over the past decade, industry data remains notably opaque. With no requirement to report performance to an industry database, and with managers that report performance doing so selectively, there is no single reliable source of hedge fund data. Due to the self reporting nature of hedge fund indices, they suffer from multiple biases. These biases fall into several categories but the main ones are survivorship bias (only managers who survive report), self selection bias (managers self select themselves into the databases) and liquidation bias (managers tend not to report when they have a severe negative performance). Academic studies suggest the data in published single manager hedge fund indices is biased upwards by 3-5% per year and for fund of funds indices by approximately 1% per year.

Our own research suggests that while these may be good longer term estimates, the bias has been much greater than this during the crisis and recovery of 2008 – 2009. Of the largest 100 funds in the HFRI and BarclayHedge Databases in June 2008 (which can be shown to be a good proxy for the index as a whole), 24 were no longer reporting in September 2009. Frontier Capital has located the returns for 10 of these funds then recalculated the performance of the group on an equally weighted basis. Inclusion of these 10 non reporting funds resulted in the one year returns of dropping from -7.8% to -12.1% and returns for the first nine months of 2009 from 10.4% to 6.5%. This research is incomplete as we need to source returns for the other 14 non reporting funds but does suggest that the reporting biases over this period are greater than widely believed. We hope to publish further findings in the near future.

At Frontier Capital, rather than rely on industry data, we have developed our own database as a basis for the construction of a hedge fund index product. The initial universe combines six industry hedge fund databases creating a pool of 22,000 funds, including open and closed funds, multiple strategies, currencies and regions. The data is then cleansed to remove duplicates, obsolete funds and fund of funds. This database is then supplemented with our own research into “hedge fund giants” – the largest funds that do not report to databases. These funds represent an additional $250 billion of assets – approximately 15% to 20% of the total estimated hedge fund industry assets. This leaves a universe of around 3,000 funds and the full assets of these are used to determine the composition of the universe by style, strategy and other metrics.

The first of these drivers is hedge fund strategy. Beckers’ 2007 paper showed that of the 9.8% average fund of hedge fund returns, 9.3% was a result of style exposure, with the remaining 0.5% due to manager selection. Each fund within our database is classified into one of 13 strategies – based on a combination of definitions from industry databases as well as the experience of the investment team. To illustrate the impact of including non reporting funds in our database, multi-strategy funds comprise 11.7% of industry industry assets according to BarclayHedge, but represent a much larger 21.5% of industry assets when we include the assets invested with large managers who do not report to any of our six databases.

Proprietary stratified sampling
Each strategy is then further classified across three dimensions: geographical focus, life cycle and size. Research shows that funds which differ on these factors produce different risk/return profiles. This process produces a multi-dimensional map of the global hedge fund industry with over 200 individual “cells” that are populated with similar managers.

An investable universe is then created by applying minimum investment criteria, including a twelve month track record. The remaining funds are then subjected to a series of quantitative filters which removes outliers leaving a short list of funds that are representative of each specific “cell”. Multiple quantitative tests are used to screen each fund across five separate time periods and multiple factors such as beta, skewness, kurtosis and volatility. The aim of the process is to narrow the list to managers who are capturing the risk-adjusted return of the cell. Due diligence is then performed on each manager before they are admitted to a final buy list.

Mixing single hedge funds with synthetics
Although the resulting funds on the buy list are representative of the industry, building a fund from this universe will still suffer from low liquidity, high underlying fees and the potential exposure to event risk. Synthetic replication istherefore used for approximately 50% of the portfolio. The blend of both a direct investment and synthetic replication allows the fund to capture the advantages of hedge funds and minimise the negative aspects of hedge fund investing. While still a relatively new strategy, synthetic replication products are gaining in popularity as actual track records become lengthy enough to demonstrate that the theory behind replication can work in practice. The cost benefits are clear – a charge of around 0.90% vs. the traditional “2%/20%” for investors in single managers with an additional “1%/10%” layer for fund of fund investors, plus the advantage of daily or weekly liquidity and no lock ups. While correlations of many of the products to hedge fund indices are high, these can be deceiving, with significant underperformance from large tracking errors. Blending multiple clone products significantly minimises this risk – of the synthetic 21 products that we track, we have found that a selected blend of five achieves correlation in the 90% range and has outperformed the HFR Fund of Funds Index over the last three years. As these hedge fund clones only capture beta and alternative beta, some strategies are not possible to replicate. The factors traded by hedge fund clones will theoretically always lag the current industry factors due to the backward looking nature of the models.


Research and empirical evidence would suggest that synthetic replication may be best used to complement rather than replace direct hedge fund investments in portfolios.

Fund of funds have traditionally been the method for many investors to access diversified hedge fund returns as they usually demand lower investment minimums and offer greater diversification. The extra layer of fees is shown to be often unjustified, with the average fund of funds underperforming the average single hedge fund, even after taking fees into account. The style factor is likely to be at play here – there is no consistent best performing strategy and studies show that fund of funds are not successful at timing movements between strategies.

There is also significant blow up risk: our research suggests that a fund needs to hold at least 50 funds equally weighted to effectively diversify away single manager risk, although most fund of funds hold far less than this and risk can be concentrated in a single manager. Global Hedge holds investments in 70 managers, equally weighted to reduce blow up risk.


Diversified, liquid hedge funds exposure
The Frontier index product, Global Hedge, was designed as an improved method for Frontier Capital’s flagship multi asset fund to access hedge fund industry returns and we are so far unaware of any other fund using this approach. The product is currently used by a variety of investors and discretionary managers as a solution for their clients’ hedge fund allocation. The fund is a complementary product to very active fund of funds and should be seen as a highly diversified and very low risk approach to hedge fund investing.

As a one stop shop approach to hedge fund investing, it is hoped that the tracker will be seen as an attractive option for institutions demanding a sensibly diversified, liquid hedge fund exposure for a reasonably low cost.

Michael Azlen, CAIA is the founder and CEO of Frontier Capital Management LLP, an evidence-based investment management company based in London, running around $550 million across multi-asset and alternative investment products using advanced indexation strategies. The firm’s focus is on providing innovative, low cost, strategic and diversified investment solutions.