HF-Analytics is highly unusual as a consultant in that its fees depend partly on how far its selected funds beat a benchmark. Schwindler’s career has straddled academia and active investment. He has conducted due diligence on hedge funds for Feri Institutional Advisors, taught on the topic for Deutsche Boerse, worked in academic research at the highly rated Edhec Risk Institute in Nice, and also taught at Bamberg University in Germany.
Q: What is genesis of the HF-Analytics proposition?
A: We follow a rigorous quantitative approach. A big opportunity exists for quantitative analysis, which is often under-rated. A good quantitative screening methodology, we think, allows you to avoid wasting time doing operational due diligence on funds likely to underperform. The methodology has been developed over 5 years and was the topic my doctorate from the University of Bamberg. The PhD thesis was entitled “Value Added By Fund of Fund Hedge Fund Managers”. It concluded, at that point in time, that 60% of value added came from strategy selection, 40% from manager selection – with nothing from tactical strategy allocations. At HF-Analytics we define all of these sources of value in a unique way that differs from how the terms are often used in other performance attribution frameworks. My doctorate also discovered that the volatility of value added from manager selection was far greater than that added from strategy selection. The system uses Matlab, but there are no plans to sell it as free standing software.
Q: What is unique about your business model?
Fees for our quantitative analysis of funds of funds are partly based on the extent to which recommended funds outperform a benchmark agreed with the client. Therefore an element of performance related remuneration is introduced as a success oriented fee into a consultancy contract. The benchmark would probably be the HFRI Fund of Funds Index. It could also be the Edhec or Eureka funds of funds indices, or other agreed with the client. Full due diligence, including qualitative and operational, requires an up front fee, to cover site visits and background checks (see Fig.1).
THE FUND OF FUNDS INDUSTRY
Q: Which areas of the fund of funds universe are currently ranked?
A: There are around 3,500 funds of funds out there, including some double counting of multiple currency share classes. Multi strategy funds are most suited to our analytical techniques. Analytically, HF-Analytics is currently sticking to USD classes to avoid the complications of dealing with currency hedging issues; further research may add those.
Similarly the analysis is presently restricted to unleveraged share classes. There is no minimum size for funds, but they do need a real money track record of at least 36 months. Analytically there is no difference between offshore funds of funds and those run on a managed account platform, and the database does also include some onshore funds of funds. After you also exclude the large number of non-reporting funds our universe shrinks to around 400 (see Fig.2).
Q: Who are the target users for the research?
A: The main focus for our offering is the many family offices that do not have enough staff to select single funds. Larger pension funds are generally doing this in house, although smaller ones will benefit from assistance in selecting funds of funds. Private wealth departments in banks are also important buyers of funds of funds. Insurance companies over time will become an important market.
Q: What is your prognosis for the fund of funds business model?
A: We are seeing some mergers of funds of funds in the US. It is not clear what clients will pay, and whether the 1 and 10 business model will be the right one going forward. Some of the larger funds of funds are also starting to offer consulting services on much lower fees. If fees continue to come down more groups will merge. The best funds of funds can of course justify fees of 1 and 10 but many others cannot. A more diverse range of business models and fee structures will emerge.
Q: So will funds of funds be replaced by synthetic products?
A: Synthetic products necessarily have to go for very broadly defined risk parameters, and a single manager can capture more granular risk premiums. The current products available are not, in the opinion of HF-Analytics, coming close to hedge fund returns, even if they are more liquid. A replication product is a good cash substitute for funds of hedge funds that allows minimisation of tracking error when they have idle cash.
The crucial weakness of synthetic products, in my view, is that they do not effectively capture illiquidity premiums. For instance, the convertible space has a big chunk of illiquidity premium. Historically when hedge funds held 80% of the market it was vulnerable. Now that more is owned by long only managers the market has stabilised. The distressed debt market also has big liquidity premiums. And managers who focus on small and mid cap equities also can pick up liquidity premiums. There is always a trade off between harvesting the liquidity premium and not being able to pay redemptions. So the qualitative diligence always seeks information on liquidity of underlying funds, and also on the underlying funds’ position liquidity profile.
Q: So do funds of funds need position level transparency?
A: Many hedge funds, in my experience, are not prepared to provide this. In any case, having line items for single hedge funds is nearly worthless, as it is far too much information for the individual investor to manage and interpret. Few investors have the resources to clean and analyse such detailed data. Risk aggregation is more useful, as a risk profile can be used in a portfolio context.
Q: What packages can be used for risk aggregation?
A: At HF-Analytics we have looked at the various risk software packages before. The company likes Riskmetrics and Pertrac’s Finanalytica, but would not judge a manager on which software they used. Appealing features of Finanalytica include its expected tail loss tests, and modelling with non-linear copulas. The tail loss is clearly important in going beyond what Value at Risk would reveal. And because hedge funds have non-linear return profiles it also makes sense to model with non-linear copulas. Risk Metrics is especially useful for position transparency as many single funds report line items to them, allowing more precise aggregations of risk. Some packages do not have access to line items.
Q: What are current performance attribution trends?
A: We intend to publish a report on attribution between strategy allocations, tactical allocations and manager selection. Just recently manager selection returns have dipped into negative territory.
However, the single hedge fund indices’ performance is overstated, leading to a measurement error. Single funds are only submitting liquid data to the indices and with-holding side pocket data.
Q: You previously wrote about the Sharpe ratio. Do you use it to select funds?
A: The Sharpe ratio in itself does not form any part of our process, as it does not have much predictive power. Ranking funds by HF-Analytics risk adjusted value added measure massively outperforms either the original, or autocorrelation-adjusted sharpe ratios.
Q: What are the components of the Risk Adjusted Valued Added measure?
A: The process is completely quantitative, objective and optimization free. Funds of fund managers are not able to designate allocations as strategic or tactical. Value Added from Strategic Asset Allocation simply compares the fund of funds’ strategic benchmark with that of a fund of funds index. To gauge Value Added from Manager Selection, the benchmark becomes endogenous: it is determined by the manager’s strategy allocations. The question is whether his manager selection choices outperformed allocating his strategy weights to the relevant strategy index. To measure Value Added from Tactical Asset Allocation, both endogenous benchmarks are used and any deviations from the strategic strategy allocation are deemed to be tactical. The model does not distinguish between active decisions to change strategy weights, and those that are passive consequences of performance differences between strategies. Any tactical overlays operated on top of a fund of funds are treated as if they were an externally fund and influence the value added from the manager selection (see Fig.3).
Q: Which hedge fund strategy categories are best for analytical purposes?
A: Our model uses five strategies: equity hedge, event driven, relative value, tactical trading, and emerging market. If a fund of fundsmanager allocates to any strategies outside these umbrellas, that will affect the value added from manager selection, positively or negatively. The strategy categories are broad to produce more robust results. More granular strategy categories may be superficially appealing, but they run into statistical problems that render the results very unstable and prone to unrealistically wild gyrations. A simple regression model produces a better coefficient, with more stable factor loadings with 5 rather than 10 or 15 strategies. HFR indices were chosen based on a robustness check, but mapping the more numerous Credit Suisse indices onto the HFR classification methodology to produce five analogous indices generated very similar results.
Q: Which hedge fund indices are best to use?
A: A study in 2004 showed that the EDHEC indices have an edge in terms of being more representative and are less likely to misclassify funds. However, since then the differences between indices have declined. There has been a convergence between them as more and more managers are reporting simultaneously to different databases, improving their coverage. Investable indices have been rejected mainly due to their short track record. But a test run showed that the Value Added components would change, while this does not alter rankings based on the value added analysis. We use a survivorship bias free database for our rankings, but nowadays liquidation bias is far more important as side pockets mostly do not appear in single strategy indices, and funds do not report for their final few months of life. It is difficult to quantify the liquidation bias, but a fund of funds index automatically includes it since funds of funds own the side pockets and indirectly report side pocket performance. They also own litigation claims for Amaranth and MotherRock. This is a much more realistic way to look at performance so funds of funds indices are much more meaningful.
Q: What types of checks will your operational due diligence process entail?
A: Asset verification is one important element. This requires talking to the administrator, and fund of funds manager, to get information, and cross checks that the NAV is the real NAV of the single funds for example. But it also includes audit checks, background checks as well as reading the official documents of the fund. During onsite visits we do not just sit in the conference rooms and talk to people but rather have a look at the operating systems and talk to people at their working desk. These impressions sometimes reveal very informative pieces of information.
Q: What are some of the key factors your qualitative process will look at?
A: Basically, we analyse all the main steps in the fund’s investment process. We consider the breadth of industry networks, and access to closed funds. We focus on the managers’ investment process – quantitative and qualitative due diligence – and monitoring activities after an investment is made, and especially look at the structure of the investment team and its responsibility structures.
For example a really independent risk manager with a veto on investment decision is a big plus compared to a CIO who wears different hats at the same time. Furthermore, the interaction between the portfolio manager and the fund analysts give some really good insights into how the responsibilities are spread between different parts of the team.
Capital Preservation rating – This is comparable to the one from Lipper. The cumulative loss over last 3 years adds up negative monthly numbers. This is designed to complement the draw down measure.
Endogenous Benchmarks – the benchmark is determined by the manager’s strategy allocations, which are in turn estimated with an econometric model.
Exogenous Benchmarks – benchmark from an external source.
Liquidation Bias – by not reporting side pocket performance, and not reporting returns after they liquidate, hedge fund index performance may be inflated.
Matlab – powerful and flexible mathematical programming environment.
Monte Carlo Simulations – using thousands of random simulations to generate a wider spectrum of possible outcomes than limited historical data will generally provide.
Non-Response Bias – the bias to indices that can arise from some funds not reporting returns
Over-fitting – using too many explanatory variables to get a better fit that will turn out to be less stable and less reliable over time.
Optimisation free – a model that is parsimonious, and avoids the problems of over-fitting
Parsimonious – model with fewer explanatory variables and more stable over time
Principal component analysis – a form of multiple regression designed to work out what proportion of a return profile can be replicated by varying combinations of specified factors, which in this case are the hedge fund strategy indices.
Strategic Benchmark – an assumption of constant strategy allocations is used to measure Value Added from Strategic and Tactical Asset Allocation.
Success Oriented Fees – payment relative to the performance of analytical services rather than on a traditional fixed fee basis.
Survivorship Bias – Blow ups not only drop out of the index, but also sometimes get retrospectively removed. This inflates index returns.
Tactical Benchmark – an econometric model estimates strategy allocations from the five hedge fund strategy indices. The tactical benchmark is not optimised combination, but is optimised to replicate the return of the fund. A multiple regression process infers what combinations of strategy allocations could have generated the return profile.
Value at Market Risk – the part of Value at Risk that is due to the volatility of the hedge fund strategy indices. This is the risk counterpart to the Value Added from Strategic Asset Allocation.
Value at Specific Risk – the part of Value at Risk that is due to the volatility of the hedge fund manager selection choices. This is inferred from the part of the return series than cannot be explained by the five hedge fund strategy indices.
Value Added – any improvement returns relative to a benchmark.
Absolute Value Added from Manager Selection – value added from manager selection, relative to the peer group of other funds of funds managers. Historically this would have been ideal for investors aiming to maximise returns without worrying about volatility.
Consistent Value Added from Manager Selection – risk adjusted value added from manager selection, relative to the peer group of other funds of funds managers. Historically this would have been ideal for investors aiming to optimise risk adjusted returns.
HF-Analytics has developed a proprietary quantitative, optimization free methodology to analyze the value added of fund of hedge fund managers. Research clearly shows that manager selection is the most important skill of a successful fund of hedge fund manager. Based on these findings we offer investors detailed peer group reports on individual fund of hedge funds analyzed with our proprietary value added model. These reports are ideal for ongoing performance monitoring as well as making investment decision from a short list. In addition we also provide hand collected short and long lists selected with our value added approach for fund of hedge fund manager searches. Moreover, HF-Analytics offers complete manager searches (incl. qualitative due diligence) at competitive fees, as most of our honorarium is based on a success oriented fee. HF-Analytics’ consulting services are complete independent and without any conflicts of interest, as our sole business is servicing fund of hedge fund investors in their quest for the best fund of hedge fund managers. Furthermore, HF-Analytics offers customized quantitative hedge fund research for clients in the hedge fund and asset management industry and supports clients in-house by their quantitative research projects.
HF-Analytics is going to launch a concentrated fund of hedge funds index, for which the constituents will be selected with its proprietary value added approach, in the second half of 2010. The index provides fund of hedge fund investors a realistic benchmark of top performing funds of hedge funds.