Portfolio Construction: A question of skill
The hedge fund world is supposed to be about skill. But Martin Steward asks, if you use a fund of funds, what skills are you buying - and whose?
It is one of the ‘old chestnuts' of the hedge funds world: What is the key driver of risk at fund of funds level - manager selection or strategy selection? It is a question that gets to the heart of the hedge fund debate: is this really all about idiosyncratic skill, or rather shared systematic exposures? And our answer to that question will determine the kind of portfolio-construction skills we think we are paying our fund of fund manager 1-and-10 for.
Hedge fund investing ought mainly to be about idiosyncratic skill, and there is plenty of academic research confirming that assumption. Comparing long-only large-cap managers with high-yield bond managers over a full cycle reveals huge dispersion of returns; compare convertible arbitrage managers with managed futures over the same cycle and the dispersion is perhaps half as wide.
By contrast, while the best and worst long-only large-cap managers compete over basis points, for the best and worst managed futures programmes it can be tens of percentage points. That leads to results like those demonstrated in a 2007 paper by Girish Reddy, Peter Brady and Kartik Patel of Prisma Capital Partners: a fund of funds investing in the median managers across the full range of strategies would have improved its annualised returns by an astonishing 7% just by moving up to the 30th percentile. Easier said than done, of course - but the results at least confirm the benefits of trying.
Hedge fund management is about identifying a specific risk-taking skill and then hedging out all the other risks. But we know that it is not possible to hedge out all those other systematic risks; another stream of academic research has dedicated itself to identifying them in hedge fund returns. Reddy and colleagues found the returns of different strategies to be similar "over multi-year periods", but that smooths out the cyclicality we would expect managers and, at an aggregate level, strategies, to have if their returns were driven by systematic risks.
This is the conclusion suggested by the table from hedge fund index provider HFR (figure.1), and it seems to be confirmed empirically by funds of funds, like Liongate Capital Management, that do take an active top-down approach to constructing portfolios. "We believe we can add additional value of just over 500bps per year onto manager selection with a very flexible approach to portfolio re-allocation with a top-down view," claims head of research and quantitative risk management Ben Funk. "The traditional approach would be to have a strategy allocation that reflects the industry and then select the best managers you can find in each strategy. We think that's too backward-looking, and a more forward-looking approach is particularly suited to the current environment."
It is important to note that this is a top-down view, not necessarily a strategy view. "You simply can't say: ‘I want to reduce credit so I'm going to reduce distressed managers', or ‘I want less equity so I'm going to reduce my equity long/short managers'," says Funk. Why? Because many systematic risks are shared across strategies - long/short equity is likely to have a small-caps bias correlated with credit strategists, for example. Again, these risks are not accounted for by standalone single-manager risk measurements like Sharpe, Sortino, Calmar or omega ratios.
At Culross Global Management a long pedigree in the fixed income world inculcates a strong top-down culture. "We start with a medium-term macro picture to give us a sense of where the most interesting areas to take risk are, and one of the justifications is that it enables us to consider risk at the portfolio level in a way that is distinct and in addition to disciplines applied at manager level," says partner Christopher Keen. But he also says it "liberalises" the firm from the traditional hedge fund strategy definitions, which are ill-suited to describing the idiosyncrasies of hedge funds. One current macro call is an expectation of future inflation: Culross is implementing that view with a global macro relative-value bonds manager but also, less obviously, with a long/short equity manager that uses pricing power as one of its stockpicking screens.
In other words, Culross picks strategies that agree with its top-down views, but also individual managers that exhibit skills that agree with its top-down views. Variations on that theme are articulated by most managers, and so it should come as no surprise that performance attribution models like those designed by Gary Brinson often find an ‘interaction' factor dominating when applied to funds of funds.
It also points us to a profound truth about hedge fund strategies: while most exhibit a degree of systematic risk and, therefore, cyclicality, some strategies exhibit more cyclicality than others; their systematic exposures can be first, second or third-order, each implying a very different pattern of cyclicality; and even different manager styles within the same strategy exhibit different patterns of cyclicality. "Alpha can be about timing into systematic exposures, it can be trading skill, it can be security selection, it can even be dealing with complex situations," notes Erik Bernhardt, an associate director with PAAMCO.
Take long/short equity, for example. The most common style claims security selection as its skill and is usually net-long, perhaps more than 100% in a bull market and 30% in a bear market. A fund of funds might not want that through the bear market, regardless of how highly it rates those stockpicking skills. By contrast, a manager swinging from 200% net-long to 100% net-short is clearly selling a very different skill: if you rate this manager's ability to make those macro calls, why not hold them through the entire cycle? On top of those distinctions, long/short equity managers will express the usual systematic biases - value, growth, size, quality, region, and so on - amplified by their unconstrained approach.
Global macro is a similarly broad strategy. Some will offer a diverse range of skills in a range of markets - discretionary, systematic, relative value, directional, fixed income, equity, FIX, developed or developing markets, high and low-frequency trading, and so on - and offer meaningful exposure to them all throughout the cycle.
Others will make a top-down decision about which styles to over or under-weight; still others will recognise real skill in only one or two markets or styles. A manager who is a whiz at G3 yield curves was great in your portfolio in 2007-2008, but will probably trundle along doing nothing until those curves start to flatten again; right now, it is emerging market currencies and curves you want your macro funds to be trading. But is that a decision for the fund of funds or for the macro manager?
With some strategies, the answer to that question seems much more obvious because their constituent managers are more obviously exposed to systematic risks. No-one ever made any money doing merger arbitrage when no companies were thinking about mergers, and conversely, distressed debt is an uphill battle in a credit boom. These strategies amplify the problem faced by our imagined global macro G3 curve specialist: performance is determined by the size of the opportunity set. The cyclicality of these strategies is clearly linked to the credit cycle, and this is also true even of strategies whose normal range of distributions show no beta to systematic credit risk, but have second or third-order systematic exposures: an arbitrageur targeting a set level of volatility of returns from trading short-term mean reversion in credit spreads will achieve that with growing leverage as those spreads gradually become narrower through the cycle. Where merger arbitrage and distressed will just sit doing nothing in non-conducive environments, this kind of strategy is more likely to blow up in your face.
This is not to say that managers in these highly cyclical strategies do not deploy skill, of course. "There are two separate questions here," says Julian Shaw, head of risk management and quantitative research at Permal. "First, is it the right time to be in a strategy? Second, even accepting that a strategy might be mostly made up from systematic risk, are there a range of alphas or are they all roughly the same?"
Merger arbitrage is so systematic that it is replicated pretty effectively by indices, but there is also no doubt that some managers are better at it than others. Hyper-cyclical distressed debt is very obviously a bottom-up security-selection game requiring very specific balance-sheet analysis and legal knowledge. And manager selection even pays off in the world of the short-vol spread arbitrageurs: if your manager is skilled and wise enough to reduce leverage as spreads tighten, and if he can keep his credit lines in place until the cycle comes back to favour him again, he will preserve the gains he made for you in the boom times. "Spread strategies are about extracting a systematic premium from the market, but they are vulnerable to crowding and deleveraging," says Christopher Fawcett, senior partner at Fauchier Partners. "Performance from those managers is driven more by risk management than by the nature of the positions they have got on - but that remains a skill."
However, these skills - two bottom-up, one top-down - are about extracting the highest information ratio from an underlying systematic risk. They arguably remain fundamentally either risk-on or risk-off, so while they can manage cyclicality they cannot capitalise on it. There would appear to be little advantage in the fund of funds delegating any top-down decision making to these strategists - as opposed to long/short equity managers with highly-variable net exposures or multi-strategy macro managers.
A few conclusions suggest themselves from these observations. While dispersion of performance between managers in the more cyclical strategists might be quite high, that dispersion will be greatest during the periods when their underlying systematic risk is delivering return, and as a result, over the entire cycle the return to the fund of funds is likely to be dominated by the top-down decision of whether to be allocated to the systematic risk or not. Furthermore, it also suggests that the contribution of manager and strategy selection to the risk and return of the diversified fund of funds will itself be cyclical.
"There are points in the cycle where manager selection is more important and others when strategy selection is more important, and when our portfolio will see bigger strategy moves as a result," says Harry Wulfsohn a director at Stenham Advisors. Brian Chung, a senior portfolio manager at SSARIS Advisors, a fund of funds affiliate of SSGA, agrees. SSARIS uses its own and its managers' macroeconomic views to identify patterns and inflexion points in markets that it can act upon. "When markets are functioning normally, manager selection is very important," he explains. "Finding the best convertible arbitrageur will serve you well. When the market is at inflexion points, on the other hand, being tactical has served us exceptionally well. Over the last three years possibly 80% of our returns have come from being tactical, whereas under more benign conditions 80% should come from manager selection."
It would be easy to conclude that the top-down calls made at inflexion points should drown out the relatively incremental benefits of manager selection that accrue through benign market conditions. Does that answer the fund of fund buyer's basic question: ‘Am I essentially buying top-down or bottom-up expertise?' Well, it certainly helps.
It is possible to argue, as the more ideological bottom-up managers do, that fund of funds investing is about selecting the best managers to whom one should delegate the top-down macro calls. º"If you run a hedge fund with a variable bias, I don't care how good your stock selection skills are - if you're 80% long when the market tanks, you're going to lose money," says Rory Hills, founding partner of Hilltop Fund Management. "For a fund of funds it's completely the opposite way around. We just don't have the liquidity to make big macro calls."
Hills describes his managers as generating three types of alpha: pure alpha, "discrete from general markets, so neither of us makes a macro call"; macro alpha, tactically going long or short systematic risks or timing markets; and opportunistic alpha, such as that derived from distressed debt. As he concedes, the last group is the only one that requires him to make the macro call himself - and it is a "little group" in his portfolio included only in cases of extreme asymmetric opportunity.
At PAAMCO, although Bernhardt sits on the strategy allocation committee, he is quick to point out that it is focused on "very large trends" and that the firm has neither the liquidity nor the inclination to be a macro trader. "Two-thirds or three-quarters of the changes to our underlying exposure levels will be done at manager level," he reckons.
However, it is notable that Liongate likes to be able to liquidate at least 25% of its portfolio within 30 days so that it can be nimble enough to make macro calls itself. "That means daily or weekly liquid strategies, and in some cases we have separate managed accounts or liquidity on managed account platforms," says Funk.
This is important, because it hints at how a fund of fund's self-perception as either top-down or bottom-up can affect its manager selection to the extent of changing the portfolio-level risk profile.
"I try not to categorise managers by strategy at all, because I think it leads to a strategy-beta mindset," says Hills. "The investment committee that tries to predict which strategies will perform best is inclined to buy a basket of funds that reflects the macro call they've just made: the last thing you want to do is go with one or two managers who are very idiosyncratic. I don't think many people are good at making macro calls, and if you are good at it you should be running a hedge fund, not a fund of funds."
Bernard Minsky, head of portfolio anaylsis and risk management at IAM, where the focus is also bottom-up, says something similar: "We've found that if your philosophy is to decide how you're going to allocate to strategies and then look to fulfill your allocation, the result is dominated by systematic asset allocation. If you work more bottom-up, followed by some adjustments to fit with certain strategy exposure limits, attribution analysis appears to favour manager selection."
Minsky issues a caveat against assuming that things are entirely black-and-white, taking event-driven to illustrate his point: a fund of fund's view on the environment for this strategy would be influenced by ‘external' macro factors, but also the expectations it has for the strategists it already knows well; if its roster is mostly distressed debt specialists its view on event-driven is to some extent a view on distressed debt, but if its roster has more multi-strategy managers who can switch between distressed and merger arb, the picture looks very different.
But that caveat may well beg the question about the kinds of strategists a fund of fund prefers that backs up Minsky's initial point. It seems fair to assume that a self-consciously top-down fund of funds that wants to retain the macro decision would look for managers that express, predictably, the systematic risk of a particular style. A bottom-up manager is much more likely to think of itself as selecting skill in making the style-rotation decision, effectively delegating the macro decision. IAM's managers are typically held for 4-6 years once they are approved, so it is no surprise that it prefers "a dislike of path-dependency and a proven ability to rotate portfolios in response to market conditions and shorter-duration trades", as Minsky puts it, whether that be variable-bias long/short equity, global macro that isn't constrained to playing yield curves or multi-strategy event-driven.
Similarly, the bottom-up-biased Stenham Advisors prefers multi-strategy managers when it comes to "cyclical strategies like merger arbitrage or distressed debt", according to investment director Javier Uribarren. Fawcett at Fauchier - who goes so far as to suggest that "if you can make a top-down call on a strategy then it is arguably no longer totally skills-based" - notes that the best hedge for 2008, CDSs on subprime CDOs, did not even exist five years before. "If you put that trade on were you style-drifting, or being clever?" he asks. "If you're making strategy calls, you may find your managers have structured their portfolio differently from that which you might have expected, because they have adapted and anticipated."
Maybe, but again, it is surely more likely that top-down-biased funds of funds would avoid these strategists because they have more systematic ideas of the strategies they are buying, and therefore a tighter benchmark for what they would regard as ‘style drift' and an aversion to managers that have strayed from their knitting in the past. While Funk notes that Liongate's current macro view informs a position in trading-oriented managers with their own macro-discretion, he adds that "it's true that we have a preference for managers with a narrowly-defined mandate". At Man Investments, co-head of portfolio management Anthony Lawler says: "We don't want style drift; we actively manage the portfolios so we invest with liquid managers so that we can make those changes ourselves." And at SSARIS Chung explains: "The best manager is not necessarily the one that produces the best results, it's the one that best fits what we are trying to do."
That does not preclude a preference for one suitable manager over another. Indeed, SSARIS is "very proud" of its manager selection, which Chung says has generated "significant alpha versus the indices" in all categories. Lawler warns that managers who are too ideological about the strategy allocation thesis can fall into the trap of assuming that "if style selection dominates, they therefore don't need to work on a manager selection to outperform the indices".
It takes us back to the earlier point that even the most cyclical strategies show manager dispersion in favourable environments, summed-up in the words of Permal's senior executive officer, Omar Kodmani: "Strategy selection does not replace manager selection - it's additive to manager selection."
No doubt this is true in best practice. But does any of this help us to identify best practice? Potentially it does, because it takes us beyond the obvious observation that performance is driven by both factors to the recognition that funds of funds' self-perception can have an impact on both manager selection and portfolio construction, introducing certain biases. That suggests key questions to ask about how these biases are managed. For the top-down manager: What efforts do you take to make sure that your long/short equity and event-driven managers maximise their information ratios in beta-favourable environments? How do you incorporate directional trading strategies? What is your liquidity profile, and what are the opportunity costs associated with maintaining it?
For the bottom-up managers: How do you monitor style drift? How do you monitor and act upon over-diversification or over-concentration of systematic risks at portfolio level? And how do you manage potential mismatches between the liquidity you get from your managers and the liquidity you offer to me?