“This volatility of insufficiently diversified portfolios lets greed and fear run portfolios, as investors stay invested for too long – or sell at the bottom and do not get back in,” observes Dravid. Indeed, there is much evidence to suggest that most investors are not adept at market timing, since money-weighted returns on most funds fall far short of time-weighted returns, because investors tend to buy high and sell low. Professional investors may fall prey to the same behavioural biases as retail investors – witness the influx of assets into CTAs in 2009, which would have been more usefully allocated the year before. Ampersand are not market timers. “There is an old proverb that the best time to plant a tree was 20 years ago. The benefits of extended diversification should hold true over time, and a well-diversified portfolio should hold up best in terms of risk-adjusted returns,” argues Dravid. Ampersand’s historical analysis suggests that the optimal portfolio to maximise Sharpe ratios might have as much as 30% in managed futures.

Overlay

Divesting assets in order to meet targets for alternative assets is too dramatic a decision for many institutions, where asset allocation is driven by committees and consensus. Bolting on alternatives, accessed through cash-efficient instruments such as futures, may turn out to be more feasible and more efficient. Ampersand’s company name comes from the & or ampersand symbol, which means ‘and’ in many languages. “AND (the ampersand) is superior to OR. We propose adding both diversifiers, and dynamic hedges, through an overlay. Rather than posting cash as collateral, existing holdings of stocks, bonds or ETFs can act as the margin collateral, which is usually between 10% and 25% of the notional exposure,” says Dravid.

The Ampersand ‘Full Monty’ enhancement goes further and replaces cash equities with equity index futures, freeing up more cash that could be invested in either fixed or floating rate interest-earning securities, which might be government, corporate or other securities, depending on investors’ risk appetite.

Conceptually, Ampersand is increasing exposure to an efficient portfolio to more than 100% of invested assets, as is contemplated by Modern Portfolio Theory (MPT) by the best possible “Capital Asset Line.” Ampersand dubs this increased exposure ‘extended diversification’. Although MPT may assume borrowing at the risk-free rate, funding through futures markets is highly efficient, based on implied financing costs of exchanged-traded futures markets.

Ampersand proposes much bigger risk allocations than have historically been typical for currency or global tactical asset allocation (GTAA) overlays (though the size of overlays will be tailored to each client). Ampersand argue that a 60% equity /40% bond portfolio expresses greater volatility and downside risk than a 100% stock and bond complemented by an overlay with 80% notional exposure that has been structured to mitigate risk while adding alpha. Again, this outcome is consistent with MPT which predicts that 100% exposure to certain assets will be riskier than greater than 100% exposure to an efficient portfolio.

Viewed from the opposite starting point of an existing alternatives allocation, the structure could also allow hedge funds of funds or multi-manager investors in liquid alternatives to increase (or reinstate) their exposure to conventional assets. Historically, many such investors have used notionally funded structures, including managed account and total return swaps, to invest in CTAs, and kept cash in Treasury bills. Ampersand suggest that the cash could be invested in equities and bonds, which should have a higher expected return.

The offering is customised to clients’ risk and return profiles, volatility and drawdown tolerance, and the basket of alternative managers can also be customised to client needs.

Diversifiers and dynamic hedges

Ampersand’s typical overlay contains two parts. The first is a basket of lowly correlated strategies, drawn mainly from both discretionary and systematic macro and CTA strategies that predominantly come under the tactical trading umbrella: “trend-followers, global macro, short-term traders, pattern-recognition traders, contrarian traders, machine learning, currency and commodity strategies; some relative value spread trading strategies can also be included, but all are expected to show correlations to equities no higher than 0.2,” says Head of Research, Dr Rufus Rankin. Managed account platform dbSelect houses over 100 such strategies, many of which profited in 2008.

The second basket is a dynamic hedge that is intended to be negatively correlated to equity strategies, but opportunistically rather than structurally. The objective is to reduce exposure to equities when they are going down, while attempting to stand aside in rising markets. “A simple example would be trend-following that becomes short of equities after a downtrend, but other strategies and asset classes that do well in a bear market could be used. One example here is volatility of volatility,” says Rankin. With Dravid having taught options at Wharton, Ampersand feel comfortable with volatility strategies, though they would not allocate to the types of short volatility strategies that blew up in February 2018. Some so-called ‘tail risk’ portfolio strategies are expected to lose controlled amounts in normal market conditions, but Ampersand’s “diversifying and dynamic hedge portfolios taken together have positive expected returns,” says CEO, Robert Enck. Where the overlay is 80% of the core portfolio, the split could be 50% in diversifying strategies and 30% in dynamic hedges. “This is just one illustration. Notional exposure to the overlay could range from as low as 20-30%, which Ampersand believe is the lowest level that makes a meaningful difference, to as much as 150-200%. Even at that level, we would seek to manage the risk of the overall portfolio and keep it fairly close to that of the original portfolio,” he adds.

Both parts of the overlay would need to be rebalanced to maintain target allocations: top-sliced after periods of outperformance, and added to after periods of underperformance, relative to the core portfolio. The rebalancing process is another potential source of added value: for instance, in early 2009, some profits would have been taken on CTAs, with proceeds redeployed into equities.

Defining risks

Ampersand participated in the Talking Hedge symposium ‘Redefining Hedge Funds for Institutional Portfolios’, held in Toronto in June 2018, where institutional investors discussed innovative trends in hedge funds, including Ampersand’s proposition.

The overlay could be difficult for investors to get comfortable with, where they view risk in terms of notional exposures. “The biggest challenge is to convince investors that more exposure does not equate to more risk,” says Dravid. Ampersand find it more meaningful to look at risk in terms of volatility. The low and negative correlations associated with the overlay, mean that it may only marginally increase overall portfolio volatility, if at all, and can potentially reduce volatility and drawdowns, under some scenarios. One example is a 60% equities/40% bonds portfolio with volatility of 10%, which only ticks up to 10.8% after adding an overlay of 80%. Other simulations show an overlay could even reduce portfolio volatility.

Additionally, portfolio diversifiers can change the pattern of returns. Many tactical trading hedge fund strategies, including CTAs, have a positively skewed return profile, which can help to counterbalance the negative skew of risk assets and produce a more symmetrical or favourably-skewed return profile.

Researching the counterparty exposure and risk entailed in a total return swap is a standard due diligence question. Notwithstanding the mitigation of counterparty risk via a third-party custodian holding the collateral underlying the swap with Deutsche Bank’s dbSelect, the perception of potential ‘headline’ reputational risk may deter some investors given the relentlessly negative media coverage around Deutsche Bank. Other platforms could also be used.

The risk that one or more of: exchanges, brokers and swap counterparties, might increase margin rates and haircuts under more stressed and volatile market conditions, should also be considered. Financial markets have been unusually calm for nearly a decade.

WTF = Why the Fee?

Ampersand proposes to be mainly or entirely remunerated based on whether its overlay enhances risk-adjusted returns. So great is Ampersand’s confidence in their proposition, that they can offer a fee structure which only pays for ‘value added’. “Many institutional investors have an allergy to performance fees and view them as bad per se. As fiduciaries, we look at after fee returns,” says Dravid. “We are willing to take a risk and put our money where our mouths are,” says Institutional Sales Director, Ryan Cosentino. The precise details of performance benchmarks for determining fees would be negotiated with each investor. Where a performance-oriented fee structure may be difficult for some investors to get signed off, Ampersand could accommodate a flat fee structure.

Of course, there are other fixed fees and costs involved in the structure. The underlying managers receive individual fees, which can sometimes be based mainly or only on performance but may involve regular management fees. Still, everything can be tailored and at least 50 alternative risk premia and alternative style premia strategies charging only flat fees now exist, with some of them accessible on the dbSelect platform. There are other fixed costs associated with the overlay structure. For instance, the TRS provider is expected to charge around 0.50% per 100% of notional exposure. For an 80% overlay, it could be 0.40%. Dravid argues that this is a small price to pay, to avoid the opportunity cost of having to sell core holdings to make space for alternatives.

Benchmarking

Having been in the markets for decades, Ampersand naturally realise that the overlay could underperform, though they expect positive absolute performance over most rolling 3 to 5 year periods. What may concern some allocators more than the absolute performance however is deviation from a relative return benchmark – or tracking error. For allocators who are subject to the straitjacket of a relative return benchmark comprised of conventional asset classes, a smaller overlay might be considered, until and unless they can reconfigure their benchmark towards an absolute return mentality. Ampersand feel that tracking error should be more acceptable anyway, if it is associated with positive alpha, which is what the Ampersand overlay seeks to provide.

Institutions with overlays

Dravid is not aware of any direct competitors: “unfunded overlays have been around for decades, but they have not been implemented in the form we envisioned as the collateral has usually been Treasury bills and cash,” he says.

Some of the world’s largest institutional investors are using overlays. Just as the largest foundations and pension funds tend to have bigger allocations to alternatives, so too they are structuring and financing portfolios in more efficient ways. PIMCO offers overlays, which can use government bond repos rather than swaps for financing. IPERS (Iowa Public Employees Retirement System) has set up its own liquid absolute return strategies (LARS) overlay to allocate to macro and managed futures and has put out an RFP for an investment overlay manager. Sweden’s AP1 is big enough to run its own internal overlay, which is another challenge – giant, multi-billion, asset owners are already pursuing Ampersand’s model in-house. Ampersand can cater for much smaller allocators and asset managers however.

Target client groups

Ampersand accommodates investors who can allocate at least $25 million to an overlay, and who like the Why the Fee concept. Potential clients could be allocation or target date funds; private funds or commodity pools; insurance company separate accounts or trust accounts; or registered investment advisers managing mutual funds, who might create a new structure where they run equities and fixed income, while Ampersand runs the alternatives sleeve through an overlay. Pension funds, foundations and endowments are also thought to be likely adopters of the concept. The business plan is for Ampersand to be running $5-10 billion for 25-50 clients, within 3-5 years. “This is a high touch business which fits a low volume of clients – unlike the funds business,” says Dravid. “Investors will gravitate towards what is best for portfolios, over time. We are talking to investors who are more innovative and who understand alternatives more than most,” Dravid adds.

Ampersand’s manager selection approach with Dr Rufus Rankin

Ampersand’s Head of Research, Dr Rufus Rankin, shared some insights into how Ampersand selects and combines multiple diversifying managers (17-18 in one fund) with the objective of producing an “all-weather” return profile.

Rankin is relatively unusual in having studied a BA in Philosophy, which he still finds “enormously helpful for clear thinking and asking good questions when there is so much behavioural inconsistency in financial markets”. In particular, Rankin points out that “investors are over-confident about forecasting returns, when in fact volatility is forecastable to a much higher degree. We are not trying to time allocations to managers, but rather seek to build stable, robust, portfolios”.

Rankin is also distinguished in that his postgraduate study took the form of a DBA (Doctor of Business Administration) rather than the more often seen PhD (Doctor of Philosophy). “A PhD should contribute new theory, whereas a DBA applies existing theory to a real-world problem,” says Rankin, and it was real world experience that helped him to clinch the final stage of the examination: an arduous, oral, viva voce examination. When interrogated about some heuristic parameters, Rankin’s riposte was that they derived from his real-world experience – something that the examiner could not argue with. Rankin pursued the DBA, which entailed some transatlantic travel to France’s Grenoble Ecole de Management – Grenoble Graduate School of Business, between 2008 and 2013 while working full time at Equinox. His thesis was entitled ‘Improving Multi-Asset Portfolio Diversification using Principal Component Analysis for Investment Selection’.

Principal Component Analysis (PCA) is one important input for Ampersand’s manager selection process, and Rankin has authored a concise, 58-page book on the discipline: Multi-Dimensional Diversification: Improving Portfolio Selection Using Principal Component Analysis. Various software packages can be used for PCA, and Rankin tends to use R and Excel with an add-in. PCA is not a silver bullet solution, as much subjective decision making is involved. “The real challenges are data and interpretation. It is data-hungry to find interesting, robust and stable factors. We need daily data for CTAs and now get it routinely on USB sticks. Then interpretation is needed to work out what components mean, and what the analysis finds. We start out with endogenous factors, and then rotate around using a statistical technique – varimax rotation – which also involves interpretation,” he explains.

Rankin is usually able to find 12-15 factors that explain a good proportion of the variance, and the mix evolves slowly, as trading strategies and markets evolve. Occasionally, Rankin finds a manager with a return profile apparently so idiosyncratic that returns cannot be explained by available factors, and this is likely to trigger further research, rather than an allocation. Rankin is aware that where managers are timing factors, the value of historical PCA is reduced.

Ultimately, “PCA plays a valuable role, in filtering strategies and confirming qualitative analysis, but it has to be complemented by qualitative understanding and does not have that much input into portfolio construction,” says Rankin. PCA has not been useful as an early warning for ‘style drift’, because much more data is needed for that. Multiple risk measures are monitored on a daily basis, at the strategy and portfolio levels.

Artificial intelligence (AI) and machine learning (ML) are hot watchwords and Rankin has lots of experience of AI – including outside finance. He once ran a translation firm and tried to use translation software for a Spanish class. “The output was close to nonsense, so I scrapped it and did the translation from scratch. There are still significant challenges in those areas,” he recalls. Rankin is wary of the hype around ML and AI. Ampersand uses some techniques found in AI and ML tool kits, is exploring neural networks, decision trees and clustering analysis, and allocates to managers who use these methods, but is not an AI shop per se. “AI may be more useful for cross-asset signals such as reading through from currency to commodity markets,” says Rankin.

Rankin sums up that: “the most important thing is to strategically allocate to well-diversified assets and strategies, periodically rebalancing them. Having lived through drawdowns and underperformance, I feel it is important to stick to a consistent process and risk framework, rather than getting pressurised into removing managers without good cause”.