OMGI’s Global Equity Market Neutral Strategy

Cayman Arbea fund augments GEAR alpha

Originally published in the March 2018 issue

Some allocators always maintain a structurally steady allocation to equity market neutral as a portfolio diversifier, both for long only equity or bond exposure, and for other hedge fund strategies, many of which have become increasingly correlated to equity markets. Others more opportunistically and tactically allocate to the liquid strategy. In early 2018, the perception that some equity markets could be trading on elevated valuations, particularly based on cyclically normalised earnings, is one argument for reducing equity beta. Additionally, rising interest rates can boost absolute returns from most market neutral hedge fund strategies. As long portfolios are substantially financed by the proceeds of short sales, market neutral strategies will often be holding the majority of net assets in cash, which earns interest. Prior to the past nine years of QE, ZIRP and NIRP, market neutral strategies typically earned several percentage points of return a year from cash, plus or minus their trading performance.

Old Mutual Global Investors (OMGI)’s systematic global equity market neutral strategy, Global Equity Absolute Return (GEAR) has, since 2009, generated some of the best and most consistent risk-adjusted returns in the space, winning The Hedge Fund Journal’s UCITS Hedge performance award on several occasions. GEAR’s fee structure has marginally helped it to outperform peers that charge higher fees. The management fee is 0.75% and performance fees of 20% apply only to alpha.

GEAR targets annual volatility of 6% and a similar spread over cash returns, to imply a Sharpe of around 1.0. It has surpassed the Sharpe target but has undershot the volatility target, due to the climate of low cross-sectional volatility.

In 2013, OMGI launched a Cayman version of the strategy, Arbea, with a higher volatility target: 9%, moreleverage, and potential to expand its balance sheet in order to hit the intended volatility.

Whereas GEAR keeps gross exposure between 180% and 220%, Arbea’s target range is between 300% and 400%. Arbea’s management fee of 0.75% works out lower than GEAR, on a risk-adjusted basis. Arbea has matched its 9% annualised return target and has averaged volatility of 7.6%, between July 2013 and December 2017 and delivered a Sharpe in excess of 1.0. Given that Arbea caps single stock position sizes at 1.5% of gross exposure, it may need more single stocks than GEAR in order to deploy its capital. Factor or style weightings are very similar between GEAR and Arbea.

Liquidity, capacity and scalability
Roughly 5,500 equities globally meet OMGI equity market neutral strategies’ liquidity criteria. “Liquidity risk management is also exactly the same for GEAR and Arbea, to ensure that returns are scalable,” says Ian Heslop, one of the three fund managers, and Head of Global Equities at OMGI. Arbea has monthly liquidity and a 3% exit fee in year one, to attract stickier money, and not because this is needed to align fund and portfolio liquidity. There are no gates or restrictions on monthly redemption. “We could manage more than the current assets ($17.9 billion*). No hard number is articulated for strategy capacity, but we would not want to grow assets to the point where returns were degraded,” says Heslop.

OMGI’s independent teams
Each of OMGI’s investment desks make their own decisions on strategy capacity, and some strategies have been hard closed before. For instance, the discretionary and fundamental, long/short UK mid cap strategy, UK Specialist Equity, is virtually hard closed (strictly speaking it is soft closed to the extent that amounts of $100,000 per dealing day, per client, are permitted). “OMGI purposefully has no CIO and strategy capacity is not a business decision,” says Heslop.

The investment desks operate autonomously, and subject to risk management oversight, while OMGI handles non-investment activity. “The whole premise of OMGI is proprietary and independent processes, signed off by teams who have control and ownership of their own investment process,” explains Heslop. Within OMGI, there are strict Chinese walls between the quantitative team and the fundamental discretionary groups, such as Richard Buxton’s UK equity strategy, which runs at much higher concentration and much lower turnover. “They are entirely independent desks,” says fund manager and head of systems, Mike Servent.

Long term vision and alignment
There are no siloes within the highly integrated quantitative team. “We meet at least twice a day, so there is a short feedback loop between markets and the portfolio,” says head of research, Amadeo Alentorn. The three fund managers are supported by three analysts and a strategist.

The quantitative process is also applied to long only strategies, which use the same return forecasts as the equity market neutral strategies but have a relative return equity benchmark instead of an absolute return benchmark. The long books of the equity market neutral portfolios need not perfectly overlap with the long only portfolios, partly because the latter typically have a tracking error constraint of between 3% and 5% versus global or regional MSCI benchmarks. This also explains why the long only books may very occasionally, and temporarily, own a stock that the equity market neutral is short of. “When the alpha forecast is adjusting, the position slowly moves from long to short in GEAR and from overweight to zero in long only,” says Heslop.

The quantitative team are committed for the long haul. The portfolio managers have been working together for over thirteen years. Their interests are aligned with investors’ interests: remuneration is closely linked to fund performance with bonuses co-invested in the funds and subject to multi-year lockups (which predated AIFMD rules on bonus retention). These managers are eating their own cooking.

Performance attribution
GEAR and Arbea are delivering their primary aim of uncorrelated returns. With beta near zero, the coefficient has been -0.05 for both GEAR and Arbea, performance is more or less all alpha, with 90% attributable to single stock selection, 10% coming from sector tilts and nothing from country exposures, according to OMGI. The strategy can take significant net exposure to industries and sectors, but net exposure to regions and countries is capped at minimal levels. Given the overriding constraints, long bets on some industries and sectors need to be counterbalanced by short bets of very similar size on others. Essentially it is beta neutral, country neutral and currency neutral, but not sector neutral. “We pick the right stocks within regions,” sums up Heslop. Attribution has been broad based: all five factors have made money, and stock selection within virtually all sectors has contributed positive returns since 2009.

Returns on capital have varied between regions, as have allocations to regions. Since 2013, Asia Pacific has been the biggest profit centre. But Europe (which, being Pan-Europe, includes the UK) has contributed most since 2009. Viewed through the prism of OMGI’s factors, Europe’s diverse and heterogeneous equity markets are ideal. “European equities have a broader spread of factor returns, so more factors work consistently through time,” points out Heslop. This underscores how different market neutral investing is from directional investing. For long only investors, European equities have of course lagged US equities by a massive margin since 2009.

Japan has been the least profitable geography. Heslop thinks that this is because the advent of Abenomics in 2012 has changed the drivers of Japanese stocks: “returns from basic value were great pre-Abenomics. Now there has been a marked structural shift where the value factor no longer works. Momentum does not work either”.

OMGI does not trade individual pairs of stocks, as some sector-neutral strategies do, but rather views the whole long book as the counterpart of the whole short book. The long side has out-performed the MSCI World Equity index and the short side has under-performed it. Over a full market cycle – which OMGI carries out simulations for – Heslop would expect the long and short books to make broadly equal contributions, though the shorts might make slightly less due to stock borrow costs. Since 2009 equities have been in a bull market and the long book has contributed the larger part of returns, but Heslop stresses that this is very time-varying. There have been multi-month periods when shorts made most of the alpha – and other periods when both books were making similar amounts.

OMGI’s team have lived through various regulatory shorting bans, usually focused on specific sectors or stocks in certain countries, without noticing much impact. “The bans prevent you from adding to your position, which stops increases in gross exposure. But regulations tend to be reactive, so we already had shorts on before the bans,” says Heslop. OMGI always obtains short exposure to single stocks rather than sectors or indices, which are often cheaper to short but reduce or remove scope for alpha generation on the short book. Stock borrow costs are part of the model and the alpha forecast, so some stocks may be too costly to short.

OMGI’s proprietary factors
Stock selection is substantially driven by OMGI’s proprietary style factors. The five factors (Dynamic Valuation, Sustainable Growth, Analyst Sentiment, Company Management and Market Dynamics) have names that sound similar to generic factors such as value, growth and sentiment. OMGI is distinguished by how it defines and measures its own factor concepts, some of which combine more than one traditional factor idea. “Our factors are carefully constructed using multiple data sources and the associated return streams are very lowly correlated with generic factors,” says Alentorn.

For instance, it is well known that a generic value factor (usually price to book value), on a long only basis, has greatly underperformed growth between 2009 and 2017. Applying the same factor on a market neutral, intra-sector, basis has also underperformed, but to a lesser extent, Heslop observes. But OMGI’s value factor has performed far better and Heslop reveals that OMGI does not invest in value independently. Amongst many nuances, OMGI considers value in conjunction with quality considerations, partly to avoid ‘value traps’. This enhanced factor has greatly outperformed a generic value factor, as shown below. “It has also done best in the US market, where generic value has performed worst,” adds Heslop.

OMGI’s growth factor, Sustainable Growth, is also distinctive. It does not go long of growth and short of non-growth stocks, but rather seeks to identify mispricings within the universe of growth stocks. One example is that risk-loving investors may overpay for stocks with potentially lottery-like payoffs that display rather erratic or ephemeral patterns of growth.

Combining the factors
Dynamically and regionally varying weights are applied to the five factors, to arrive at scores for individual stocks, which can be positive, neutral or negative. The factors, and other constraints such as stock borrow and transaction costs, collectively do not throw up a strong enough positive or negative signal to warrant either a long or a short position for the majority of the investment universe, at any point in time. GEAR and Arbea typically have positions in between 750 and 1,250 stocks, which works out at between 14% and 23% of the investment universe of 5,500 stocks.

Whereas some generic factors, such as growth and momentum, can become highly correlated over some periods, OMGI’s factors are designed to be orthogonal over multi-year periods. “On average, the five factors should have zero correlation through the cycle,” says Servent.

Dynamically varying factor weights
Dynamically varying factor and style weightings is a hallmark of the OMGI process. Some strategies, particularly alternative risk premia, style premia, or smart beta strategies, tend to rebalance to maintain roughly constant factor weights. OMGI has previously managed strategies with constant factor weights. OMGI today are varying weights to express active model signals on timing the factors. “The investment process evolved to deal with the cyclicality of factor returns, which is endemic in factor and thematic investment. There are periods when some factors are not rewarded, and some factors, most importantly value, can become correlated to the market. We have re-engineered the process to manage the correlation and downside risk,” says Heslop.

The other factors are also cyclical, and the timing decisions are taken from many angles. For instance, “Momentum performs differently in high and low volatility climates. Dispersion tends to generate positive returns for all factors, so we try to be as efficacious in low as in high dispersion markets,” says Heslop.

OMGI will not “bet the ranch” on, or completely remove, any one factor however. Style weights fluctuate within bands, at the overall strategy level, as shown below.

“We dynamically adjust towards, or away from, styles, to be consistent over time,” says Heslop. This process is governed by how the models categorise market regimes, which can vary by region. “Applying the same factors to all regions is seen as a test of robustness. The weightings of factors within regions can vary, as different regions may be in different types of market regimes,” adds Servent.

Factor tilts in early 2018
When we met OMGI in late January 2018, growth and analyst sentiment were being emphasised while management quality was being de-emphasised, at the overall strategy level. “The Sustainable Growth factor is towards the top end of its weighting range, not because growth has performed well for so long, but because the market has rewarded this factor during historical regimes that are categorised as being similar to the current market environment,” explains Heslop. The Analyst Sentiment factor – which uses consensus earnings forecasts and recommendations because OMGI did not identify extra value from ranking different analysts’ output – was also a relatively high weighting, as it had predictive power in the current market regime. Heslop observes that analyst sentiment signals tend to work better in a bull market when the chain of causation flows from analysts to the market, whereas the reverse applies in a bear market. “The Analyst Sentiment factor is a short-term component that helps with the timing of entry and exit,” adds Alentorn. In contrast, the Company Management quality factor was being downplayed because the prevailing market climate was not historically conducive to this factor.

Old Mutual define market climates using sentiment and risk appetite indicators which map maximum pessimism, to include the Lehman failure and the TMT bubble bursting, while maximum optimism includes the start of Abenomics in Japan. In early 2018, sentiment had, unsurprisingly, been drifting up in all regions. Old Mutual’s risk appetite measure is perhaps less intuitive. Though many observers, such as Goldman Sachs, suggested that risk appetite was reaching record levels in early 2018, OMGI’s indicator suggested low risk appetite due to the types of stocks investors were buying. OMGI gauges the animal spirits from a cross-section of investor behaviour. Heslop observed that “the combination of low volatility at the aggregate market level, and falling risk appetite at the investor stock preference level, was unusual”. This proved to be rather prescient as volatility exploded in early February.

Academic consultants
OMGI does not exist in a vacuum. The team have advanced degrees or PhDs and they keep in touch with advances in academic research. As Head of Research, Alentorn oversees long-standing relationships with external academic consultants. They include four eminent professors – Dr Ian Marsh of Cass Business School at City University; Dr Steve Satchell, Fellow of Trinity College at the University of Cambridge; Dr Mark Salmon, Professor of Finance at the University of Cambridge; and Dr Peter Pope, Professor of Accounting at the London School of Economics. “Each has unique expertise. They come up with ideas in draft papers and can act as a sounding board for our research, giving us an academic perspective,” explains Alentorn. In addition, OMGI wants to get exposure to more cutting edge and blue-sky research done by a pool of less experienced researchers: postgraduates including PhD candidates, who are associated with the above academics. This “virtual lab” is tailored on a project by project basis. Naturally this research is riskier. It involvestrial, error and blind alleys but the means can matter as much as the end. “Projects do not always succeed but we learn a lot along the way,” reflects Servent.

Systems and technology
Part of OMGI’s edge comes from its differentiated analytical approach and technology is vital to process vast amounts of ever changing data. OMGI is distinguished by having a systems person, Servent, as portfolio manager. “We realised early on that systems are integral to what we do. The investment process is integrated with technology. We write systems, estimate proprietary models, conduct back-testing and simulation, and run trading processes, with our systems,” he says. Virtually all of the systems are proprietary. “We write nearly everything ourselves so that the systems are uniquely responsive to the needs of the team. The only thing we do not write is mean/variance optimisation, where we use Axioma’s package,” he adds.

All of the managers and analysts are coders, who can read and write code. The MATLAB language is used. “We are not into technology for technology’s sake. We use technology that gets the job done in the cleanest and most robust and most transparent way,” says Servent. Within the OMGI team, the ethos is very much open source. “There are no black boxes. We take care to write concise and transparent code. The analysts spend most of their time reviewing code,” adds Servent.

The models are constantly evolving and have so far been through 22 incarnations, informed by a simulation process using 20 years of data that naturally includes the famous “quant meltdown” month of August 2007. “It is easy to draw incorrect conclusions from simulations, but we are experienced users,” says Servent. OMGI’s alpha generation is the product of multiple, incremental enhancements over many years since OMGI began running equity market neutral strategies in 2001. There is no single secret or silver bullet. “Our models come from a multi-decade evolution of gradual improvements,” sums up Servent. The managers are confident that this process will continue.

*as at September 2017