Millennium Global Investments (MGI) was founded in 1994 and is one of the world’s oldest currency managers, which has survived and thrived over nearly three decades including through some challenging performance and regulatory episodes for the wider industry that have seen other currency specialists shrink or perish. In contrast, MGI’s currency assets under management have recently reached an all-time high of $20bn, thanks to strong net inflows (and no outflows) in 2021 and the firm regularly receives awards for its currency management expertise. Clients are predominantly large institutions in North America and Europe, including pension funds, endowments, foundations and central banks. These institutions have always had a long term strategic need to manage currency risk, and some of them also view currencies as an increasingly opportune source of alpha and portfolio diversification as inflationary shocks and higher interest rates threaten historically high valuations for equities and bonds.
Investors are gravitating towards currencies as the epicentre of macro, where finance, geopolitics, economics and the pandemic response all touch.
Mark Astley, Co-CEO, Millennium Global Investments
Some $16bn, or over 80% of MGI assets, are actively managed, making MGI the world’s largest provider of active currency solutions, according to IPE. MGI’s active currency overlay or dynamic hedging mandates allow clients to strategically define or tactically shift the balance between active or passive exposure (and passive hedging is also offered on a standalone basis).
Active fundamental discretionary management has been a hallmark of the firm from the outset. What has changed in recent years is diversification into systematic strategies, and MGI therefore now has two co-chief investment officers, for its discretionary and systematic strategies. CIO for discretionary, Richard Benson, has been with the firm for 13 years of his nearly 20 years of currency market experience having started his career in the Royal Navy. CIO for systematic, Umberto Alvisi, has spent over 20 years in currency markets, with the last 8 years at MGI after roles at Citadel LLC and Credit Suisse.
Most clients now blend MGI’s discretionary fundamental strategy with its systematic currency strategy. Some clients do focus purely on discretionary or systematic solutions, and the systematic strategy is also now investible via a UCITS fund that was seeded by a large European institutional investor with over $100 million. Beyond the investment style, most mandates are customised to some degree in terms of volatility targets and benchmarks where relevant, as well as operational structures.
The discretionary and systematic investment professionals, numbering twelve in total, are aware of each other’s positions and trades, but have independent processes, applied to different investment universes, and implemented via partly different instruments and risk management approaches. The systematic strategy currently trades only G10 developed market currencies whereas the discretionary strategy can also trade emerging market currencies. The discretionary strategy can use a variety of vanilla and exotic options in addition to FX forwards to express its views, while the systematic strategy implements signals via linear instruments (though it also gleans insights from options markets). Discretionary management often uses stop losses to manage risk while systematic awaits signals.
Investors appreciate the low historical correlation between the two strategies, and they are philosophically different: “Discretionary macro is more forward looking and exploits a deep understanding of macro but can fall prey to inherent human behavioural biases and heuristics. A quantitative approach avoids these biases, but can be backward-looking, though we have managed to identify some forward-looking inputs. Most clients opt to synergise alpha from both strategies, and gain from a multi-factor framework,” says co-CEO, Mark Astley.
Millennium Global Investment (MGI) was founded in 1994 and is one of the world’s oldest currency managers
MGI founder, Michael Huttman, who was once CIO, is now Chairman. He maintains a regular day-to-day dialogue with the investment team and retains strategic oversight of initiatives such as the launch of FinTech, MillTechFX, which helps clients to optimise their currency execution. He developed an early interest in cryptocurrencies some years ago, but MGI has cautiously awaited the emergence of an institutional regulatory and service provider framework before preparing its upcoming cryptocurrency offerings.
Headquartered in London, the firm opened a Paris office in 2019, which makes it easier to service European clients post-Brexit and there are also relationship management offices in Tokyo, Boston and San Francisco, serving institutional investors in Asia and North America. MGI maintains strong relationships with major global investment consultants who advise large asset owners and the firm’s strategies are listed in institutional investor databases such as eVestment.
ESG is manifested in the firm’s governance structure, diversity of staff, and approach to proper and ethical trading of currencies. MGI contributed to, and is a signatory of, the FX Global Code, which was first promulgated in July 2017 with the latest version released in July 2021. The firm is also a signatory of the UNPRI. MGI’s macroeconomic research heeds ESG country rankings, and some clients, including a large US West Coast public pension plan, receive ESG research reports prepared by the economics and strategy team. ESG is thus integrated into the discretionary investment process.
In early 2022, investor interest in currencies is partly based on concerns about valuations of equities and bonds, and fears that a 1970s style “stagflation” could be adverse for both asset classes. Already the first half of 2022 has seen developed world inflation reach multi-decade highs and simultaneous losses for bonds and equities, which upends the simplistic but widely followed approach to portfolio diversification predicated on long exposure to these two asset classes alone. “Some currency strategies are negatively correlated to equities and bonds and can even become part of tail risk strategies. This can help to truncate the left tail risk issue,” says Astley.
Global macro funds have often made the right calls in 2022, and currencies are a popular arena for expressing macro themes. “Investors are gravitating towards currencies as the epicentre of macro, where finance, geopolitics, economics and the pandemic response all touch. The potential for reversing the multi-decade globalisation trend could mark a sea change in global economics. Demographic trends such as an ageing population are reducing labour supply. Meanwhile, ambitions for net zero carbon emissions are increasing costs and inflation,” argues Astley.
Timing the patterns of inherently cyclical currency markets can contribute alpha streams that have historically been uncorrelated with equities and fixed income. There are also a priori arguments for the decorrelation: equities are broadly driven by earnings growth and valuations, and bonds by interest rates and inflation, whereas currencies are driven by a wider range of factors that also change through market regimes.
For some clients, the focus is mainly or wholly on risk management and mitigation rather than return maximisation or optimisation.
Raw currency risk is recognised as an uncompensated source of volatility. Conceptually, global equity investors have two portfolios: a local currency equity basket with around 14% volatility and a historical long term risk premium of 6-7%, and a currency basket with around 8% volatility and expected return of zero. Currency risk overall is the second largest source of volatility in many institutional portfolios. Over some time periods, it can account for over half of the volatility of an equity portfolio, and it typically explains over 80% of the volatility for a global government bond book. “It is irrational to hold this sort of portfolio exposure, when it can be dynamically managed to reduce drawdowns and mitigate cashflow constraints that can arise from a 100% hedging approach,” says Astley. In between the extremes of zero hedging and 100% passive hedging, more opportunistic approaches, such as dynamic hedging which varies hedge ratios, and active currency overlays with more flexibility, are increasingly sought after.
Even if investors believe they are fully invested, they may have some latitude to at least reconfigure their currency exposure. Currency hedges, alpha overlays or standalone mandates can be structured to increase or reduce overall portfolio volatility targets, or simply to maintain constant risk but redeploy the amount and type of currency risk, with sensitivity to investors’ liabilities or performance benchmarks.
The ability to run currency strategies on a substantially unfunded basis, usually via managed accounts, can make them capital efficient. Beyond volatility budgets and cashflow constraints, strategies and mandates can also be customised to other objectives, tolerances, investment universes, preferred instruments, vehicles, counterparties and service providers.
With daily turnover of $6.6trn in 2019, according to the triennial BIS survey, currency markets are more liquid than equity or bond markets and about 25% of turnover now comes from emerging market currencies. Bid/offer spreads as low as one basis point can make currency transaction costs very competitive, though careful attention to market conditions and counterparties is needed to obtain best execution.
FX execution needs to be optimised for MGI’s own investment and hedging strategies with turnover of over $600bn a year. These large volumes of trading combined with longstanding relationships with a multibank platform of counterparties including the top 12 wholesale bank providers, help MGI to shop around for keen prices and execution. Transaction cost analysis (TCA) reports are provided to demonstrate the quality of execution.
Sister company, Millennium Global Treasury Services, (trading as MillTechFX) trades with the same counterparties as MGI, facilitating high quality multibank execution for clients, such as asset managers, private equity groups, and corporates. MillTechFX not only reduces these clients’ execution costs but also reduces their operational burden with a service underpinned by currency experience and technological investment.
MGI has been developing its systematic strategy since 2011, originally for dynamic hedging before moving on to pure alpha strategies. The models have evolved, with the current suite of signals applied since 2016. They have generated an information ratio of 0.7 with a positive skew, negative correlation to conventional asset classes and hedge funds, and exceptionally strong profits generated after the Brexit and Covid events.
The alpha engines of the model are risk premia, and momentum management. “The investment philosophy evolved to retain macroeconomic fundamentals as a medium-term anchor, while acknowledging that shorter term factors viewed as “market dynamics” can also push currencies away from fundamentals. The combination is needed to create a process to deal with all market dynamics and environments,” says Alvisi.
The risk premia module opportunistically trades two well-known factors – value (defined by purchasing power parity) and medium-term momentum, while its third factor is more unusual: relative equity performance to gauge business cycles.
The risk premia module uses traditional backward-looking momentum, but market dynamics are gauged through an innovative indicator for the momentum management module. “Traditional trend following momentum signals extrapolate from price history. We do not follow a traditional momentum approach based on historical price patterns. Assuming persistence runs into problems with the stochastic, rangy nature of currencies, so we use forward looking indicators based on implied volatility in the OTC options markets. We look at risk reversals, which measure the difference between call and put option pricing at the same delta,” explains Alvisi.
These currency option markets are another way in which the asset class is differentiated. Whereas in equity markets put options will nearly always cost more than call options of the same delta, in currency markets the ratios between upside and downside volatility move around more. This makes relative option pricing a more symmetrical signal. Forward looking information from the implied volatility surface is therefore a snapshot of market expectations of future returns, which immediately includes all information in the market. “It immediately reprices new information such as the Russia/Ukraine situation or the Covid pandemic, before market prices. The market started pricing in Russia/Ukraine around February 14th/15th 2022, while the global Covid pandemic was reflected in option prices as early as the second half of February 2020. Thanks to this early signal, March 2020 was one of our best months. Brexit risk was also flagged up by option markets. Cable (GBP versus USD) rallied ahead of the Brexit vote, reflecting consensus expectations of a “Remain” vote, but the option market skew – with puts priced 4 times more expensive than calls – reflected the downside risk. Even the Swiss Franc de-pegging in early 2015 was to some degree reflected in option skew pricing,” says Alvisi.
The strategy is not an explicit tail risk protection strategy, but it has so far profited from some big shocks. “We have tended to be long of tail risk because option markets are a very useful signal of tail risk. The portfolio can quickly move into risk off mode, as it did in February 2020 and February 2022,” says Alvisi.
Option markets can of course throw up false alarms, but profits from their accurate warnings have outweighed losses from the wolf-crying. “Our hit ratio is nearly 60%, but our win-loss ratio (or “slugging ratio”) is so high that we do not need to be right all the time. Even with a hit ratio of only 50% we would still have made good returns,” says Alvisi.
Option market information is used purely as a source of information for trading signals. Positioning is always expressed through the directional trading of FX forwards, which are “delta one” linear instruments, rather than options.
We use what is called non-linear regressions in traditional econometrics, because the world is non-linear.
Umberto Alvisi, Co-CIO, Millennium Global Investments
The program’s starting point is to equally weight risk premia and momentum management, but in practice it dynamically rebalances in two ways: between momentum management and risk premia, and amongst the three factors within the risk premia strategy.
Risk premia versus momentum management rebalancing varies the balance between the two broad modules. “In late 2020, risk premia signals almost disappeared, because USD depreciation was the overwhelming market driver at that time and the risk premia signals were not adding value. The risk signals are endogenously driven by the strength of the signal, reflecting the models’ conviction,” says Alvisi.
Though option market signals can be fast moving, the portfolio does not shift to a new regime overnight. In 2020, it was “risk-off” in March and used April to transition to a “risk-on” stance by May. “It would be difficult for anyone to pick the exact bottom of the market. We lost some money in April 2020 but less than what we made in March 2020. We then stayed “risk on” for the rest of the year, when USD depreciation became an important driver,” says Alvisi.
Rotating around value, medium term momentum and relative equity performance, also reflects how different factors work better in different regimes. There is no structural bias to be long or short any of these factors, but there can be temporary and tactical exposures to them.
“They are switched on and off as markets move between value and other regimes. A risk premia approach had traditionally been very static, which can be quite dangerous for market dynamics and regime shifts. We therefore wanted to make risk premia aware of market conditions, so that we can identify the factor risks in each position and reduce unrewarded risks. Machine learning statistically discerns which factors are relevant,” says Alvisi.
The detection of regime changes is informed by advanced analytical techniques, which include non-linear regressions using machine learning techniques, such as sparse learning, to identify regime shifts and find out which drivers are at play in different environments. “We use what is called non-linear regressions in traditional econometrics, because the world is non-linear,” says Alvisi. The machine learning approach is “supervised”. “We drive the process of signal investigation, positing relations between currencies and different financial/macro variables, so that we can fully explain the model signals. This fits in with our fundamental background. It also means that in periods of drawdowns we can understand better the reasons that are driving the losses,” he adds.
The blend of dynamic risk premia and momentum market dynamics components is unusual and provides two dimensions of diversification. “The two models have a low but positive correlation and additional layers of diversification come from trading ten different currencies, and trading over a range of frequencies,” says Alvisi.
All of this marks an advance on a traditional, first-generation approach to trading currency factors. “A traditional carry-value-momentum factor approach, which produced strong performance between about 2000 and 2010, essentially stopped working over the past decade. The information ratio went from 1 in the first decade of the millennium to 0.1 in the second,” says Alvisi. The reasons are unsurprising in hindsight: “The collapse of developed market interest rates meant that there was simply not much if any carry left amongst G10 currencies. Simple momentum signals have struggled due to the lack of persistent volatility in foreign exchange markets. And value only proves useful over very long timeframes”.
Trading time horizons range from one week for shorter term, momentum management models, to one or two months for slower moving, medium term risk premia models. Annual transaction costs average out at 1%-1.5% for a 10% volatility portfolio.
Transaction costs are monitored and optimised against the objective of executing signals in a timely fashion. “Bid/offer spreads are competitive compared with other asset classes. They average 2-3 basis points and are rarely above 6-7. Sometimes timely execution is more valuable than the transactions costs. Sometimes the trading desk need to work the order to get better prices and execution for certain types of trades and currencies,” says Alvisi.
The maximum leverage for the strategy targeting 10-12% volatility would be 55% for each of the nine currency pairs, theoretically adding up to 495% but neither leverage nor volatility are kept constant. Both overall leverage and exposure per currency are opportunistically varied. In practice, the maximum overall leverage has been 290%, and the average has only been 120%.
Though the maximum risk budget per currency pair is equal, the utilisation of this risk budget does vary with the models’ conviction. “There is often exposure to all nine pairs, but it is rarely equally weighted,” says Alvisi.
The Value at Risk (VaR) UCITS approach, rather than the commitment approach, is used to allow flexibility on leverage. “We have never come close to the VaR limits,” says Alvisi. The strategy naturally fits into UCITS so there is no need to create indices and associated swaps. G10 currencies were chosen because they are the most liquid. “The strategy could contemplate adding emerging market currencies, but their gap risk and illiquidity would need to be allowed for,” adds Alvisi.
There are differences of opinion about whether various cryptocurrencies should be defined as currencies, securities, or even commodities, given the importance of electricity pricing inputs for some of them. All these labels and others besides could be relevant for different units. In any case, MGI has been interested in the space for over 5 years and has now set up a dedicated unit named Sonar Digital by Millennium Global, and appointed Lisa Scott-Smith, Managing Director (Digital Business), and Simon Lack, Director (Digital Asset Solutions), to lead it. Its purpose is to provide the opportunity to participate in the digitisation of finance with institutional grade products and governance.
“Crypto is not a single asset class. Various units could be used for transactions, store of value, transfer of wealth and financing. For storing value, cryptocurrencies are potentially more attractive than gold because they are completely immutable, more divisible and have no transport or storage costs,” argues Lack. “We want to participate in and contribute to the institutionalisation of the space, but the right infrastructure is needed because operationally it does not look, smell or feel like any other asset class. When we first looked at the space in 2018, custody issues were an obstacle because the established definition did not fit in with blockchain or distributed ledgers in the cloud. That has changed, with sophisticated new custody providers such as Gemini, Copper, Fidelity and Coinbase. Exchanges are adapting their KYC and AML to cope with sanctions issues. We are assessing the ecosystem of service providers. We use 27 years of fiduciary experience and relationships with banks, regulators and auditors to carry out due diligence assessing the risk management and processes of custodians, administrators, liquidity providers, to gain a deep insight into the implications of working with each provider. We are building an institutional grade offering.”
Sonar Digital by Millennium Global contemplates offering direct exposure to cryptocurrencies, rather than using futures, ETFs, closed end funds or other derivatives and wants to create a next generation product: “Existing offerings have weaknesses. Closed end funds or futures can trade at discounts or premiums. Most vehicles are long only whereas we aspire for an absolute return, alpha strategy,” says Lack.
In common with MGI’s main currency management business, Sonar Digital contemplates using a blend of discretionary and systematic investment processes, which could include some distinctive approaches: “We are also exploring some innovative technologies and data analytics,” says Lack.
Cryptocurrencies are spawning their own Decentralised Finance (DeFi) ecosystem, and some strategies now engage in borrowing and lending often at much higher yields than developed market fiat currencies. Lack aims to obtain a holistic perspective on the range of risks and returns from DeFi: “For borrowing or lending of cryptocurrencies, we will develop a very cautious and deep understanding of the risk profile of the activities and protocols. We would carefully assess a wide range of risks, including exchanges, counterparties, smart contracts, anonymity, technology, and cyber risk,” he points out.
“Overall. we are excited about the breadth of opportunity. Participation is growing and crypto is here to stay,” he sums up. Investors should watch this space for more on Sonar Digital’s crypto launch.
The last five years have been a challenging climate for discretionary currency investing: “Alpha tends to be episodic and lumpy,” says Richard Benson, CIO of discretionary investment management. “Financial repression, interest rate suppression, subdued volatility and lack of differentiation between economies and cycles meant it was not so easy to exploit alpha from a traditional discretionary approach,” says Astley. “Extreme liquidity provision to the system blocks macro discrimination between currencies. The Fed’s balance sheet expansion swamped all else and currency volatility was low even as bond market volatility began to pick up in late 2021,” underscores Benson.
2022 could be a game changer for several reasons. “QE is finally coming to an end and the yield curve has flattened and even inverted in some areas. We now have a rising rate environment and differentiation between central bank policies is a good environment for discretionary management. In early 2022 it seemed that the Federal Reserve, Bank of Canada and Bank of England were moving rates up while the Bank of Japan was on hold. The ECB pivot to a hawkish stance in February, and its rate rise, was a real game changer because it means fixed income yields are no longer negative in Europe. The ECB’s switch back to a more dovish stance after the war was another surprise,” says Benson.
Russia’s invasion of Ukraine has also been momentous for FX. “Currency implied volatility in early March 2022 had quadrupled over multiple time windows, as investors wanted to express risk off portfolio tail hedges through Euro downside,” says Benson. The war led to steep drops in Eastern European currencies that were perceived as a proxy for the Rouble, which became difficult or impossible to trade. There were also fundamental concerns about how the war might impact supply chains in certain countries, such as automakers in Czechia. Benson decided to use reverse knock out option structures, and European digital or binary potions, to gain exposure to the Polish Zloty and Czech Koruna currencies. “These exotic options provide a cheaper way to wager on upside while limiting outlays just as a plain vanilla option does. Option exposures can have a long or short volatility bias depending on the pricing of implied volatility. The Euro exposure had some short volatility exposure to take advantage of the gap between implied and realised volatility,” Benson explains.
The discretionary strategy has typical trade time horizons of one to three months but does also adjust positions and react to news and events daily. “Either information changes or price changes relative to the information,” says Benson. Risk budgets can expand with opportunities, and the independent risk managers sometimes encourage more rather than less risk-taking.