How Asia FX Futures Can Improve FX Momentum Strategy Performance

• G10 FX momentum strategies have delivered poor returns over the past decade.
• Part of this could be due to a lack of diversification; Asian FX futures could therefore add significant value to such strategies.
• We test a wide range of momentum trading rules and find that Asian FX futures deliver superior returns to G10 currencies (Chart 1).
• Therefore, adding Asian currencies to existing G10 FX momentum strategies could help redress underperformance issues.

FX momentum or trend-following strategies are some of the most common types of FX trading models that investors use. Their use began in the late 1970s as currency markets in developed markets started to get liberalized. Early on, academics were sceptical of such models because they violated the efficient markets hypothesis, which states that past prices should contain no information beyond current prices. However, with the persistent profitability of momentum strategies over the 1970s and 1980s, academics have since developed theories to justify those returns. These range from limits to arbitrage to the presence of sizeable non-profit-seeking players in FX markets like corporates or central banks.^[1]

However, just as interest has grown, the profitability of momentum strategies has fallen. Taking two versions of FX momentum strategies that are regularly updated – one that the author developed in the mid-2000s and the other first reported in 2013 by the systematic investment fund AQR – we find that the best returns were seen in the 1980s and then the 2000s.^[2] But the past decade has delivered negative returns (Chart 2).

These negative returns have challenged systematic investors. Should they ditch FX momentum altogether? Is it a waiting game for the profits to return? Or do they need to adjust the strategy? We would argue that the most fruitful route is to adjust.

The most obvious adjustment would be to expand the universe of currencies used. Most model funds use FX futures in developed markets, and within that they focus on DXY, EUR, JPY, GBP, CHF, AUD and NZD. These are traded on exchanges like CME and ICE. While there are differences across these currencies – notably, some are commodity currencies and others not – they are still developed currencies with increasingly similar interest rates.

Given the rise of China, especially over the past decade, it would make sense to bring Asian currencies into the mix. This would bring in currencies that are China-centric rather than just US-centric and offer a wider range of fundamentals from high interest rate currencies to tech currencies. The five most liquid Asian FX futures markets would be CNH, SGD, KRW, INR and TWD. These are commonly traded on the SGX exchange.

Why Do Funds Use Futures Rather Than OTC Markets?

An obvious question is, why do model funds tend to prefer using futures rather than over-the-counter (OTC) markets for trading FX? Some of this is historical: many funds developed models using futures in other markets such as commodities or equities, and so an eco-system of model trading around futures developed. This naturally led to using futures to trade FX. Then there are practical reasons such as exchanges offering a centralised market to trade multiple markets, and transparency around pricing, fees and commissions.

But since the global financial crisis of 2008, another reason has come to the fore – changes in regulation. In essence, the thrust of post-crisis regulation has been to shift markets to central clearing, electronic execution and introducing margin requirements for non-cleared trades also known as the uncleared margin rules (UMR). In FX markets, this has most impacted FX non-deliverable forward (NDF) markets, which cover many emerging market currencies, and also FX options markets.

UMRs are forcing a fundamental shift in how investors trade NDF markets. By the end of 2020, most investors will have to post initial and variation margin with the bank with which they trade. Historically, this was not the case. What makes matters worse is that FX positions cannot be netted across banks – this means investors would end up posting much more margin than had they traded with an exchange where transactions are netted and centrally cleared.

Not only that, but banks have to hold capital against trades they take on and also post margin to other dealers. The greater use of balance sheet and worsening leverage ratios would introduce additional costs and margin requirements for banks. These would be passed on to investors. All in, some estimates suggest that, for a typical portfolio of Asia FX trades, the margin requirements of trading bilaterally with banks could be more than 2.5 times more than trading with an exchange^[3]

Therefore, there are compelling reasons for investors to continue to use FX futures, rather than the NDFs in the OTC market, to implement their systematic strategies.

Choosing Momentum Strategies

Now that we have established that FX futures will be used for any momentum strategies, we need to determine whether adding Asian currencies makes a difference to investment returns. We can construct momentum strategies for both developed and Asian currencies and then compare performance. To do this, we must first choose the type of momentum strategy.

There are broadly two types of strategy: momentum on a time series basis, which generates signals on individual currencies; and momentum on a cross-sectional basis, which ranks currencies against each other.^[4] In this paper, we opt for the time series approach because we are adding currencies to an existing set.

As for the momentum rule itself, the two most common rules are using lookback windows or moving averages:

• A lookback window is simply the % change in the currency over a recent period or window – for example, the past month or the past three months. Then, if the return is positive, you buy the futures; if the return is negative, you sell the futures.
• Moving averages are where you take the average of the currency over a recent period – say the past month, and then in one version you compare the current futures price with that average. If the price is above the average, you buy the futures; otherwise you sell.

In all types of strategies, you can also have different holding periods. For example, if you have a one-week holding period you would generate the signal once a week and hold the position until the following week’s signal.

We opt to use the lookback window strategy, which both investors and academics commonly use. We look at one-, three- and twelve-month windows to capture short-, medium- and long-term trends respectively. For holding periods, we pick 1-day, 1-week and 1-month periods.

Here are two examples of how the strategy would work:

(1) Trading short-term momentum in USD/CNH

As it is short term, we use a 1-month lookback window. For this strategy we will use a 1-day holding period, though we could use 1-week and 1-month as alternatives too. We also decide on a dollar notional size that would apply to all trades. For example, it could $1 million notionals.

For the trading strategy, the first step would be to calculate the 1-month percentage return in the SGX USD/CNH (UC) contract. Let us say that an investor is trading on 20 April – on that day, the 1-month return in USD/CNH was +0.5%; that would imply a buy signal. The investor therefore buys the active UC contract in $1 million notional.

Now, as our holding period is one day, the investor holds that position until the next day. On that day, the investor recalculates the 1-month return – now it comes to -0.5%, which triggers a sell signal. The investor therefore exits the long position and enters a short position of $1 million. That would imply sell $2 million in notional CY overall to go from long $1 million USD/CNH to short $1 million USD/CNH. This exercise would be repeated each day, and the investor would always be either long or short the future.

(2) Trading long-term momentum in INR/USD

As it is long term, we use a 12-month lookback window. For this strategy we will use a 1-week holding period, though we could use 1-day and 1-month as alternatives too. As we used $1 million notionals for USD/CNH we use the same notional equivalents for INR/USD.

Again, the first step would be to calculate the return over the lookback window – in this case it would be the 12-month percentage change in the SGX INR/USD (IU) contract. The investor is trading on 20 April – on that day, the 12-month return in INR/USD was -9.5%. That would imply a sell signal. The investor therefore sells the active IU contract in $1 million notional.

As the holding period is one week, the investor holds that position until the following week. On that day, the investor recalculates the 12-month return – it comes to -7.6%, which is still a sell signal. The investor therefore does nothing on that. This exercise would be repeated each week, and the investor would always be either long or short the future.

What Are the Returns?

We assess the trading returns for G10 and Asia FX from 2015 onwards – a period during which Asian FX futures have gained in liquidity and China has become a more dominant force in the global economy.

It is clear that momentum strategies that involve the majors, DXY, EUR/USD and JPY/USD, have struggled in recent years. We looked at nine permutations of momentum strategies for each currency. The best performing rule for DXY, 1-day holding with the 12-month lookback, delivered flat returns with a Sharpe ratio (excess returns over volatility) of zero. The best rule for EUR/USD futures delivered annual returns of 1% with a Sharpe ratio of 0.1, and the best rule for JPY/USD delivered similar 1% returns with a 0.1 Sharpe ratio (Chart 3, Table 1 and Appendix).

For G10 currencies as a whole, across all strategies and currencies, only one third delivered positive returns (see Appendix). The Sharpe ratios range from -1 (AUD, 1-month holding, 3m lookback) to +0.5 (GBP, 1-week holding, 3m lookback). The median Sharpe ratio is -0.1. In general, GBP AUD, NZD and CAD performed best, while CHF lost money on all strategies.

In contrast, Asia FX momentum strategies have delivered positive returns, with almost 60% in the black. Notably, the best strategy for USD/CNH delivered annual returns of 5% with a Sharpe of 1.2. Elsewhere, the best strategies for USD/SGD and INR/USD delivered annual returns of 2% with a Sharpe of 0.4, and the best strategy for TWD/USD delivered annual returns of 3% with a Sharpe of 0.5. KRW/USD performed less well, with the best strategy delivering annual returns of 1% with a Sharpe of 0.1. The best rules are shown in Table 1.

Adding it All Together

Starting with just the G10 currencies, if we combine them all into one equally weighted portfolio, we find it delivers a median Sharpe ratio of -0.3 across all strategies. The best strategy, 1m holding with a 12m lookback, delivers a Sharpe of -0.1. Meanwhile, an equally weighted Asia FX portfolio delivers a median Sharpe ratio of 0.1, and the best strategy, 1d holding with 1m lookback, delivers a Sharpe of 0.3 (chart 12 in Appendix). These numbers reflect that the strategy returns of individual Asia FX are stronger than G10, and so any portfolio of Asia FX would likely outperform G10.

But there are ways to optimise the returns. Of course, this runs the risk of data mining; but nevertheless it would give a sense of how to improve returns. The most obvious optimisation would be to use the best rules for each currency and create ‘G10 optimised’ and ‘Asia optimised’ portfolios.

We use the rules for each currency as shown in Table 1. We find that the G10 optimised portfolio delivers annual returns of 2% with a Sharpe ratio of 0.5. The Asia optimised portfolio delivers annual returns of 3% and with lower volatility, so the Sharpe ratio jumps to an impressive 0.8.

If we show the returns over time, G10 FX momentum returns have gone sideways since early 2018 even with optimised trading rules (Charts 4 and 5). Meanwhile, Asia FX momentum returns have continued to deliver steady returns since 2015 and 2018. Therefore, adding Asia FX futures to existing FX momentum strategies would add significant value.

Bottom Line

G10 FX momentum strategies have struggled over the past five to ten years. Some of this could simply be cyclical – any strategy could have a poor run – but some of it could be due to a lack of diversification of currencies in a more globalised world. Asian FX futures are the obvious candidates to add to such strategies. They are liquid and are tied to the China cycle, which could bring different characteristics to G10 currencies. We find that, using a wide range of momentum rules, Asian FX futures appear to show trending behaviour and as a result are profitable in momentum strategies. We further find that adding Asian currencies to existing G10 FX momentum strategies improves returns. Therefore, an important part of rehabilitating FX momentum strategies would be to include Asian currencies.

Appendix

1. Hafeez, “Benchmarking Currencies” (2007); Menkoff et al. “Currency Momentum Strategies” (2012). ↑

2. Hafeez, “Benchmarking Currencies” (2007); Asness et al. “Value and Momentum Everywhere” (2013). These papers were selected as they are transparent on the trading rules used and they use long back-testing periods. The former piece also was one of the first to be associated with a tradeable FX momentum index – so the returns since publication are truly out-of-sample. The latter has been publishing trading returns of the models regularly since publication. ↑

3. Collateral Damage – How Uncleared Margin Rules Will Revolutionise the FXPB Model, Citi (2019) ↑

4. See Hafeez (2007) and Menkoff et al. (2012) ↑