FX Algo trading in Asia – setting out on a voyage of discovery

The use of algorithmic FX trading in the Asia Pacific FX markets may still be relatively nascent compared to the uptake in Europe and the US, but the potential for growth is significant. According to a study of the region’s FX markets by KPMG China and the GFMA, the increased use of algos for FX orders are one of the key trends taking place in the region, with international banks expected to migrate more order types to algo execution on their SDPs to improve surveillance and eliminate potential conflicts of interest in the future. Yet how significant is demand from Asia’s buy-side clients for FX algo and TCA toolsets and what should banks be doing to tailor their services to the region’s unique trading conditions? Nicola Tavendale investigates.


“If we look at the growth in FX algos globally, then clearly Asia is starting from a much lower base in terms of penetration usage of algos,” says Ralf Donner, Executive Director and Global Head of FX Client Algo Execution at Goldman Sachs. “But even though clients in Asia may be starting from a lower base, the growth in algo usage is also more rapid than for any other region. As a business, we are very focused on Asia and are seeing a lot of interest from the region for algos.” The motivation for clients in Asia to use FX algos is, however, very much the same as in other markets, Donner adds. Typically clients will use algos in order to reduce their existing costs of execution and add to their execution toolkit. According to Donner, in some cases these clients were not trading electronically at all until recently, but now that they have gone electronic they are discovering that in addition to risk transfer, which is still a very valuable source of execution for them, algos can also make another useful execution tool.

Ralf Donner
Ralf Donner“The standard algo products which are being used with a high degree of success in Europe or America will need to be recalibrated or adjusted for the liquidity conditions during Asia hours.”

Allan Guild, Head of Alternative Execution Services, Foreign Exchange & Commodities at HSBC also notes a marked upturn in interest for algorithmic FX trading in what is already a key region for the bank. According to Guild, FX algos particularly come into play for clients when they have significant amounts of foreign exchange to trade. The exporter community in China in particular, but across the region in general, is increasingly interested in having these conversations with the bank about how to manage their foreign exchange risk, he adds. “We have a number of very good client relationships based on our footprints in the region and certainly FX algos are an increasing part of those conversations with a range of those clients,” Guild explains. “Asia is a really exciting place for our foreign exchange business because all of our clients, it doesn’t really matter what client segment you are looking at, need to undertake foreign exchange transactions. It’s driven by the amount of cross-border activity that goes on in the region in terms of trade investment, which in turn drives the need to do foreign exchange and the curiosity and interest in different forms of execution.”

Increasing electronification

In addition, the region notably boasts a large and diverse community of buy-side players, with the top end of the asset management and corporate space tending to need that high degree of cross border activity which therefore involves foreign exchange. Even if the client is an asset managers investing in equities or bonds, or a corporate focused on manufacturing, the degree and size of the foreign exchange flows, which are naturally part of the Asian macro economy results in their interest in HSBC’s range of algo products, explains Guild.

Allan Guild
Allan Guild“Many will begin with their TWAP algo in terms of getting a first taste of using an algo product.”

However, Asia’s FX markets are still markedly fragmented, with trading taking place in different venues using different technologies and non-standardised data, warns Leo Tay, Managing Director, Head of FX Trading Asia at ING Bank. FX algo adoption has also resulted in smaller participation by human traders, he argues, with more flash crashes seen during illiquid periods such as during Japanese holidays or the Asian open. “Algo adoption has also, for obvious reasons, been more pronounced on the more liquid Asian FX pairs as compared to other instruments, including non-deliverable forwards,” Tay explains. “To develop a robust algorithm, we do need greater levels of market standardisation, ie by adopting FIX protocols and working with regulators globally to unify a fragmented market.”

Levels of electronic execution are also expected to rise further in response to improved algo trading, as well as requirements for greater price transparency and best execution, according to the findings of the KPMG China and GFMA report on FX trading in the region. “This will present opportunities to enhance price discovery and efficiency in onshore markets that could benefit local economies,” the report adds. According to Jonathan Woodward, Regional Head of Transaction Sales, Asia Pacific at Refinitiv, the issues surrounding fixings over the last few years have also prompted the whole FX industry, Asia included, to address these concerns with the creation of agency desks that are completely separate from the existing teams. “The agency desks offer tools to enable clients to get as close as possible to the fix through the use of algos,” he adds. Woodward agrees there has been a significant increase in the adoption of algos in FX in Asia over the last few years and notes that the biggest uptick in the usage of bank provided algos has been among the large real money managers. However, hedge fund activity in FX has also increased over the last few years, either as part of a macro strategy or a dedicated FX fund, and most will use in-house built algos, he says.

In terms of algo strategies being deployed, this depends on the client’s objectives just as it does in the US or European markets, Woodward adds. The most common algo in Asia is TWAP, which helps clients clear larger orders automatically over a period of time. But if the client is not trying to replicate a fix and the firm has access to a central limit order book (CLOB), then there is an opportunity to participate in the market and not cross the spread, he explains. “There are many ways to do this, and in Refinitiv’s FXT the trader can peg to the bid/offer for a defined time period and then, if not executed in that time, aggress the offer/bid and before replenishing the order,” Woodward says. “This process is repeated until the order is completed.”

I BIS. Includes only transactions with G10 currencies on both sides of the transaction

I  BIS. Includes only transactions with G10 currencies on both sides of the transaction

Unique trading landscape

Banks should also be aware that the Asian markets have a lot of unique characteristics which have typically been overlooked in the past, warns Donner. “When you start focusing on the region, you begin to realise that it does require its own approach,” he explains. “For instance, offering algos on a basket of currencies. If you are a Japanese client, then pretty much everything in your basket of currency pairs is a synthetic execution. So if you are trading all these different yen pairs there is naturally a great netting opportunity between the various legs that you need to execute. It suggests a very different type of execution from how you might otherwise proceed.” Goldman Sachs has also noted strong levels of interest for Asian NDFs, he adds. “Because many Asian currency pairs are not deliverable, to have an attractive algo offering in Asia you really need to have NDFs. That is critical to having a complete algo offering,” Donner says.

A further important consideration is liquidity. In many cases, the liquidity conditions during Asia hours on even the top ten Asian NDF pairs are worse than they are than in the rest of the world. According to Donner, this means that the standard algo products which are being used with a high degree of success in Europe or America will need to be recalibrated or adjusted for the liquidity conditions during Asia hours.

“This is key to having a successful offering for Asia clients,” he adds. The bank has also had clients who have tried to build their own algo solutions in-house by clipping a larger trade into smaller chunks and then farming these orders out to a set of liquidity providers. However, Donner notes that the bank is careful to point out that the mark-out characteristics of this kind of execution can be quite poor compared to a genuine algo offering. “But that is not a particularly Asian phenomenon,” he explains. “We also educate our clients in Europe and the US and point to some of the dangers of clipping trades and what that does for execution quality.” Woodward agrees, adding that most buy-side clients will use bank provided algos because they do not have the skills or the technology to build their own. Some of the specialised FX or macro houses are also using use their ability to continually change and develop their execution strategies to differentiate their funds’ performance, he adds. “To be successful in this arena, you have to invest in your people and technology and be on the leading edge in technology adoption eg.

Leo Tay
Leo Tay“Algo adoption has been more pronounced on the more liquid Asian FX pairs as compared to other instruments, including non-deliverable forwards.”

AI, Machine Learning etc. The execution team also needs to have an in-depth understanding of the market and all the factors that affect the way a market behaves,” says Woodward. FX algos also provide banks with an important opportunity to help their clients achieve their trading objectives and to add value to the client relationship though educating the client about the different strategies that they offer. However, one common pitfall can be that the strategies are over complicated for the client’s needs or they may misunderstand the effects of some strategies, warns Woodward.

Role of education

From the banks perspective, algos enable them to automate the flow and remove human intervention. This in turn reduces the banks costs and gives them the ability to scale the service, explains Woodard, which can help ease post-trade processes and makes the provision of trading analysis easier. According to Guild, the most important focus for a bank serving clients in the region should be on providing high quality, innovative solutions but which are presented in a clear and easily accessible way. HSBC offers four principal algo strategies – including VWAP and TWAP.

“Our clients like having this choice and the simplicity of the choice,” Guild explains.“ Many will begin with their TWAP algo in terms of getting a first taste of using an algo product. Some will naturally move on to an implementation shortfall or liquidity seeking mindset product. Our approach is to keep the offering relatively simple in terms of those choices, but also to continue to make that investment in terms of quantitative research and modelling. This makes sure we are right at the cutting edge to make sure the execution outcome for our clients, whichever of those options they chose, is market-leading.”

TCA is also an important part of HSBC’s offering to its Asia clients and is a big part of its service proposition, adds Guild. However, the bank is also careful not to overwhelm its clients with new benchmarks or ways of looking at their execution, particularly if they are not necessarily going to be impactful and meaningful for them. Guild explains: “Instead, it’s important to have good visualisation. We also see a drive from clients for independence and we are comfortable with enabling our clients to use third-party providers, such as our agreement with BestX. But ultimately what a client is looking for in TCA is an understandable and easy to access presentation which breaks down their order into the individual child trades.”

Jonathan Woodward
Jonathan Woodward“Simply automating the FX trade off the back of the primary asset-class execution is also not necessarily the best plan.”

Banks also have a key role to play in the education and awareness of their clients around the various benefits of different types of execution, including algorithmic. Tay explains that a good FX algo should be able to clear decent clips of FX with minimal market impact. “In addition, FX algos also give clients direct market access by not having to cross bid-ask spreads,” he adds. “This reduces transactions costs from algo users point of view.” However, algos also depend on “clean” data, and “clean” data depends on market liquidity, warns Tay. “Asian currencies are generally less liquid and trading venues more fragmented. Unless market liquidity in select Asian FX markets improves dramatically, the usefulness – and reliability – of algos will be questioned,” he adds.

Outlook for growth

As not many Asian buy sides have a full multi-asset algorithmic approach, there is also currently limited demand for a wider spectrum of FX execution algos from banks, explains Woodward.

“Most of the hedge funds leave their FX with the prime broker or custodian, which is an interesting decision given the intense focus on the quality of execution of the primary asset-class,” he adds. “However, simply automating the FX trade off the back of the primary asset-class execution is also not necessarily the best plan.

There must be a clear execution policy that can be communicated to interested parties.” Yet the regulatory drive in recent years for firms to demonstrate best execution has also served to increase awareness and is encouraging clients to begin having these discussions with their market makers, says Guild. The implementation of MiFID II, for example, created a notable uptick in algo usage, with algo based execution and the additional transparency and benefits of transaction cost analysis increasingly being recognised by clients as a way of demonstrating their compliance. “As we see best execution regulations being discussed and coming to the market in Asia, that leads to similar conversations in the region as well,” he says.

Furthermore, the introduction of the FX Global Code in the region is also expected to improved controls over algo development, deployment and testing, according to the KPMG China/GFMA report. Guild agrees, adding that its introduction has further increased the level of conversation between market makers and their clients around what their choices are.

FX ADV by product BIS, volumes adjusted for local and cross-border inter-dealer double counting

FX ADV by product
BIS, volumes adjusted for local and cross-border inter-dealer double counting

“What the Global Code has done, which we think is a very good thing, is to encourage more of those conversations and it is natural that as more of those conversations are happening more clients will take an interest in some of the new elements of that product range, such as our algo offering,” he adds.

According to Donner, this is yet another key reason why the leading banks are noticing more interest in algos from the Asia region. In addition, he notes that the liquidity profiles of many leading Asian currencies have also significantly improved in recent years. For instance, the amount of USD/CNH that is now traded electronically no longer looks like an emerging market at all, but instead now looks like a completely developed market, according to Donner.

“If we look at the top 10 traded currency pairs, then USD/CNH is within the top five. It is hard to imagine what a change that from only a few years ago,” he explains. Even a currency pair like USD/THB has been known to make it into the top 20 most traded pairs in recent times. In NDFs, Donner explains that the USD/INR and USD/KRW are also both relatively liquid, or at least more liquid than some of the less liquid emerging markets deliverable pairs available. As a result, he expects to see the growth in demand from Asia for FX algo execution continue apace. “We are moving to a regime where the market is becoming much more electronic and although the role of algos in that is yet to be determined, I would expect it to continue to boom,” he concludes.

Related Posts

Scroll to Top