By Mike O’Hara, A-Team Special Correspondent.
Daily trading volumes in global foreign exchange markets topped $1.79 trillion in November 2020, according to CLS, making FX by far the most heavily traded asset class. Most of this trading volume is executed either indirectly through electronic platforms or bilaterally between counterparties, rather than on traditional ‘exchange-type’ venues.
This fragmentation of liquidity – combined with an explosion of data as more electronic venues come to market – can make the marketplace tricky to navigate. As a result, market participants across both the buy side and the sell side are now demanding more from their technology vendors. So how are the vendors responding?
FX data challenges
The increasingly electronic nature of much of the market’s trading volume has had a wide-ranging impact on the way FX business is conducted.
“The relationship between individual liquidity providers (LPs) and their clients has evolved as E-trading has developed,” says Antony Brocksom, Global Head of Sales and New Business at FXSpotStream LLC. “When it was voice trading, clients would be on the phone with different LPs. E-trading has changed this and clients want to be able to have a view of the market across different LPs, including to meet best execution requirements.”
This ability to see and to trade on prices from multiple sources is a key requirement in the FX markets. Because Tier 1 LPs are highly selective about the counterparties they trade with, the majority of FX market participants receive executable prices either through aggregated platforms or from lower tier institutions, which act either as intermediaries (passing client flow through to their own LPs), or in a market-making capacity (taking the opposing side to their clients’ trades).
As brokers and lower tier banks can operate in either or both of these capacities, the technology they use to support their trading activities have to be able to handle the data flows accordingly, and at high volume.
“A challenge unique to FX is the OTC nature of the marketplace,” says Medan Gabbay, Chief Revenue Officer at Quod Financial. “Everyone’s data is different, and the physical way that they’re interacting with liquidity is very different. For a spot FX trade of a certain size, different customers will have a totally different experience of how that will execute, even with the same liquidity providers. And the way that each of those LP’s market-making engines will react to those executions is also different, so you get this multiplication of data. That creates both a data storage and a data analysis problem.”
Electronic data flows in FX can be complex and involve a number of steps. Traditionally, a buy side firm would initiate a trade request from a portfolio management system (PMS), which would then be routed to an order management system (OMS) for risk and credit checking before being routed to a broker or an electronic trading platform, often through an execution management system (EMS), which would typically look at available liquidity and provide the algo trading capability to split and combine orders appropriately, before routing to the appropriate counterparty. Increasingly, these three functions are being combined, particularly the OMS and EMS functions, into so-called ‘OEMS’ solutions.
The EMS component will typically receive streaming prices from a range of liquidity sources, some of which are themselves aggregated from other sources further upstream. And the increasing adoption of algo trading in FX means that trade volumes are rising, while order sizes are getting smaller. This means that platforms need to scale.
“Our platform is ready to handle more and more transactions,” says Ludovic Blanquet, Chief Product Officer at smartTrade. “Whereas firms used to trade $100 million in 10 orders of $10 million, now they trade 1,000 orders of $100,000. And that flow means the volume of data is just exploding, and we are ready to handle it.”
Of course, not all trading in the FX market is conducted electronically. Whereas spot, forwards, NDFs and swaps are generally electronically traded, large ticket or more structured trades tend to be negotiated, typically over the phone or via chat/instant messaging, which adds an extra layer of complexity.
“The buy-side has fully embraced electronic trading and as a result, trade automation is now being adopted as well,” says Christopher Matsko, Head of FX Trading Services at FactSet. “That said, there is still a large portion of real money executions going through high touch workflow scenarios such as mixed currency baskets, negotiated cross-currency trades and resting limit orders. From our experience, asset managers want to holistically capture all of their trading data points, whether it be phone, chat, RFQ or algo executions.”
Differentiating the sell side
From the sell-side perspective, banks and brokers are constantly looking for ways to differentiate themselves in what has become a heavily crowded marketplace.
Many of them operate both an ‘A Book’, where they pass their client flow through to their own LPs, making money either on commissions or spread markup, and a ‘B Book’, where they internalise client trades, often against their own positions in the market. However, running a B Book is a zero-sum game and the margins are slim.
“Some banks are trying to extract alpha from their LP’s connection but this is a risky limited reward strategy,” says Blanquet. “What most banks have to do now is to reach their client franchise with the right attractive price through as many channels as possible, smart distribution yields more profits.”
Many banks are now making wider use of data analytics, both to improve profitability when trading against LPs, and to better service their customers. Having the right tools to analyse liquidity in real time is essential in this regard.
“You want to be able to look at the performance of your liquidity providers,” says Richard Kiel, Global Head of FX at KX. “Are they meeting their commitments in terms of the quality of pricing? What’s the average spread over time? How do they rank versus each other? Are they top of book, and if so, for how long? You also want to look at client analytics, in terms of how your clients are performing; Hit rates, reject rates, and profitability often come down to pricing or performance, so you need to be able to look deeper into the data. Transaction Cost Analysis (TCA) analytics looking at mark outs, cost of reject, slippage and decay provide further insights while supporting future automated decision making.”
Another important requirement for the sell side is to be able to create custom price streams, tailored specifically to their clients’ trading styles, which can vary considerably.
“If you can, in real time, generate an analysis of things like impact hit ratio or likelihood of reject, then you can create as many streams as you want and curate those streams, skewing the price based upon the volatility measure, or widening it based upon the reject measure, very simply without having a quant team,” says Gabbay.
This is key, particularly for banks that pass client flow through to their LPs. Being able to identify toxic flow for example, and route it accordingly so that preferred LPs are not adversely affected – which can have the knock-on effect of those LPs widening their spreads, or worse, rejecting that flow altogether – can make a massive difference to the quality of the pricing that banks are able to distribute.
Liquidity providers themselves are also looking for ways to better service the market. Some are now getting together to provide various utilities to improve efficiencies and share costs. One such example is FXSpotStream LLC, a subsidiary of LiquidityMatch LLC, which is owned by a number of tier one banks.
“We’re a bank owned utility,” says Brocksom. “We’re here to allow the banks to maximise the opportunity to face their clients over a cost-effective service, provide better pricing to their clients and for clients to save costs when executing with their desired banks. And because we’re a bank owned utility, the cost of running the service is met by our LPs. From a client perspective, we’re completely free.”
Probably the most important requirement of the buy side is to find better ways to source and interact with liquidity. This has led to a growing demand for analytics tools that can help identify where liquidity is available. Vendors are responding by coming up with ever more innovative ways for firms to analyse executable liquidity and route orders accordingly.
Itiviti, for example, offers a smart order router (SOR), which sends ‘child’ orders to venues, calculated using quality parameters such as expected fill ratios, latency, and broker cost to get an adjusted route for each specific venue and currency pair. Their SOR also includes ‘double hit’ protection features to avoid hitting a venue twice within a certain timeframe, to maintain a good standing with LPs.
A number of platform vendors, including Bloomberg, now offer API Trading, which allows clients to connect directly and trade via an API. Bloomberg also offers rules-based trading, where a client may specify the trade parameter under which the order can be automatically sent for execution to a specified list of liquidity providers, thus saving time and reducing the risk of markets moving when a firm has multiple orders and multiple accounts to service.
Other vendors are looking at tools that have proven popular in the electronic equities markets, such as algo wheels for example, and adapting them for FX trading.
“We had a lot of clients approach us last year requesting, for best execution reasons, to not just send flow using one LP’s algo, they needed to randomise it in a ‘round robin’ way or include ‘weighted’ or ‘random’ selection criteria,” says John McGrath, Chief Revenue Officer at BidFX. “The algo wheel allows them to do that graphically. It operates almost like a smart order router in that it will allow the clients to trade full amount algorithmically with where they determine the best offer of liquidity.”
There is also a growing requirement from the buy side for trading platforms that can accommodate FX alongside other asset classes, particularly as many firms now operate multi-asset trading desks. Additionally, there is an increasing focus on workflow efficiency and best execution, with many investment firms now looking to integrate their PMS, OMS and EMS functions through a single technology platform.
“Technology shouldn’t be siloed,” says David Tattan, Head of European Sales at Tora. “We’re big believers in that and so everything that we do is trying to unify the workflow into one system, so that the trader can operate more efficiently. The other trend is multi-asset. Having different specialist FX, bond, equity and futures platforms isn’t ideal. So the idea is to have simplicity of workflow through a single system, with as many destinations and markets as possible, offering the best in breed UI and functionality.”
Matsko at FactSet agrees that the combined need to consolidate systems across asset classes and across business functions is a growing trend among the buy side.
“Large asset managers are consolidating,” he says. “As a result, all the disparate technology they’re using globally needs to be harmonised. And along with that harmonisation, they want to achieve trading efficiencies through automation via a single consolidated multi-asset interface.”
As with other asset classes, artificial intelligence (AI) is being increasingly adopted in FX. With recent advances in big data and machine learning, and the availability of the cloud to store and analyse large data sets, AI and machine learning are being used for a variety of purposes, from predictive price modelling, to recommendations of how and where to execute a trade.
One particular area of focus for AI in FX is around behavioural analytics, where machine-learning algorithms are used to analyse the behaviour patterns of both clients and liquidity providers under certain conditions, so that workflow can be automated accordingly.
“Artificial intelligence and machine learning capabilities can be used to look at customer behaviour in order to understand how the customers are interacting with banks, and the impact on risk management and growth strategies’” says Alexander Culiniac, Chief Operating Officer at TickTrade. “It is by and large becoming a key component in a sales trader’s toolkit, allowing them to manage a large portfolio of clients yet focusing on the existing and potential high-value clients identified by the system.”
Using AI in this way can offer significant advantages, Culiniac reckons. “Our analytics platform offers that real-time view along with historical data, which is driven by our Notification Engine, highlighting to the sales trader changes in behaviour and the customer’s profile, facilitating upselling, cross-selling, detractors and fraud-detection as a few examples.”
While the majority of global FX trading operates as an OTC cash market, exchanges around the world that list FX derivatives are also taking initiatives to reduce friction in the marketplace.
One exchange of particular interest is the Chicago Mercantile Exchange (CME), as it operates both an electronic cash market via the EBS platform and an electronic derivatives market, which runs on its Globex platform. The exchange plans to migrate the cash market from EBS to Globex this year, which means it will offer all FX products through a single technology stack, aiming to make access to the combined market more efficient for its customers.
The exchange also launched three tools last year to help clients navigate both the cash and derivatives markets.
“One is the Swap Rate Monitor, which shows the spread between cash and our futures marketplace, and how that data can be used in our markets,” says Paul Houston, Global Head of FX Products at CME. “It also shows interest rate differentials between the eight currencies of FX Link, and how investors or traders could use the CME marketplace to trade interest rate differentials versus the cash or the bond market.
“The second is the FX Vol converter,” Houston continues, “which shows thousands of FX options prices converted into an OTC volatility surface, allowing users to directly compare the CME’s FX publicly available option prices to the prices they’re getting in the OTC marketplace. The third is the FX market profile tool, which shows CME listed and EBS cash liquidity side by side, with the spreads currently available in the marketplace, and it shows how the liquidity profile changes during different times of the day. A volume weighted price shows how you can trade both markets.” All three tools are available on the CME website and the exchange is offering them free of charge, says Houston.
What’s clear is that the institutional FX technology landscape is rapidly evolving, and it can be difficult for firms to keep up with what is available. One key theme is that despite the size and volume of the global FX market, there are still many points of friction that can be reduced. But technology vendors are aware of the challenges that firms face, and are constantly taking steps to help address them.