Despite liquidity now being recognised widely as one of the most pressing issues facing institutional FX trading firms, the challenge is not really all that new. According to Fergal Walsh, Managing Director, Global Head of Algorithmic Execution for Foreign Exchange and Local Markets at Citi, liquidity has actually been migrating away from venues for a number of years to a growing range of alternative FX venues.
“To resolve this challenge trading firms, both systematic or voice driven, spent notable tech dollars to ensure they have an efficient manner of accessing all the relevant markets where previously they may have had two simple access points on traders’ desktops,” Walsh adds. “As a result, market access via a smart order router/aggregator is now one of the most important components to all trading business, while from a client perspective, there is the additional consideration of banks internal liquidity pools.”
More than ever, clients now need to be able to understand and evaluate each bank’s internal offering in order to determine how effective it is. But those banks with larger e-trading businesses efficiently integrated into their algo offering should then in turn now be able to offer a value-add to their algo clients, Walsh argues. “For Citi, this is a key differentiator,” he explains. “The breadth and depth of our globally distributed client franchise makes our liquidity pool unique and valuable, coupled that with the sheer size of our trading and sales business, all who can use our algo platform – and you start to see how special and “differentiating” Citi’s liquidity pool can be.” He adds that the real challenge now for banks and clients alike is how to effectively manage and benchmark the various liquidity pools available, which is where he believes independent TCA can really start to come into its own.
Creating a clear view
Even so, liquidity in the FX space remains as fragmented as ever, while the number of sell side and non-bank sell side participants continues to grow, says Brad Bechtel, Global Head of FX, Managing Director at Jefferies. According to Bechtel, the real difficulty for clients lies in figuring out which are the quality venues and having the ability to interact with those venues, which can be hard for a client to achieve on their own. “The focus is shifting from the quantity of venues available to the quality of venues that clients interact with,” he explains. “But the quality of venues varies as does their liquidity profiles, so having a partner that actively manages and curates the liquidity within the venues is essential.”
While this fragmentation of the FX market has led to spread compression, it has also meant the real interest in the market has become harder to find, warns Tom Appleton, Executive Director, e-FX Algo Execution, Crédit Agricole CIB (CACIB). This in turn is makes it difficult for traders to assess price and liquidity without the use of additional tools. “For those firms that don’t have access to quality trading and liquidity tools, the result is a harder trading environment and concerns about liquidity, mirages etc,” Appleton explains.
FX algos, however, work by aggregating the fragmented liquidity channels to get a complete view of the market. “Having this clear picture of current liquidity allows the algo to compare to historical levels, adjust parameters and make predictions in microseconds,” says Appleton. “This speed allows algos to assess the market, find the liquidity and take advantage of situations that humans just cannot react to. As a result, overall execution costs have been significantly reduced by using FX algos.”
Having the right relationships
Algos also have the ability to add liquidity and capture spread by placing resting bids/offers on open portals/ECN’s, as well as interacting with their own franchise liquidity for banks, adds Ron Klipstein, Head of Algo Trading, Global Foreign Exchange, Northern Trust. In addition, he explains that as credit constraints continue to increase, finding reliable counterparties is key. “We work closely with our internal credit and network management teams to monitor these relationships from a credit perspective,” Klipstein says. “We have also invested significantly in our FX technology recently through a series of acquisitions and partnerships, all in order to meet our client’s increased dependency on best-in-class toolsets to help capture liquidity in this increasingly fragmented environment.”
In addition, a poll of attendees at this year’s TradeTech FX US found that 80% believed the liquidity the algorithm executes on was more important than the algo itself. So how important is the ability to capture liquidity to the overall effectiveness of algo strategies?
“I would not put it down to ‘capturing liquidity’ per se,” explains Walsh. “Liquidity management with good process and rigor is the critical first step; only then can you consider exposing each strategy type to the appropriately curated liquidity and achieve an optimal balance. This is really where data science meets art.”
According to Walsh, for this to be effective it also has to be accompanied with a robust, regular systematic and quantitative review of all the venues in as granular a manner as the data allows – most of which Citi performs in real-time. “Only once that basic building block is in place to ensure the right liquidity is presented to each strategy type can the quantitative work for effective order placement, or liquidity capture, kick in,” he adds. In addition to performing analysis of market microstructure, Citi is also working on predictive algorithms to dynamically select venues for each micro-placement, which in turn enables more effective use of liquidity. “Measuring and predicting the cost of liquidity has become a useful and effective area of work in this regard,” Walsh adds.
Engagement and education
Yet just as each venue in the FX space has its own strengths and weaknesses, each client also has their own different needs, adds Bechtel. It is then a matter of marrying the client’s needs to the right liquidity at the right time and then hitting the venues where they are strongest, he explains. “When using an FX algo product, it is as important to have an understanding of the liquidity behind the execution as it is to understand how the algo works,” Bechtel says. “We focus on providing both value added services.” However, he warns that managing multiple connections to various venues and the different pools of liquidity within those venues is no easy task. “Yet when liquidity management is done effectively, it can greatly enhance the ability for the algos to execute with maximum liquidity and minimal market impact,” adds Bechtel. “Jefferies is particularly good at this active management of liquidity on behalf of our customers and it is a service which they find especially valuable.”
According to Appleton, all FX venues ultimately aim to have unique value propositions, from uncorrelated pockets of liquidity to offering iceberg order types or even to offering latency floors that level the playing field across customers. However, the liquidity they offer is only as good as the counterparties they have in their pool, so there is also focus on attracting good liquidity, where ‘good’ can be measured by metrics such as fill ratios, market impact, spreads, cost of retry etc, he warns. Appleton adds: “Policing the venues to optimise these metrics allow us to ensure the quality of the liquidity we provide our clients. Many also allow us to customise liquidity with clients beyond the standard setup in order for clients to gain tailored access, depending on the type of execution they desire.”
In order to better navigate this increasingly complex liquidity environment, leading FX practitioners are keen to engage with clients about the underlying liquidity as well.
As a result, many FX algo specialists are now leveraging their knowledge of liquidity by providing consultancy services for their clients, Appleton says. “Providers are also aware of how the client aims to trade and are helping them with decisions in real time through the use of liquidity analysis tools, market insight, correlations between markets etc. At CACIB, we have recently created a liquidity dashboard that we use to help us navigate this and help provide our expertise to the client,” he adds.
Internal vs external?
The availability of liquidity also depends on the individual profile of the bank in question, argues Klipstein. “Some banks do not have much internal liquidity to offer, so they go out to the market via ECN’s and exchanges as well as other banks, while some very large banks have a franchise they can leverage to utilise internal liquidity,” he says. In fact, just a few years ago FX algo users simply expected banks like Citi to manage the inputs, including using liquidity effectively, adds Walsh. “While that expectation remains true, our clients today are a lot more versed in the market structure and liquidity fragmentation effect, as well as the impact that good or bad liquidity can have on their executions,” he says. “As a consequence, we regularly have conversations with our clients about market structure, microstructure, venue analysis, platform rules etc in order to help them to make more informed decisions. We believe this can also be helped by gaining a deeper understanding of how each ECN or CLOB behaves, as well as the work Citi has done to leverage each venue effectively.”
In many cases, clients may also require clear and transparent explanations of how and when internal liquidity is used, including when it is preferenced over external liquidity. For example, according to Walsh, Citi’s internal liquidity pool is an obvious differentiating factor for both its algo platform and its principal e-trading business due to the breadth and depth of Citi’s overall franchise. “We have also recently added an internal-only crossing engine, CitiFX Cross, to provide even more opportunities to match flows at mid, reducing execution cost and impact even further. Notwithstanding the benefits we see from internalisation, we also give options to include exclude liquidity if clients so wish,” Walsh explains.
Internal liquidity ultimately comes in two forms, either as a true mid-market match that offsets one customer’s interest vs. another, or as execution against skewed pricing based upon what is happening within a liquidity provider’s customer franchise. According to Bechtel, a true mid-market match is extremely beneficial to clients, but skewing pricing can be both positive and negative. “External liquidity also has its pros and cons,” he explains. “It all depends on the access there is to that liquidity and how that liquidity and the connections are managed.” Yet while market makers are very good at trading in these pools and maximising spread earned for themselves, customers should have access to the exact same tools in order to participate on a level playing field when engaging with external liquidity pools, Bechtel argues.
The element of choice
Appleton agrees, adding that FX algo providers are always trying to find an edge for their clients. “Once you have a certain maturity in a category of algos, then it is actually liquidity access that provides this advantage,” he explains. For example, within the category of opportunistic algos many may look very similar, however performance may vary greatly due to the quality of liquidity being accessed, Appleton says. “As a result, we believe that allowing clients to dynamically select different liquidity pools can give them an edge when used with a partner algo provider who is also actively engaged in policing those sources,” he adds.
In order to more effectively navigate the changing FX liquidity environment in general, FX providers are now exploring the potential of new technologies in order to create the next generation of algos for clients. One of the most interesting innovations in this space is the application of AI in FX algo development.
Appleton explains that in practical terms, this will create the ability for the algo to quickly adapt to changing market conditions. “However, some of the more state-of-the-art AI concepts, such as Deep Learning, result in systems that are essentially a black box where any trading decisions are very difficult to reason about,” he warns. “Instead, the most effective FX leaders are taking those AI concepts of adaptability and applying them to specific aspects that can be reasoned about, then also employing sound liquidity management processes, consulting with clients on what is best for them, giving access to pre-trade tools while maintaining a suite of algorithms that offers proven reliability and effectiveness.”
Algos are also becoming smarter in the way they approach thinner liquidity environments, Klipstein says. He explains that new types of algos can offer increased functionality and workflows or the ability to algorithmically trade a new set of instruments, such as NDF’s. While some clients may also wish to modify the liquidity pools being used, this ability really depends on the type of liquidity the client is accessing and whether it is an agency or principal trading relationship.
Clients may also be worried about signalling risk to the market and prefer to remove certain exchanges or providers to mitigate that risk, according to Klipstein. “However, although there may be slightly nuanced ways of looking at FX liquidity, by and large there is nothing new under the sun,” he adds. “We believe instead that well educated clients are the best clients - and we do our best to provide those clients with a clear picture of the prevailing liquidity in the market and how we interact with it.”
Adding innovation to algos
Some clients may also wish to modify liquidity pools in order to manage the effect of last look and its impact on their orderflows, Walsh adds. “We provide the ability to remove ECNs, leaving in place only ‘no-last-look’ CLOBs. However, because of our liquidity management process, few clients choose this option,” he notes. According to Walsh, the major providers of FX algos are instead focusing on engineering predictive algorithms which can dynamically select venues for each micro-placement, enabling the algos to seek out liquidity wherever the opportunity arises.
“The leading providers, including Citi, are also focused on helping clients to measure and better predict the cost of liquidity and using that as input for algo execution,” he says. “Next generation algo development also includes building smarter strategies for accessing hidden, or iceberg, liquidity, coupled with greater internalisation of algos and our internal liquidity.”
“Ultimately, the goal here should be to provide clients with the best possible tools to interact with the best possible liquidity, maximising liquidity while minimising market impact,” Bechtel adds. Banks have always been good at engaging with liquidity in order to maximise the amount of spread they earn on client volumes, Bechtel continues, with bank clients having access to the best-in-class tools available in order to capture spread.
“At Jefferies, we fundamentally believe that all clients should have the same exact edge and access to this high standard of market-leading tools,” he argues. “The next generation algos that we have developed are centred around this core belief, providing the client with the same edge that only the clients of tier-one banks have traditionally enjoyed.” Such tools are of particular benefit to clients who want to take control of their order and not outsource the risk via an RFQ or voice trade.
“Yet for a lot of firms on the street, FX algo execution is still a relatively new concept,” Bechtel continues. “But as buy-side participants increasingly learn about the benefits of using algos and become more sophisticated users, they are now beginning to ask the right questions - not just about algo execution, but also about the liquidity available. This, coupled with having access to best-in-class toolsets, is essential to navigating today’s increasingly fragmented FX liquidity landscape.”