FX Liquidity Management: Addressing the challenges of a seriously complex undertaking.

FX Liquidity Management continues to be a challenging and complex task as more trading styles, execution venues, platforms, liquidity providers and market makers enter the market. This is not a new development. A cottage industry of liquidity management providers and services has emerged as a result, designed to give market participants a chance to keep up. The question is whether the solution providers are able to keep up and where both the technology and trends in liquidity are heading.

“FX Liquidity Management is likely to remain complex but new technology is helping more participants to access solutions that give them a competitive advantage as they manage access to the largest and most electronic financial market,” says Stephen Totten, director of quantitative analysis at oneZero.

Artificial intelligence

He highlights the potential impact of artificial intelligence (AI). “AI is set to transform a sector that is already benefiting from sophisticated execution algorithms and increasingly quantitative analysis for both pre- and post-trade liquidity management, for example. And traditionally voice areas in FX, such as NDFs and swaps, are also seeing a huge amount of innovation from new entrants and novel offerings on existing platforms. That makes access to a neutral third-party technology partner with the ability to help manage a range of products and liquidity at scale in highly volatile FX markets more important than ever,” says Totten.

At the same time, the role of the FX liquidity manager has evolved, says Totten. “It is no longer enough for FX liquidity managers to simply quote tight spreads – which is hard enough anyway in the volatile currency markets seen since the end of the zero rates era. Skew sensitivity, skew leakage and analysis of market impact are all vital factors to evaluate for both makers and takers to achieve the best execution.”

oneZero has developed a range of tools to measure skew effectiveness and automatically identify clients with high market impact to help liquidity managers allocate their time and resources to the areas where they can make the largest impact, underlining why technology has become so important to the FX Liquidity Management process, says Totten.

“The sheer volume of market data and the growing number of market participants mean that it is now impossible to function without advanced analytics. At oneZero we are currently seeing 2.5x the data that we were just 18 months ago, for example. We regularly handle over 10 million transactions a day and averaged more than 9 trillion of quotes a month last year and data volumes are set to rise further,” he says.

“Some market players are struggling with this, so having a technology stack that can handle scale is absolutely crucial. Apart from regulatory needs that are non-negotiable, liquidity managers can also use new analytical tools to adjust to market conditions more quickly, which drives customer revenues and delivers quantifiable dollar benefits. Data analytical tools are already the only way to tackle most of the key challenges facing FX liquidity managers. And AI is set to speed the pace of progress in helping technology solutions evolve to give liquidity managers an advantage over their competitors,” says Totten.

It is also useful for FX trading firms to try to audit their existing FX Liquidity Management activities, says Totten. “Every FX firm needs to audit its liquidity management outcomes and plan for improvement. If an ECN or a bank price maker has a key goal such as winning a higher wallet share or reducing market impact, then that has to be measured, for example. And if the goal is not met then the liquidity manager needs to understand what has happened and why.”

Despite all the talk of technology, relationships are vital in FX Liquidity Management and Totten believes that the right use of technology can improve rather than detract form that relationship. “Technology doesn’t displace relationships in FX Liquidity Management, it enhances their efficiency. Faster transmission of data, interactive feedback on new liquidity pools or order types and greater use of quantitative metrics all help FX liquidity managers to do their job better.”

Technology will also be a critical component in the next generation of FX Liquidity Management tools and services, says Totten. “AI is coming to FX Liquidity Management, and at oneZero we are finding new ways to uncover patterns and improve efficiency for our clients. We also already have a tool called Maker Pool Replay that is proving popular with our clients. By combining our own data with the trading history of our clients we let them see what would happen to their PnL and trading statistics if they added or removed a specific pricing stream. Simulating orders against alternative liquidity pools can therefore be used to optimize the liquidity supply chain and improve decisions,” says Totten. “Because of the sensitivity of trade data this is another area where a neutral third-party technology partner can add enormous value – and help FX liquidity managers to exploit the coming boom in AI solutions.”

Fully disclosed trading

Historically, liquidity management was viewed as an undertaking exclusively of anonymous ECN trading platforms with the onus on the ECN liquidity managers to effectively match different participants flow profiles together, says Henry Durrant, head of business development and liquidity management at Reactive Markets. “In recent years, the FX market has seen a seismic shift away from anonymous ECN trading to fully disclosed, bilateral trading setups ultimately with the same participants. This shift has been driven by an increasingly data driven approach to profit and loss generation and an industry-wide recognition of the risks of information leakage, both through the distribution of pricing skews for streaming LPs and signalling risk resulting from passive order placement on order books. Trading via disclosed relationships removes a lot of this uncertainty and risk as well as providing the ability for a fully bespoke trading setup between the client and LP.”

With this growth in disclosed trading the challenge for clients then becomes finding a scalable technology solution to access all these trading relationships in one place, normalising not only the trading connectivity but also a single source of data and analytics for effective liquidity management, says Durrant. “This has created an opportunity for service providers like Reactive Markets to reduce complexity for clients and in turn enhance the trading relationship between LP and client through actionable liquidity management data and analytics.”

Despite the evolution of the liquidity manager’s role, the fundamental goals remain the same – to optimise the liquidity pool to source the best liquidity and improve execution quality across the full range of instruments, time zones and market conditions, says Durrant. “How a liquidity manager performs this role, however, has evolved from a legacy manual and often subjective approach to a fully quantitative and automated one. The role now demands a level of proficiency in data science combined with the market knowledge and context to apply this and effectively communicate with clients,” he says.

“The role of the liquidity manager within our clients has also evolved, where we observe clients taking a much more active role in analysing their liquidity pool and ensuring each of their liquidity relationships are adding meaningful value to their stack. Traditionally this engagement would have been in the form of a static point in time review of the last month or quarter’s performance across a series of high level metrics. What we are seeing now is a more dynamic, involved and granular process with clients constantly evaluating trends and changes in their liquidity pool, taking action as frequently as intraday if required.”

The liquidity management process has also become increasingly data-driven and quantitative and therefore more reliant on technology, says Durrant. “Technology is simply the only way to make this happen. Specialist data platforms need to ingest and store billions of data points per day. Quantitative models need to sample this data and distilled down into useful, digestible business information. Ultimately, technology does the heavy lifting and allows the liquidity manager to spend their time analysing results, drawing conclusions and planning actions rather than performing the analysis itself.”

A good data and analytics platform will capture data from every liquidity source and provide unique insights into the behaviour of each, says Durrant. “It will slice this data across any axis that the liquidity manager desires providing the ability to understand exactly how each instrument and liquidity provider performs at any time of the day and under different markets conditions. It will clearly identify trends in the data in a timely manner so liquidity management is no longer consigned to a quarterly review but can instead have daily, or even real-time, visibility. The benefit of this will clearly differ from company to company but we have clients who have saved millions of dollars on execution costs by optimising liquidity so the benefits are very clear.”

“Such platforms are not cheap or easy to build and maintain. Fortunately, the advances in cloud infrastructure for elastic compute and storage along with the availability of cheaper time-series databases have made building and accessing data platforms easier than the once were. Building a full in-house liquidity management platform, however, is still a major undertaking that should not be taken lightly. There are some excellent third party providers in this space and many execution platforms offer such a service. Liquidity management is central to the offering at Reactive Markets and we have invested heavily in this technology for our clients.”

Effective liquidity management also requires the ability to quickly action decisions in order to change the shape of the liquidity pool, says Durrant. “Consequently, a firm’s technical connectivity strategy must be agile enough to support these changing requirements. Similarly, platforms that provide centralised access to liquidity must be flexible and scalable enough to support bespoke pricing for each client and allow the LPs to provide a tailored pricing stream to meet their requirements.”

Important metrics

Clients are becoming more selective and deliberate about how they construct a pool of LPs, focusing less on the explicit number of LPs and more on the value each LP brings to the table, says Durrant. “Metrics such as market impact and rejection cost provide insights on overall cost of execution across counterparties, whilst pricing insights such as the percentage of time at top of book and spread comparisons versus a benchmark give clients a clearer picture of the current and potential value of these relationships.”

This selectiveness also makes the auditing of FX Liquidity Management a more useful process, provided firms know who to achieve this benefit, says Durrant. “The benefit of an active liquidity management process should be clear. In its absence pricing quality and performance will often deteriorate over time. The hidden costs of trading in the spread and costs of rejects will inevitably increase and reduce execution performance. Opportunities to include specific LPs that can add value to areas of the liquidity pool may be missed.”

Once a firm has decided that liquidity management is valuable, it is important to put in place a consistent process that is performed regularly, says Durrant. “They should have a clear idea on the design of their liquidity pool including the numbers and types of LPs they would like in that pool. Once they have the pool in place it is important to have a good understanding of how they will measure the performance of the LPs and the relative costs and benefits that each brings to the pool. For example, having a view of the costs of rejects can be as important as an overall spread number. Armed with the quantitative data, a regular discussion with the active LPs should be scheduled.”

In order to drive this process, the firm needs access to the data and analysis insights, says Durrant. “If they can’t access it themselves then there are a number of third party providers who can help with this function. Indeed, there is often additional benefit with going externally as the providers will often have access to information that is not publicly available or available to the firm; for example, reference mid rates that can benchmark LPs or access to LP data that the firm is not currently trading with.”

Relationships continue to be at the heart of liquidity management, says Durrant. “Building a trusting relationship with open dialogue between counterparties is key to a symbiotic pricing setup that works for all parties involved. Regular active conversations between liquidity takers and liquidity providers keeps these relationships fresh, builds this trust and gives an opportunity to align concerns.”

Companies also understand the importance of liquidity management to optimise their execution quality and costs, says Durrant. The new generations of liquidity management tools bring the data and analysis that were not readily available to the masses directly to the end customer. Aided by the elastic nature of cloud computing, these platforms are doing this in the face of increasingly market fragmentation and a massive growth in data availability, he says.

“We have already started to see a number specialist companies offering liquidity management as a service which makes accessing liquidity management insights and information much easier. These companies have access to a much broader range of data so a firm’s performance can be benchmarked against the market. This is not without its challenges, however, and the predominately bilateral nature of the FX markets make capturing and sharing of a firm’s bespoke pricing data an issue,” he says. “Of course, it is impossible to talk about data and analytics without mentioning the potential of AI to supplement and accelerate the tools of the future. Given the rapid expansion and innovation in the AI space this is bound to play an important role in the next generation FX Liquidity Management tools.”

Technology advancements

According to Guy Hopkins, chief executive and founder of FairXchange, liquidity management is becoming more complex because of ongoing fragmentation but it is balanced by the advancements in technology. “There will always be new platforms, new liquidity providers and new ways to execute but that is offset by the advent of liquidity management software that enables you to assess all of those options. We help firms to understand the impact of changing your liquidity environment,” he says. “When you add a new LP there is trading documentation and credit agreements to sign, integration work to be done and APIs to be used, and much more. It is a complex exercise that takes time and money and success cannot always be guaranteed. This should no longer be the case. It should be much easier to add and assess new LPs. This is as important for LPs or venues selling their services as it is for takers; both sides need to know whether a relationship is likely to be worth consummating”

“One change we have seen in recent years is the greater number of firms that have appointed liquidity managers to take responsibility for the process. In some firms it is a full-time dedicated role. Others have liquidity management departments. But even though it is a specific role, it can vary depending on the entity and its trading objectives,” says Hopkins.

“For example, an ECN will have a different approach to liquidity than a macro hedge fund in terms of centralised dealing desks or automation. There may even be different trading entities and styles within the same organisation. So liquidity management can be a completely different process depending on the type of organisation, and who is having the conversations with LPs.”

What is increasingly clear across the FX market, is the impact that liquidity can have on a firm, says Hopkins. “Liquidity management has become both more complex and more important. For example, there are new ways of trading (some firms have replaced sweeping with full amount execution styles) and a liquidity manager has to be able to assess the impact of that change. Sales teams at liquidity providers (both banks and non-banks) also now spend far more time talking about liquidity and execution than they used to, as opposed to the traditional model of solely discussing directional market moves and macroeconomics. This is reflected in the significant appetite on the buy-side for this type of information, the importance of which in many cases outweighs that of more traditional types of content supplied by the sell-side.”

All of these questions are unanswerable without technology; it is no longer sufficient just to be eyeballing spreads on a GUI or relying on a quarterly market share report. Liquidity management technology is an ongoing exercise, and hugely important as the counterbalance to the growing complexity of liquidity, says Hopkins.

In terms of AI, a lot of the problems with liquidity management are multi-dimensional so are very suited to machine learning (ML), says Hopkins. “There are many areas where ML can have a lot of value. However model explainability is key – the ultimate objective is to understand liquidity and to increase transparency. If you come to rely on complex ML models that are effectively black boxes as the basis for your liquidity management approach, the danger is that you are just moving the complexity from the trading process to the measurement process.”

That said, there are only so many ways that FX can be traded. Algos introduced a new way of trading but also brought more visibility into the market. And the changes to last look are good examples of how the technology has evolved. Another example of the evolution of liquidity management is the greater use of liquidity audits and using data analysis to better understand how to use and rotate a panel of liquidity providers.

When it comes to the next generation of liquidity management solutions, Hopkins identifies three phases of the process – measurement (what does my liquidity look like?); simulation (what would happen to my liquidity under certain conditions?); and optimisation (what is the best outcome?)

“The much longer-term objective will likely be to take the results of those simulations and feed them into execution systems and then automate that process and create a closed-loop circuit,” says Hopkins. “The sell-side is already making those adjustments in how they manage the liquidity they distribute to their clients, and it is inevitable that this will start to happen on the buy-side. At some stage, the solutions also have to go beyond explicit transaction cost measurement to include factors like the cost of capital or the cost of clearing so that you have a more complete picture of liquidity and the associated costs.”