Algorithm usage in FX trade execution has grown over the past several years and the majority of market participants realize that in order to achieve best execution in this fragmented and fast moving market, execution algorithms have become a necessity.
The spot market and increasingly NDFs lend itself well to algo usage, because of liquidity fragmentation, the evolving functionality offered by venues, rapid quote updates and regulatory requirements around monitoring execution quality all of which make trading by voice often suboptimal. As a result, many traders are looking for execution algos that cover a wide breadth of their trading needs which means much more than just TWAP or an urgent style algo which were the initial FX algos – initially easy to build and support.
However, this picture is changing quickly. First, as with any technological trend, one expects sophistication to increase over time. Both sell and buy-side participants have begun demanding increased visibility and transparency as well as increased trading sophistication in how the algos trade. Technology holds immense promise in this regard and commoditized algo services do not always cater to these demands.
Second, regulatory changes have affected how banks and institutions look at their trading operations. For instance, Uncleared Margin Rules (UMR) have changed the economics of a trading operation. Cost minimization is all-important, while digitization has increased exponentially. Many of the regulatory requirements are impossible to satisfy without widespread technological adoption.
Danielle Caravetta, Head of Sales at Pragma, details how firms can realize greater cost efficiency by upgrading their technology via the private or white label route. “Outsourcing can be offered at a fraction of the cost of building internally, both in the initial platform spend and the ongoing maintenance costs,” she says. “This includes both the cost of the FX algo platform itself and the tools required to support and maintain an institutional grade platform.” She lists real-time monitoring tools, trading databases, and post trade TCA reports as examples of the complex infrastructure needed.
Greater tech adoption has spurred efforts to increase efficiency in trading operations, and execution is in the spotlight. Transaction Cost analysis (TCA) is in high demand, despite the FX market not lending itself well to such analysis because of the lack of a consolidated tape. Determining the market impact of an FX trade when there isn’t a central tape from which to compare the execution price, has been a significant challenge.
However, this hasn’t dampened participant demand. This demand has been high enough for the Global Foreign Exchange Committee (GFXC) to review its Code and publish guidance regarding algo usage disclosures and TCA templates. Standardizing these disclosures has gone a long way towards helping institutions evaluate their execution through TCA adoption which in turn has also increased algo execution demand as algorithmic trading lends itself well to TCA since both are driven off of numerous data points all stored in easily accessible databases. Participants now have greater confidence in their ability to evaluate execution, which has spurred them to further explore the execution performance advantages algos provide.
However, as traders have now been using execution algorithms in spot and NDFs for a number of years, their level of sophistication has grown and thereby their expectations and demands for what is required from an execution algo. Additionally, in order to provide what traders are looking for, the technical requirements have grown as well.
Adopting algo execution requires firms to upgrade their technological infrastructure and provide resources to monitor everything from daily operations to security. These are over and above the quant resources needed to monitor algo execution and efficiency.
Caravetta lists an infrastructural example that illustrates the difficulty many firms face. “Real-Time UIs are a very expensive and specialized tool to provide,” she notes. “They must provide viewing functionality, real-time analytics, the ability to interact with orders and alerting features so that the coverage desk supporting the algorithms can provide proactive support to their institutional clients.” When faced with the costs of such infrastructure, institutions will naturally wonder about the ROI building algo execution capabilities in-house can bring. In the case of smaller business lines like NDFs, the cost of adoption might be unjustified. Similarly, smaller buy-side firms with low volumes will find such investment impractical.
This is where outsourced solutions enter the picture.
White and Private labeled solutions offer efficient sophistication
While the costs of embracing technical sophistication are high, it doesn’t mean firms ought to sit on the sidelines and let important trends pass them by. Outsourcing offers a cost-effective solution for firms looking to embrace technology and modernize their FX trade execution.
“By outsourcing a bank or fund can be fully live in three to six months with an institutional grade platform, as opposed to the years it may take to build internally,” says Caravetta. “The time to market can be as little as three months, with a production ready, client facing offering that includes a full suite of customizable algorithms and tools required to support an institutional business.” A major reason for this is the expertise white label providers bring from alternative markets. Most service providers are well-versed with implementing algo solutions in equity markets and carry this over to the FX world. While the markets are different, critical back-end infrastructure remains largely the same.
“Banks, ranging from global to regional, and large hedge funds are candidates to outsource their FX algo trading requirements,”
Danielle Caravetta
Caravetta explains, “If you examine the equities and futures markets, outside of the top 5 global bulge bracket broker-dealers, the vast majority of participants outsource their solutions.” This aids a service provider’s ability to implement algos for FX desks. “Providing an algorithmic trading service that can be viewed as high quality by traders at the largest money managers or banks in the world requires a huge amount of expertise,” she says. “Market microstructure, managing low-latency software, and network connectivity experience is carried over to FX implementation.”
Caravetta is quick to stress the differences between a white label solution and a private label one. On the surface, market participants might think they’re the same, but this isn’t the case. The typical white label solution comes bundled with liquidity pools offered by the provider. For instance, a broker might offer a white label solution of their platform, backed primarily only by their liquidity pools.
In such cases, clients cannot create custom liquidity pools, and conflicts of interest might arise. In addition, a white label solution does not provide the degree of sophistication that an institutional or corporate client might need and it doesn’t allow a bank to inject its own intellectual property into the execution algos the way a private label service does. For example, firms that leverage private label solutions often have specific ways they want the algos to route and interact with their own internal liquidity which isn’t possible in an off-the-shelf white label solution.
The breadth of algo behavior customization is also greater with a private label, with regards to what order types are used and at what pace the algos trade at. These configurations can be done at a granular level for each specific client a private label client has and even for each trader at the private label’s client. In short, private labeling offers a custom, low-touch service that will meet a client’s sophisticated requirements.
Outsourced algorithmic trading services can thus offer cost-effective solutions due to the scale of services they provide and the markets they service. Compared to an institution that requires a complex infrastructure and highly specialized software to serve a few desks, service providers spread costs across multiple clients.
Typically, private label platforms lend themselves very well to customization and even offer quantitative research services to support their clients with questions or requests they may get from their large, sophisticated institutional clients.
Adopting an outsourced service provider also provides clients instant access to upgrades and new features that can expand lines of business for which they would never have the time or resources to build in-house. For instance, algo usage in NDFs has been rising as more liquidity providers stream NDF pricing, but it is not automatic that an algo built for spot can immediately start trading NDFs. Adjustments need to be made to the algo to support NDF market data as well as take into account the difference in liquidity sources relative to spot. “By working with the right private-label provider, mainly an independent specialist in the FX algo trading space with a proven track record of developing high performing institutional grade algorithms, a bank or fund can maintain the same benefits (anonymity, independence, customization, private branding) of an in-house build,” says Caravetta.
Given the range of features on offer, choosing a provider might seem like a tough task. What features does a good service provider have?
Choosing the right provider
For starters, expertise in an alternative market is always a good sign. The FX market is fragmented enough to pose complex challenges for an algo service provider. The last thing a firm needs is a service provider using it as a sandbox or proof-of-concept. Evaluating the level of customization needed is a critical factor firms must evaluate.
Every firm has different needs. If costs are the primary area of concern and a firm just needs something basic, with limited customization capabilities for a couple of clients, white label solutions can be useful. Private labeling is better suited for firms that want to insert their own IP into the algo behavior have deep liquidity curation needs and whose clients are more discerning of execution quality. Generally heavy customization is required for both the trade scheduling and order routing that an algorithm handles, coupled with the trade support tools such as real-time monitoring and interaction with orders and real-time and historical TCA which are all necessary to support a firm’s largest institutional clients. Thus, the question of whether private or white labeling is “better” largely comes down to a firm’s requirements and priorities.
Whichever option firms choose, Caravetta lists a range of criteria that firms must evaluate such as track record in implementing algorithmic trading, conflicts of interest, customization , access to liquidity, breadth of product suite , additional support tools , and institutional know-how. Regulatory knowledge plays an important role too. “Ensure the partner firm has policies in place to meet the algorithmic trading requirements under regulations such as MIFID II,” she says.
Volatility offers a good test of both an algo and the service provider. How did service fare during the previous bout of volatility, and does the service provider have backup options for such conditions?
An ideal private label partner is entirely focused on algo development and integration, not on internalization or proprietary trading, which may not be the case with a white label provider, where a potential conflict of interest may exist This ensures there will never be any conflicts of interest.. Firms should also thoroughly evaluate the platform and examine the depth of analytics provided. TCA must be viewed as standard these days, and customization concerning views and data is essential.
Caravetta stresses that independence and transparency are central pillars when evaluating a private label service provider. “We believe in the importance of partnering with a provider who is independent and does not have any conflict of interest,” she says. “This also includes providing a high degree of transparency not only in how the execution algorithms work, but also through providing clients access to their trading data so they can do their own analysis if desired.”
Conclusion
The complexity of integrating algos and the necessary infrastructure to support FX execution is high. Outsourcing provides firms, both on the buy and sell-side, the ability to enter the market quickly and less expensively than building in-house.
Best of all, the practice is in line with the trading industry’s new business economics that requires cost-effectiveness. The right partner will boost a firm’s ability to trade effectively and enhance relationships. Caravetta observes, “Unless a firm has the scale and budget, like a top five global bank, creating and then supporting an algorithmic trading service can be cost prohibitive.”
Frequent delays often lead to banks realizing this two to three years into the project. “This experience has happened in equities, futures and FX,” she says. “It sounds a lot easier than it actually is to build and support an algorithmic trading suite for the world’s largest institutional traders.”
Clearly, outsourcing is the best way forward for firms on both sides of the market.
