It is not always wise to look at continents as homogenous blocks but when it comes to the adoption of FX algos across Europe, there is little regional disparity, especially when it comes to the more advanced markets of France and Germany, home to some of Europe’s largest banks and some of its most active FX algo providers.
So what are the challenges for FX banks in two of the more advanced European markets – France and Germany – as well as the technology vendors with a distribution franchise across Europe?
The adoption of FX algos in France is especially developed. Its major banks, from Credit Agricole to Societe Generale to BNP Paribas, are all active developers and providers of FX algos. The central bank, Banque de France (BdF), is also a prolific user of FX algos. In its 2017 annual report, the central bank stated that over 60% of its trades were handled electronically, 20% of which were executed via algorithms. And then in June 2018, the BdF struck a deal with Quod Financial to use its adaptive FX platform to develop algorithms for liquidity seeking and managing market impact.
“As a French bank, we’re known as strong quantitative players. This has helped us to develop strong algos and constantly innovate. The technology development at the bank is mirroring that at Silicon Valley.”
And adoption among buy-side firms is increasing, particularly among European corporates, says Asif Razaq, global head of FX Algo Execution at BNP Paribas. “We have definitely seen the use of FX algos growing across all sectors and across Europe. Last year saw a big rise in the number of new users, partly due to the pandemic. The usual trade execution channels were not available to them.”
The new adopters are using algos selectively at first, deploying them for specific use cases rather than across their portfolio. However, as they become more familiar with the technology, the adoption is likely to increase. And even though the return to normal conditions may see the rate of adoption slow initially, it is unlikely to see anyone give up their use of algos.
The fact that FX algo adoption has increased at a time when the majority of people are working from home speaks volumes about the sophistication and ease of use of the algos, says Razaq. “It will be a catalyst for further adoption,” he says.
“The algo market in general has seen new technologies brought to life. Algos are now a lot easier to use and buy-side participants are more comfortable with the technology. There is also a massive difference in the technology available from five years ago. This made it easier for buyside firms to make that leap of faith. It is now so much easier to use algos and to integrate them into your workflow,” says Razaq.
France has also developed a strong fintech culture with its banks at the forefront and helped by government support. In March, the French government announced a $4.3 billion support package for startups designed to help them find seed capital and to avoid them being acquired by larger firms.
With the advent of Covid, liquidity was a big concern and in June the government announced additional measures. The public investment bank Bpifrance launched an investment fund for key technology companies offering a number of financing options. Meanwhile France’s financial regulator, the Autorité des Marchés Financiers has launched multiple fintech initiatives over the last five years such as the Fintech Forum in July 2016.
“Demand will naturally continue to increase as providers continue to deliver new technology that enables client’s access to next generation strategies for other FX products such as NDFs, forwards, and FX swaps.”
Artificial intelligence (AI) has been a particular focus in France. There are currently more than 450 AI-based startups registered in France, many of them based on financial services. According to some French fintechs, the prevalence of AI startups is rooted in a cultural and educational heritage – from the pioneering work of French computer scientist Yann LeCun to the Grand Ecoles and their engineering graduates, many of whom have gone on to banks’ quantitative teams.
“As a French bank, we’re known as strong quantitative players,” says Razaq. “This has helped us to develop strong algos and constantly innovate. The technology development at the bank is mirroring that at Silicon Valley.”
BNP Paribas, for example has developed an AI tool, called ALiX, as part of its FX algo offering. It uses AI and natural language processing (NLP) to replicate human trading traits and react to real-time data and changes in the market. It also ‘talks’ to the client, informing users about volatility in the market or market intelligence.
“It gives the algo a voice and tells the client exactly what it is doing,” says Razaq. “It is a big step away from the black box approach of the past. Now users are getting live commentary from a chat screen. Clients can also talk back to ALiX. It can crunch data and give eurodollar trade volumes for example. Clients are able to get a feeling for the underlying market.”
The adoption of FX algos has increased as a result of the move to working from home and the launch of ALiX has especially helped, giving users market feedback in their own home, says Razaq. The move to working from home has created some compliance challenges so all messages are recorded and ALiX is accessed via Bloomberg and its app store, meaning that clients are interacting with ALiX through an approved channel.
The corporate sector has hitherto been slower to adopt algos, says Razaq. “Many of them don’t have daily orders and were happy enough with traditional execution methods. However, they are increasingly executing larger order sizes. And as adoption of FX algos increases elsewhere, there is a greater curiosity.”
There remains some trepidation when it comes to taking an autopilot approach to execution, says Razaq. “One of the fears is getting an order wrong. So we have developed a new service where we can enter the order on the clients’ behalf, using the algo and then providing transparency and feedback on each order via transaction reporting,” says Razaq.
“We have also developed a Get Me Out tool that acts like a panic button. We’ll show them a traditional risk order so they can switch back and forth. That’s helping clients become more comfortable with algos,” says Razaq. “Technology plays a big part.”
“Without a central order book, it is very expensive to get all the correct data points and to undergo all the back-testing. This will become more of an issue as the industry becomes more data-intensive.”
The appetite for algorithmic trading has increased markedly over the last five years, says Nickolas Congdon, head of electronic trading services at Commerzbank. “While FX algo usage still lags equities, the industry has recently witnessed the proliferation of FX algorithm strategies via dozens of mediums,” says Congdon.
There has also been an increase in the development of post-trade tools like transaction cost analysis, something more commonly associated with equities but has become increasingly important in FX trading as best execution demands have spread across asset classes.
A more recent driver of FX algo adoption has been the pandemic and the market volatility. “During the heightened price volatility, as a result of the pandemic, FX algorithms proved their worth,” says Congdon. “As spreads widened algorithms were instrumental in sourcing deep liquidity in what continues to be an increasingly fragmented marketplace.”
The October 2020 Bank of International Settlements (BIS) report highlighted the increase in electronic execution for spot from almost 55% of total turnover in 2010 to approximately 70% in 2019. In addition, usage of FX algos has increased significantly and BIS estimated algos account for 10–20% of daily spot volume in G10, which equates to approximately $200–400bn worth of FX spot traded via algorithms each day globally.
And as the demand for FX algos has risen across Europe, its leading banks have sought to develop their FX algo offerings. “Deep data drives electronic markets and the leading banks are jockeying for an increased market share through an army of quants working to optimise trade execution and incorporating the next generation of artificial intelligence, machine learning techniques as well as mobile and tablet trading making it more efficient for traders to execute orders outside the office,” states Congdon.
The leading European banks continue to make significant investments in their proprietary technology and use their expertise to guide their franchise,” says Congdon. “Commerzbank is uniquely placed as both a market maker and liquidity consumer and our team of in-house quant traders are continuously developing and analysing techniques to minimise our own hedging costs. This extends to our clients who exclusively leverage our extensive experience and investment in innovative technology.”
The imminent arrival of new regulations and best practice guidelines, such as the FX Global Code of Conduct, is also likely to positively impact the adoption of algorithmic FX trading across Europe and also create demand for new pre and post trade analytical toolsets.
“Several global regulatory bodies and committees continue to review and improve best practice guidelines for execution algorithms for pre trade, real-time, and post execution around both controls and order transparency,” says Congdon. “On a European level MiFID II, further defined in RTS6, introduced a comprehensive set of rules around algorithmic and high-frequency trading as well as various aspects of market microstructure that promote the transparency provided but these order types.”
“ ..the buy-side is taking more control of their FX trading and using FX as more of an asset class than a back-office process”
“Transparency is at the essence of algorithmic trading therefore it is essential that providers ensure clients are aware of how their orders are handled and interact in the market. As this continues all signs indicate that the demand for this execution style will continue to grow.” he says..
At the same time that best execution is increasing demand for TCA tools, FX algo providers have also been taking steps to move away from the black-box technology approach that typified the first waves of algo technology. “It's not the case that algo providers are attempting to obfuscate the inner workings of their robots for fear of giving away valuable intellectual property,” says Congdon.
“It’s more that bank providers need to continue to educate clients around the interworking of these strategies. Banks should ensure clients have clear, easily accessible and understandable disclosures regarding trade execution and information handling allowing them to make informed decisions about the other market participants with whom they interact,” says Congdon.
“We actively support third party anonymised peer benchmarking and continually encourage our clients to quantitatively evaluate the performance of the strategies across providers. A new initiative, championed by the Global Foreign Exchange Committee, is looking at encouraging banks to provide standardised disclosures or cover sheets on how their client execution algorithms operate which will help in demystifying these strategies,” says Congdon.
There is also growing demand among European buy-side firms for a wider range of FX execution algos. “Demand will naturally continue to increase as providers continue to deliver new technology that enables client’s access to next generation strategies for other FX products such as NDFs, forwards, and FX swaps,” says Congdon. “These developments coupled with additional regulatory headwinds and best practice guidelines to promote transparency will further contribute to demand for FX algorithms. This ongoing increase in demand is reflected in our own client algo activity, which is up 36% year on year.”
It is therefore no surprise that European banks are still the main providers of FX algos, even if the presence of independent tech providers is growing. “Execution strategies by banks still account for the majority of executions,” says Congdon.
“However, there are some tech providers that are offering turnkey solutions across several components from liquidity aggregation and order routing to algorithms, risk management and data analysis to regional banks to white label to their franchise. As a bank provider we are ingrained with our clients across multiple facets allowing a deeper relationship than just executing FX,” says Congdon.
One of the technology providers offering FX algos to both regional banks and buy-side clients is smartTrade who are therefore able to provide perspectives from both sides of the fence. According to John Stead, global head of pre-sales at smartTrade, one of two key drivers of FX algo adoption among regional banks has been competition. “FX banks see that their competitors are using algos and now they have to show their clients that they have them too.”
Not only are there corporates trading FX as an asset class and looking for more advanced execution methods, there are also corporates looking for other ways to hedge their FX exposure and are willing to look beyond their traditional banking partners. At the one end are the aggressive buy-side firms looking to FX as a means of alpha generation. And at the other end are the firms that are looking to make their FX trading more efficient than relying on their custodians to execute all of their trading.
The second driver has been the shift in responsibility, from bank to client, when it comes to FX execution. This is partly due to regulation, such as the likes of MiFID II, and the greater emphasis on best execution.
However, while all the large banks and the majority of regional banks offer FX algos, there is a still a challenge in persuading buy-side firms to adopt algos. “We see barriers to adoption every day,” says Ludovic Blanquet, chief product and strategic planning officer at smartTrade. “A level of education is needed from the liquidity-providing banks to their liquidity-taking clients.” Some of the barriers are around resources. For example, does the client trade in great enough volume to warrant the outlay on FX algos and the ancillary costs for some firms like colocation in a data centre. Then there are the data costs, ensuring that the right formats are used for inputs and outputs. And then there are the staffing requirements and ensuring that there is someone with enough experience to manage these data issues.
Another interesting area is best execution, says Blanquet. More algo providers are offering TCA services and reporting for best execution but buy-side firms need independent TCA rather than reporting from the algo provider. However, best execution and transaction costs is a much more complex issue in FX compared to centralised asset classes like equities.
“While this has been a critical factor in the use of algos for other asset classes, like equities, it is a much more complex issue in FX,” says Blanquet. “In the absence of a centralised order book or consolidated tape, what does best execution really mean? Currently there is no regulation and a use of improper benchmarks. At the same time, the buy-side is taking more control of their FX trading and using FX as more of an asset class than a back-office process. This should make the industry address the issue because the buy-side firms want the same standards they have for other asset classes.”
However, the costs involved could prove to be a barrier for all the non-bank liquidity providers looking to get closer to the buy-side by offering FX algos. “Without a central order book, it is very expensive to get all the correct data points and to undergo all the back-testing,” says Stead. “This will become more of an issue as the industry becomes more data-intensive. There are more liquidity providers trading on a 24/7 basis.”
Another barrier to adoption is liquidity appropriation, where liquidity providers are unwilling to offer algos that can split liquidity to their competitors. “If you have 10-12 liquidity providers, do you take an equal amount from each one? How do you split them? That decision has to be made by the trader but they need to be well informed and educated about making that decision within an algo context and ensuring that they get a good return for their fund and achieve best execution,” says Blanquet.
It is the kind of issue that can also affect the relationship between liquidity providers and takers, says Blanquet. In order to address some of these issues, smartTrade has developed a new module, AlgoBox which Stead describes as a “comprehensive container within which code can sit directly alongside the trading platform”.
Through AlgoBox, users can access the key parts of the smartTrade platform, whether to source data or make changes to trading strategy or monitor their key algos. “We see this as a major step for algo users to escape the liquidity appropriation,” says Blanquet.smartTrade is one of the first technology providers offering algos. As FX algo adoption increases among European buy-side firms, so will the number of FX algo providers. It will therefore be critical that market participants are well-informed about the options available to them. As Blanquet says, “Everyone will be an algo user in the next five years. Fundamentally it is a matter of education.”