How is technology helping to make the latest generation of FX algos much smarter?
Nowadays algos are consumers of data events, both historical and in real time. Data availability has exponentially grown in the past couple of years. As such, algo design has to be more centered on their capacity to process, analyze and model data in both real time and in simulations. As the capacity to process data increases, algos have the ability to take more factors into consideration while making real time decisions, therefore potentially both improving the performance of the algos and helping our clients access liquidity more efficiently.
How might the application of new technology open up algorithmic FX trading to a broader client set?
Technology is playing a critical role in the advancement of not just FX algo Trading but FX electronic trading in general. In particular, improvements to OMS/EMS technology are playing a critical role on the enhancement of the client workflows. We see: a) vertical integrated solutions providing not just liquidity/algo access to clients but also post-trade and independent cost analysis, critical to provide a feedback loop to enhance execution and/or b) horizontal integrated solutions which enable similar workflows across different asset classes.
These technology improvements are a critical tool for workflow execution automation, enabling buy-side traders to focus further on strategy and optimization. Trading on larger sizes and more illiquid pairs through algos is becoming increasingly common. Our data suggests that clients benefit on more efficient execution in those type of orders. As workflows make data available, naturally I would expect those flows to open up to further use of algorithmic trading strategies.
Artificial neural networks (ANN), of which DNA is a type, are inspired by the biological neural networks of the brain
Where is technology being applied to improve algo liquidity management and give clients a better way to engage with liquidity over time and much faster?
DNA (Deep Neural Networks for Algos) is a great example of the use of machine learning (more specifically neural networks and reinforcement learning techniques) and technology to improve the way algos engage with liquidity. Clients provide the high level schedule of the algo and DNA takes care of the interaction with liquidity, a highly nonlinear problem, well suited for this type of solution.
Speed is important but only to a certain extent. In an environment where venues are implementing speed bumps, market data is throttled and response times are on average several ms, it seems more relevant to understand liquidity interactions than shaving a couple of micro seconds in the trading strategy.
How is technology facilitating the design of new quantitative toolsets to help clients fine tune their FX algo parameters?
From a client perspective, new quantitative toolsets can only be as good as the data used to build them. Technology and electronification enables further traceability which in turn means better benchmarking. Transaction Cost Analysis on the data will empower clients to close the feedback loop to fine tune their choice of strategies and parameters.
In what ways is technology being harnessed to release the power of data and analytics in FX algo trading?
Technology has evolved quite rapidly in this area over the past couple of years. Standards of technology and integration have enabled banks and third party providers to offer unique solutions to empower clients with cutting edge visualizations for Pre-Trade, Real-Time and Post-Trade Analytics.
A good example is Algo Central, the J.P. Morgan Data analytics and algo entry tool which is now accessible within several third party platforms. This integration means that clients have the best of both worlds, they can access the latest bank algos and functionality as well as the latest developments with regards to data visualization.
Algo Central, the J.P. Morgan Data analytics and algo entry tool, is now accessible within several third party platforms including Bloomberg
Why is the investment they make in technology increasingly becoming a key differentiator amongst FX algo providers?
Ultimately in order to be competitive dealers we have to enhance the execution experience for clients. The actual algos and models within the algos are an important part of it. On top of that, there is a range of technology services around them which provide real value-add. Some notable examples are third party integration, booking workflows, data services, data analysis and visualization, automated pre-post trade actions, alerting… and I could keep going. All these important services around the algos require important technology investments which will differentiate the overall quality of service by reducing client costs holistically.
Looking to the future, how might technology be applied to take FX algo trading to the next level?
The future is already here. We now have algorithms which make use of neural networks for liquidity optimization (DNA). There are data visualization and cost analysis tools (Algo Central) integrated in client workflows pre-trade, real-time and post-trade. Feedback/alerts are provided while the algo is executing (Algo Insights).
All these tools will no doubt continue to evolve. The biggest likely change is that all these advancements will cause the number of order types and parameters to consolidate. This transparency will enable simpler and more intuitive ways for clients to select how they want to execute.
