Automating and Streamlining the Front Office
Data science enabled technologies, specifically machine learning and deep learning, have become an integral part of front office processes. Initially, these tools were used for analysing large data sets to improve decision making. Now, more and more firms are using these technologies to automate front-office processes, such as pre- and post-trade analysis, regulatory reporting and, in some cases, even trading itself.
In recent years, the availability of large structured and unstructured datasets, combined with the exponential increase in computing power, has seen a huge growth in machine learning applications across the financial markets.
Across the front office, banks are applying this new technology and machine learning techniques to high frequency trading and algorithmic trading. “Systematic market-making is something that banking clients are definitely getting involved in, especially the bigger banks have already started implementing it a while ago. And now, I think that most other dealers will have to join in this trend and it will become a prerequisite and a necessary ingredient of the market-making in the future,” comments Alexei Tchernitser, Director, Product Management at Quantifi.
Quantifi recently hosted a webinar with Risk.net on the “Front-office reboot: how new technology, ML and data science are reshaping trading”. During the webinar, participants were surveyed on how new technology is being utilised.
Are you aware of firms using fully automated machine learning solutions for decision making in any of the following areas?
Machine Learning & Decision Making
Algorithmic trading (50 percent) and systematic market making (38 percent) are two use cases where machine learning can be utilised. Whilst humans are very good at making decisions, we are limited as to how fast we can make them which is why this type of activity is so well suited to machine learning. However, one of the downsides of the machine learning model is that whilst it allows for faster decision making, it can be very slow to learn. For example, if there is a change in the market – a decision from the Fed or something happening geopolitically, such as the war in Ukraine – firms will always need humans on the back-end to reassess these decisions and take over the control from the machine if necessary.
These two use cases are also examples of areas that can benefit from a deep learning or reinforcement learning approach. “In these cases, where you have got high frequency data, you will have a lot of training samples potentially, so this is technically ideal for machine learning to be applied and less so, for example, if you want to do portfolio rebalancing which may occur every month, where you will not have a lot of data,” comments Francois Mercier, Vice President, Senior Data Scientist at Citi.
Participants were also surveyed about how far along they are in their cloud migration. With increasing demands from customers, greater regulatory requirements and cost efficiency pressures, more and more firms are adopting cloud technology to reduce costs, increase resilience and agility, as well as support the scalability of resources.
How far along are firms in their cloud migration?
0% of firms being 100% complete in their cloud migration does not come as a surprise – especially when it comes to the larger firms. “It’s not surprising that’s the view, I think at SG [Société Générale] we are probably around that [80%]. You know, 80% is the target where we want to get to and we started our journey, probably about ten years ago now,” comments Sohail Raja, Head of Execution Platforms & UK Chief Digital Officer at Société Générale.
For the firms considering cloud computing in the context of data science and machine learning, the next few years are going to be very interesting. Banks are going to be able to interact with FinTech providers much more seamlessly and once firms have this level of access, there will be countless benefits for market participants.
Quantifi has taken full advantage of the trend towards cloud technology by collaborating with cloud providers like Microsoft Azure. The move to cloud offers clients greater speed and agility. By leveraging Microsoft Azure, Quantifi can deliver an elastic platform that can easily scale processing capability up or down and be configured and changed at a granular level to accommodate our clients’ business needs.
When it comes to machine learning, automation and efficiency gains are always central to these kinds of discussions. However, it is important to consider the other benefits of machine learning including innovation around new products, new trade ideas and new functions firms can offer their clients.