FairXchange Unveils AI-Driven FX Liquidity Analytics

Whilst it is undeniable that data and analytics advances have greatly enhanced the relationships between FX liquidity providers and customers, the advance has not come without new challenges – namely, the sheer weight of data available and how that inhibits institutions’ ability to pick out the pertinent changes in service that are important to both parties.

Building upon its Horizon product, which delivers that data to users, FairXchange has launched a new AI-driven liquidity data analytics solution, Horizon Sentinel, that seeks to optimise financial institutions’ use of liquidity and counterparty data by reducing the time spent on building datasets to discuss with clients.

Sentinel is the first product resulting from the work of FairXchange’s Research and Development team, which was augmented by the hire of former hedge-fund manager, Will Holt to run the unit. “We have been putting the building blocks in place as more and more clients have asked us what we are doing with AI and machine learning,” explains Guy Hopkins, CEO of FairXchange. “This has allowed us to put the more complex pieces in place, enabling us to look at how we could get the critical data out – to easily highlight to clients exactly what the data is showing, so they can act upon it. That is what Will has been working on for the year he has been with us.”

One of the challenges with data services aimed at bolstering the trading relationship is that the data is analysed too late, often on a monthly basis, and misses intra-month changes in behaviour at either LP or client, or both. Additionally, the sheer size of the datasets means that small, but important, changes in behaviour are easily missed. This, says Hopkins, is what Sentinel addresses, by bringing the important datapoints to the fore.

“There is also the challenge of maintaining a large number of relationships,” he continues. “In many organisations, sales teams are naturally drawn towards the largest customers and spend a lot of time managing those relationships, which is fine, but it does mean that the opportunity to strengthen other relationships can be missed, because the teams simply don’t have the time and resources to identify opportunities. With Sentinel, they can identify potential issues early, and act pre-emptively to ensure the relationship runs smoothly.”

It is vital that changes in pricing and liquidity are visible to both sides of the relationship, and are able to be discussed on a timely basis

As was the case with Horizon, an early adopter of Sentinel has been Sucden Financial, Wayne Roworth, global head of FX at the firm says Sucden has been actively using the Sentinel AI tool in Horizon for several months. “It has had a substantial positive impact on our business,” he asserts. “We have a complex trading business with many important counterparties, both on the liquidity provider and client side. FairXchange’s AI driven alerting empowers our team to keep abreast of every change in the business.

“Sentinel allows us to quickly identify potential changes and act upon them immediately,” he continues. “It allows us to proactively manage dialogue with our counterparties. In particular, our liquidity providers very much value the fact that we can play an active role in monitoring the flow, highlighting areas of potential concern that we can then work on together. This makes for a much more balanced relationship, which importantly allows us to differentiate ourselves from our peers.

“The dialogue that this data facilitates, and the speed with which we can act upon it, has resulted in numerous improvements to our liquidity that have allowed us to increase our client volumes and associated revenues,” Roworth adds.

A clear benefit of earlier identification of issues is earlier remedy, a point Holt makes when he highlights the evidence from a series of weekly calls with clients. “We have seen some streams widen out that have been picked up by the counterparty within a day or two,” he explains. “Before, these would maybe not have been picked up for three or four weeks.

“This is not a negative thing,” he stresses. “If you know there is a problem, you can do something about it before the issue potentially gets bigger, by having a conversation with the counterparty.”

Effectively, Sentinel is bringing the exception-based concept, that is prevalent in trade processing, to the relationship, something that frees up time and resources. This no longer means, as was perhaps the case previously, that the onus is on the LP to discuss the data with the customer. “We have been told this is changing the dynamic of the relationship a little,” says Hopkins. “Whereas before the client often waited for the bank to start the conversation. By using alerts, they can initiate the discussion. Previously this was difficult because how do you make the time to analyse such big data sets, especially if it is across 10 or more LPs?”

Different Approaches

Hopkins says that Sentinel has maintained FairXchange’s philosophy of providing neutral data that can benefit both sides of the relationship and this is reflected in the functionality of the platform, which allows users to access specific data relating to their business.

From the LP side perhaps one of the more valuable functions is how the platform can alert the wider business to dynamic adjustments made by the trading team. These changes can be the result of different market conditions, and how the risk book is coping with the day’s flow as two examples. “For a variety of reasons, the e-trading team could widen or tighten spreads, but they may not transmit these changes internally,” explains Hopkins. “This means that a changing of spreads to a client, which could, for example, be the result of a P&L drawdown, may not be transmitted to that client because the relationship team are unaware of it.

“It is vital that these changes are visible to both sides of the relationship, and are able to be discussed on a timely basis,” he continues. “This is especially so if there are, perhaps, cumulative events that all contribute to a liquidity change. It can be hard to analyse the drivers of a change if they happened on multiple days and had different impacts. By identifying them sooner, LPs can head off any meaningful changes in liquidity.”

Sentinel allows institutions to manage their relationships on a daily, rather than monthly basis, and adds value to both LP and clients

To actually access the data highlighting any change, Sentinel allows users to filter alerts by different criteria. At a high level it displays all alerts generated by the AI models that are seen as statistically significant, these can then be filtered down to study the data by different aspects of the business, for instance through the prisms of yield, mark out or volume, and again by specific currency pairs. As is the case with the Horizon platform, Sentinel provides data in both basis point and dollar terms. “That has been an important message from our clients,” says Hopkins. “The more you can quantify the dollar impact of the changes, the easier it is for institutions to escalate any issues.”

The data is presented both numerically and visually and users can then interrogate the model for any significant underlying drivers. The relative changes are also highlighted, to only pick out occasions when an LP’s changes are an outlier from their peer group, thus ensuring that the noise associated with a general market regime change can be removed. Specific improvements that LPs have made are therefore much more easily identified, allowing them to more clearly demonstrate where they are actively engaged in winning more of their customers’ flow.

“One of the things we have noticed from our initial deployments is that the platform is open to nearly every  area of the business,” says Holt. “It’s not just about liquidity management or trading; sales and compliance teams can also use its functionality to help manage their business.”

On this point, Holt explains that users are able to save different views, once they have the alerts process filtered to their requirements, however as an example, the business management team are able to filter by aspects that the client-facing teams are not studying. Overall though, Hopkins says clients are establishing a daily cadence whereby Sentinel helps them prepare for the day ahead in a much shorter time. “We have had clients tell us that what was taking each staff member 30-60 minutes in the morning, is now only taking five or 10. That is a huge efficiency gain,” he points out. “Clients are also using the platform to prepare for their LP reviews.”

Ultimately, Sentinel is aimed at providing desks on both sides of the relationship with a very vital commodity – time. As desks continue to be time-constrained, access to the right data at the right time will become a crucial differentiator for service providers especially. “Sentinel improves productivity, minimises opportunity cost, and allows users to focus on the critical aspects of their business,” Hopkins concludes. “It allows institutions to manage their relationships on a daily, rather than monthly basis, and adds value to both LP and clients.”