Matthew Hodgson
Matthew Hodgson

FXLIQUIDITY Providing the FX market with the insights that it needs

Matthew Hodgson, CEO and Founder of Mosaic Smart Data discusses FXLIQUIDITY, a collaborative project set up between Mosaic Smart Data, CLS and MUFG, delivering free FX liquidity insights to the market in response to the COVID-19 related market turmoil.

Matthew Hodgson, CEO and Founder of Mosaic Smart Data discusses FXLIQUIDITY, a collaborative project set up between Mosaic Smart Data, CLS and MUFG, delivering free FX liquidity insights to the market in response to the COVID-19 related market turmoil. The service analyses data from 11 key currency pairs (USD/JPY, EUR/USD, GBP/USD, USD/CAD, AUD/USD NZD/USD, USD/CHF, USD/MXN, USD/ZAR, USD/HKD, USD/SGD) and will continue to be delivered on a weekly basis for the remainder of 2020, with spreads, volumes and liquidity changes posted weekly to mosaicsmartdata.com/fxliquidity.com. 

The impact of COVID-19 has led to stressed market conditions and enormous volatility with huge volume fluctuations and dwindling liquidity.  The lack of liquidity and volatility insights has presented new challenges for many firms and market participants. At the end of March 2020, when COVID-19 was leading to international lockdowns and wide-ranging economic uncertainty, Mosaic Smart Data, along with CLS, the largest market infrastructure group delivering FX settlement, processing and data solutions, and Mitsubishi UFG Financial Group (MUFG) held initial discussions on creating a new data analytics service. 

As the most liquid global market with the largest variety of participants, FX underpins all international capital flows.  As such and working in collaboration with CLS and MUFG we came to the mutual agreement that the creation of a free service that provides greater transparency of currency moves and the resulting associated liquidity would benefit all FX market participants. 

This new service was christened FXLIQUIDITY and it offers transparency to FX participants, enabling them to focus their attention on what is most relevant to their requirements during stressed market conditions and periods of high volatility – for instance, when the best windows to express risk are available, and when the best time periods to access the markets for risk mitigation or hedging present themselves.  This was the use case that all participants consider valuable. 

Steering the project

Embarking on this ambitious project, we quickly came to the realisation that FXLIQUIDITY was a sizeable multi-disciplinary endeavour requiring inputs from the functions of global company management, information services, data science, quantitative finance, FX trading, customer relations, legal and regulatory compliance, technology, user experience, marketing and communications. 

At Mosaic Smart Data, we led the management of the project, developing machine learning algorithms to model liquidity patterns, creating and hosting the microsite, as well as integrating the periodic data sets supplied by CLS and MUFG, which drives the weekly time series, matrices and commentaries. Mosaic Smart Data and MUFG then analyse the data findings to produce the main summary highlights for each weekly report. 

CLS, as a company that provides FX settlement and data services for approximately 50 per cent of the world’s FX trades, is highly regulated by international authorities and central banks. This meant that many hurdles had to be overcome for the project to gain global compliance before it was launched, including sign-off by the US Federal Reserve.

FX Liquidity

Logistics

The unparalleled market turmoil and activity presented a number of hurdles for the three collaborating firms with the most pressing being speed to market.  It was becoming clearer by the day that market participants needed insights and they needed them fast. As a result, we realised that the site needed to be launched as quickly as possible. With this at the very forefront of our minds, we ensured that the FXLIQUIDITY prototype microsite was created within 24 hours of the initial agreement to proceed between the three parties.  Thereafter and within a few short weeks the analytics parameters were decided upon and the vision to offer the service for free, as a benefit to the global FX community, was agreed. 

From the technical infrastructure being configured, legal compliances between the three firms were established, the site was stress-tested, and a delivery and communications plan across the collaborating companies was formed.  FXLIQUIDITY was launched on 18th May 2020 – only a month and a half after the idea was initiated. 

The metrics and results of FXLIQUIDITY

The data analyses volumes, spreads and liquidity of 11 key currency pairs across a rolling three-week period with commentaries available for each, which is updated weekly.

The analysis of key currency pairs compares three time periods: ‘pre-COVID-19’ (1st June 2019 to 20th February 2020, as a baseline), ‘COVID-19’ peak (21st February to 20th March), and ‘recent’ (a rolling three-week period to provide a realistic overview). Currency pair spreads, volumes and liquidity term structure are analysed, and the findings are presented with supporting time series and commentaries. 

Initial findings revealed that G10 currencies were returning to between 70 and 80 per cent of their pre-COVID-19 liquidity levels while emerging market (EM) currencies were not faring as well at around 50 percent of their pre-COVID-19 levels. On the other hand, CHF was the very first currency to recover back to pre-COVID levels. Recent trends show that EM currencies have seen improvements in their liquidity scores, although still remain a long way from their pre-COVID levels. 

Other key points have included spreads widening during New York trading hours and volumes going up between 10 and 25 per cent in all currency pairs analysed during the first three weeks of June.  In particular, we have recently seen a deterioration of Euro liquidity, driven by highly directional flows and imbalances building up towards the end of the holiday season in August, as dealers and the buy-side scrambled to adjust their flows. Amongst the G10 majors, AUD remains the worst performer in terms of liquidity pick-up.

These findings have all been posted to the FXLIQUIDITY microsite and were made available to the media, to customers and prospective customers via widespread mail distribution and via social media. On the whole, liquidity levels are struggling to return to pre-COVID levels, underscored by minor weekly spread and liquidity changes. Confidence is still to return, highlighted by just how slow the market has been to recover over this trying period.  

FX Liquidity
GBP/USD Hourly spread (pips)
FX Liquidity
GBP/USD Hourly volume (MM Base Currency)
FX Liquidity
GBP/USD Hourly liquidity profile

Modelling the data

Once the parameters were decided upon in analysing the data across the 11 key currency pairs – in terms of volumes, spreads and liquidity – the modelling was carried out by Mosaic Smart Data’s data science team and validated by Professor Rama Cont, Chair of Mathematical Finance at the University of Oxford. 

The goal for the data science team at Mosaic Smart Data was to establish liquidity profiles on each of the 11 currency pairs that were analysed. To do this we were provided with a rich source of data from CLS on executed FX transactions in all currencies where we can observe the price, volume and trading volume dynamics, which includes the flow directionality.  MUFG supplied the order book data which allowed us to compute a view on the dynamics of supply and demand where we can observe the bid/ask imbalance, the spread and the depth of the order book all of which gives us an idea of the cost of execution at a particular time. 

These aspects of market liquidity and market indicators are what aided us to deliver a benchmark liquidity profile in each currency pair, and which is made available on the FXLIQUIDITY microsite. 

For this, the Mosaic Smart Data team used machine learning to take a set of variables contributing to liquidity and look for the best combinations which give the highest explanatory power for the cost of execution in each currency for different volumes. The aim is to explain the cost of execution for market participants, as this is the fundamental principle of liquidity; the ease with which you can execute a given trade in a given currency. The result of the machine learning approach is what we deliver to users on a weekly basis and the graphs that are featured alongside this article.  
The next logical step is to model these patterns to generate a forward-looking out-of-sample prediction in the 24-hour period ahead and is work we are currently embarked on at Mosaic Smart Data.

Analytics in the time of turmoil

The fundamental driver behind FXLIQUIDITY is the need for smart analytics. The real issue across markets today is that financial institutions currently sit on BIG data that is expensive to gather and store and offers little commercial and strategic value unless it is refined. Market participants across the capital markets are drowning in an overwhelming deluge of this incoherent data and have limited capability to separate the signals from the noise and value within it.

Analytics has always been vital when navigating the financial markets, let alone in the midst of a global pandemic, and the market is quickly coming to the realisation that it is no longer simply a ‘nice to have’, but is tantamount to survival. 

As result, there is very little point in having lots of data if you are unable to analyse what it means or how it might impact your business. As such, at Mosaic Smart Data, we have developed advanced analytics where we inform our customers in real-time – What has happened, Why it has happened, and What is likely to happen next. Because of our technology and expertise, our advanced real-time analytics tools provide cohesion to FXLIQUIDITY and we’ve pointed our analytics engines at this solution. One of the exciting technologies we have leveraged for this initiative is Mosaic Smart Data’s Natural Language Generation (or NLG) technology to provide easy-to-read commentaries in the reports that give real insights to users, and this is done without human intervention, so there is no time-lag from any manual processing.

FX Liquidity
SPREAD ELEVATION MATRIX: Shows the % change in the spread levels between the recent period and the pre-covid period

A look to the rest of 2020

2020 has certainly been eventful across the financial world with no slowdown in sight. With the ongoing uncertainty from the fallout of COVID-19 and as the world continues to adapt, uncertainty across the capital markets is likely to continue. The US Presidential election in November 2020 is undoubtedly set to inject yet more volatility into the markets and as a result, the need for value-enhancing analytics is set to continue.  

FXLIQUIDITY is proof that three market participants of varying size, working in tandem, at speed and for the benefit of the entire FX community, can deliver tangible value and as we move into the final quarter of 2020, we will continue as long as there is an ongoing demand for the service we have provided.