CEO CHAT: Rowland Park, Limeglass

Research drives the trading bus.

The buy-side needs it and the sell-side must provide it. But compared to technological progress in other areas of finance such as trading systems, innovation in the arena of financial research has been languishing in the dark for years. This limits the value of high-quality reports and leads to a loss of opportunity for market participants.

Enter Limeglass, a fintech firm that has set out to transform the way financial research is distributed and consumed. In a conversation with Traders Magazine’s Editor John D’Antona Jr, Limeglass Co-Founder and CEO Rowland Park shared how his firm is employing “cutting edge” artificial intelligence (AI) along with Rich Natural Language Processing (NLP) to fill this gap. He also discussed his team’s decades of experience in financial research and how it has created a brand-new way to access research, focused on picking out the relevant paragraphs in documents across the whole research library.

TRADERS MAGAZINE: Tell us a bit about the background of Limeglass and how the company was founded.

Rowland Park: I’ve only worked for four companies, three of which I co-founded. All the previous companies I worked at were research companies, the last one being 4CAST.  4CAST provided short-term financial market analysis on macroeconomics, policy, foreign exchange, bond and money markets from offices in each trading time horizon.

Rowland Park, Limeglass

Our analysts were producing hundreds of reports each day on anticipated short-term movements in the markets and our clients were banks and brokers who themselves were producing a huge amount of research. They liked our service because they could find relevant analysis much faster than on their own portals.

Traders would often complain that their mailboxes were deluged with unread research. Having a mass of reports that are difficult to navigate leads to information overload. Even with inbox rules set up to move incoming research to various folders, market participants rarely had time to read even a small proportion of it.

If they were searching for a specific topic, they would resort to traditional methods, such as using the Control+F function in documents or searching through their email folders. These methods were – and still are – inefficient, frustrating and a poor use of time. It is so easy to miss important details by searching through research this way. As a result, these cumbersome, yet prevailing, methods provide an incomplete view of the markets.

Several of us left after selling 4CAST, and this issue of under utilization of research and information overload came back to mind. I realized there must be a better way of producing and consuming research. I discussed this problem with Simon Gregory, the ex-CTO of 4CAST with whom I had worked for 15 years. We decided to combine our expertise to provide a better solution to the way that financial research is disseminated and consumed. So, we set up Limeglass to do just that.

After a couple of years of R&D, we received early-stage funding from several angel investors to begin developing a financial research platform. The company was also awarded an innovation grant by InnovateUK, the UK Government agency set up to foster research and innovation across the country.

We’ve been fortunate to take part in J.P. Morgan’s highly selective In-Residence Program, which assists emerging tech companies to develop production-ready solutions for solving critical wholesale banking problems. The whole experience was very useful and resulted in J.P. Morgan investing in the company. We were delighted with this vote of confidence in our solution and we’ve continued to work with the bank to make our platform suitable for use by other world-class financial organisations.


TM: Why is Limeglass technology applied to financial research in particular?

Park: The importance of data and analytics is talked of a lot in today’s financial markets. There is a long-running arms race for the biggest institutions to collect the most data and produce the most sophisticated technology to best leverage that data. But the rich trove of unstructured information in financial research has received relatively little attention from technology teams. We think that is a missed opportunity.

This has been given greater urgency by changes in legislation in major international markets that have led to more scrutiny on buy-side and sell-side operations. Since the 2018 introduction of MiFID II in the European Union, designed to improve accountability of costs to customers and unbundle research and execution services, there has naturally been a change in the way financial services companies operate.

A review by the UK’s Financial Conduct Authority found that research budgets across the board have fallen by 20-30 per cent. Many of the largest buy-side firms are putting more effort into developing more research in-house. Additionally, tightening margins and an increasingly demanding global environment are, of course, factors that have an impact on all sectors of the industry.

This means that market participants need to make better use of their research.

There has been practically no innovation in this area for years, so we saw an opportunity to use technology to make real progress. In harnessing AI  and rich NLP in conjunction with human expertise, we created a platform with the ability to comprehensively assess research documents.

Moreover, the functionality means that this works for individual pieces just as well as for a bank’s entire research library, for an all-round view of relevant topics. This offers huge scope to maximize the value of financial research, for both the producers of reports and the market participants who receive them.


TM: What makes Limeglass so innovative?

Park: Limeglass uses a process we call ‘document atomization’. In essence, this means breaking down a document to identify the relevant topics and ideas at a paragraph level rather than treating the whole document as one item. We then smart tag these ‘atoms’ with relevant identifiers, organizing them into an asset-specific taxonomy that has further layers of classifications for country, maturity, and so forth.

However, we not only tag particular phrases themselves, but also their synonyms and related phrases through the use of rich NLP. For example, if you wanted to find paragraphs that referred to the US-China trade war, you would also find sections that mentioned ‘tariffs’ and ‘US-Sino trade tensions’ (and more than 40 other synonyms), so that these particular passages would be highlighted as being relevant to your search.

Once the insights in the research have been broken down in this manner, they can be reassembled in any number of different combinations to perfectly suit the needs of the individual market participant at any given moment. Our technology also enables institutions to personalize their research for any audience, maximizing the value for users and ensuring that the correct research reaches the right teams.

It’s this combination of providing both a wide-scale context and a granular perspective that we believe is unique in the research space. Furthermore, with AI and machine learning, our platform is continuously evolving as more and more phrases and synonyms are added and tagged.

In creating the platform, we considered not only how research is used, but also how it is produced. With our technology, research report writers are able to track and trace how their material is used, providing useful metrics that can be taken into account when producing any new project reports.


TM: What are Limeglass’s plans for the future?

Park: We are continuing to develop relationships with financial institutions to showcase the capabilities of the platform. The dissemination and consumption of financial research is a vital tool in indicating which trading decisions will be the most profitable.

Technological advancement in this arena is only the start of what we can offer. Once research is atomised on a regular basis and at scale, it provides all kinds of opportunities for innovation. Once paragraphs are identified in research documents, they can be linked to other applications in many ways. For example, they can be linked to an execution platform or messaging window where the user is looking at the price of an instrument and wants to read any related research.

Our aim, in creating a document atomization platform, is to transform the financial research market, turning the liability of too many reports into the valuable assets they are designed to be.