Sentiment analysis in particular has quickly gained supporters in financial services and beyond over the past 12 months. The USP of the sentiment analysis approach is that it can take unstructured or subjective data from a vast myriad of online sources and sift, sort, extract and quantify insights from this mass of data based on specific objectives and search criteria.
Importantly, AI-driven sentiment analysis can in a practical sense determine the emotional tone within a textual source, and as such can be used to gain a much more detailed understanding of the attitudes, opinions and emotions involved. How and why are increasing numbers of UK firms using sentiment analysis to innovate in financial services?
Irrational markets
The recent showdown between retail traders organising via Reddit pages and hedge funds has provided a visceral reminder of the potential for extreme volatility latent in certain corners of the US equities market. The two main factors explaining the dizzying ride GameStop’s share price went on were the growing popularity of retail day trading – in part a function of the arrival of low-cost online brokerage platforms such as Robinhood and eToro – and the sheer depth and liquidity of the US’s capital markets.
However, it is also important to point out that a certain ‘sports betting’ culture, often merging with a sort of loose anti-establishment sentiment, further fuelled the buying frenzy that saw the market cap of a moribund company temporarily spiral upwards to absurd levels, many multiples ahead of what reality justified.
Whilst it would be difficult (but certainly not impossible) to imagine a similarly irrational yet coordinated mania erupting on the same scale in UK or European equities, the incident has underlined the extreme importance of subjective online data for understanding market movements. In fact, the growing divorce between company fundamentals and stock performance has reached such highs that the hoary old debate over active versus passive investment management has received a new lease of life.
Carson Block, founder of infamous short seller Muddy Waters Capital, has recently argued that the ‘stonks’ phenomenon provides definitive evidence of the superiority of the active approach, citing ETFs buying and selling on autopilot as an exacerbating factor in the recent volatility.
Data, data everywhere…
There is a sense of fighting the inevitable in Block’s claims, and the rout of passive funds over their active counterparts is in all likelihood set to continue in 2021. But he is right to point to the profound shift that is taking place in terms of what data matters for financial markets.
Even just a few months ago, few financial commentators would have believed that non-professional traders swapping tips back and forth on an online forum would amount to anything substantial enough to cause a price swing such as occurred with GameStop. And whilst there is still no effective challenger for the awesome functionality of the Bloomberg terminal, the rise in significance of what could be termed ‘alternative’ sources of data used by retail traders simply cannot be ignored.
The expansion in the quantity of online notice boards, data providers, research aggregators, chat forums, and all manner of other digital communal spaces has meant the amount of data that could be of relevance to understanding market movements is now bigger than ever before. All of this represents additional data on top of the company balance sheets, press releases, research notes, forward P/E ratios, and so on that traders have always relied upon to make decisions. As such, data is becoming ever more dispersed and diverse, and access to and correct evaluation of these flows of information has never been more important.
Sentiment analysis in the UK: established successes and who to watch for 2021
There has been a strong UK presence in this growing field for a while now, and it undoubtedly represents a key area to watch for 2021. For example, Brighton-based Brandwatch has expanded rapidly through organic growth as well as strategic mergers and now have offices in the US, Spain, Germany, France, Singapore and Australia. Brandwatch utilises sentiment analysis to provide insights on how brands are talked about by their customers, and have now set their sights on gaining further partners in financial services.
Another early adopter is Preqin, a London-based research firm offering insights on the private equity sector. This is based on both qualitative and quantitative AI-led techniques to harvest data, which is then fashioned into insights. Often referred to as the Bloomberg for unlisted investments, Preqin simply couldn’t do what they do without the use of sophisticated AI/ML web scraping programs that gather the data that its analysts then forge into insights.
Alongside these established successes, there is also a thriving startup scene here too represented by the likes of SentiSum and Warwick Analytics.
Warwick Analytics' main focus is on how to use AI to improve customer service in the high street banking sector. The business argues that their tech can lead to numerous efficiency savings by quickly identifying customer needs, and more specifically which channel is most appropriate for meeting these needs.
The combination of Warwick Analytics’ AI-powered form of sentiment analysis with the large data-set of all customer interactions generated by a high street bank enabled them to know what channel will work best for specific types of interaction; understand the causes of channel failure and what drives customers to switch; and reduce customer effort by delivering good service in the customer’s preferred channel first-time. In other words, sometimes a smart chatbot is a great time-saving device for both banks and customers, and sometimes it simply isn’t appropriate and will get a negative response from the customer.
A similarly promising prospect is AI/ML startup SentiSum. The London-based firm, which was founded in 2015, has created a customer experience analysis and insights solution powered by AI aimed at companies in the insurance industry, but also with possible applications in retail and elsewhere.
It would be fair to say that the insurance industry can’t always be held up as a shining beacon of good customer service, (see Hiscox and the recent small business interruption scandal), and this is where SentiSum’s tech can help. It allows the processing of high volumes of data in real time to identify key areas of concern for companies and make actionable recommendations for improvement based on how customers respond. It can also automate repetitive support processes, with the aim of improving efficiency and, as a result, improving customer satisfaction and reducing costs
In summary, today there is more data out there than ever before, and even in finance it is far from being purely or even mostly numerical these days. A recent report by Renfinitiv found that 72% of financial services firms consider AI/ML to be essential to their core strategy, and 40% expected large increases in investment here owing to the impact of COVID.
Making good business decisions requires good data, and there’s a wave of SMEs in the UK and beyond who are developing the tools to make this easier for everyone from individual investors to banks to insurance companies. As such, sentiment analysis in particular and AI/ML in general is a highly promising growth area to watch out for in 2021.
