In the initial months of 2021, a Reddit message board, WallStreetBets, caused an upheaval in the financial markets. It highlighted the power of social media and democratic trading apps in bringing down Wall Street power players. After the GameStop saga, many US investors started flocking to sites like Reddit to gather stock tips. A survey by Travis Credit Union in March 2021 revealed that over 70% of retail US investors depended on these unconventional sources of information, rather than listening to traditional news sites for stock tips.
This is just an example of how the institutional segment has been impacted by shifts in consumer behaviour, unexpected events, and changing market signals. Financial institutions must now readily look into different sources of data to gain a competitive edge and respond to fluctuating market dynamics.
Data has always fuelled the financial industry. But today organisations that can leverage alternative data sources and expand their analytics capabilities through AI, ML and NLP, are future-ready in the truest sense. It therefore comes as no surprise that over 92% of market leaders in the hedge fund industry not only use alternative data extensively but are looking into new ways of quantification and codification of data from unconventional sources on a large scale.
In the simplest sense, alternative data is what investors use to evaluate a stock or an investment, derived from non-traditional data sources. By traditional sources, we mean the usual company earnings previews, financial statements, press releases, macroeconomic data, geopolitical news from traditional media, central bank meeting minutes, SEC filings, and presentations by the management, and so on.
However, sources of alternative data are as diverse as:
This includes information related to demographics, click-through rates, popular keyword searches, number of unique visitors to a website, etc. These metrics can enable investors to evaluate the user base of a business, the popularity of an advertising campaign, the stickiness of the user base, future trends in various products and services. Web traffic data can provide insights into diverse industries, for both small and big, B2B or B2C, companies. It can provide information about the condition of the labour market in a sector. Some investors are looking into topics that employees in a company are searching for to understand internal dynamics and predict upcoming products of the organisation.
Posts and comments on social media platforms, blogs, management communications, and more can provide great insights into public sentiment regarding a product or service. This includes online reviews on various platforms. Apart from current trends and brand virality, today’s traders also use Twitter to gauge overall market sentiment in the stock market, after a particular data release.
Satellite data providers can offer information about foot traffic in stores and malls, giving traders a picture of a company’s sales growth (or decline). Satellite imagery can help track empty or full parking lots near places of interest. Similar data can be available from Wi-Fi signals and Bluetooth beacons. Geo-location data from flight and shipping trackers provide insights into global mobility and supply chains. It can also help track construction activities, agriculture yields, and oil/gas storage. This can help investors ascertain the health of sector-specific entities or the global economy.
Data provided by aggregators of credit and debit card transactions and point of sale can provide insights into the sales figures of a company. This data set can be highly accurate when including a large number of users from similar backgrounds. Payment data cannot be tracked extensively, but it can also provide investors with a direction of where the stock might be heading ahead of public company disclosures.
This includes electronic receipts obtained on product deliveries or purchase of services. It also comprises responses to opt-in emails, invoices, and other forms of such data which can help track retail revenues. This data can be accurate, but smaller than the datasets offered through transaction data.
Considering that a huge number of companies offer apps today, their usage data can provide insights into the business. This includes data like the number of downloads, the number of times per day the app is used by a user, and other quantitative and qualitative data. This can provide information about projected revenues, especially for companies in eCommerce, food delivery, and streaming services.
Three main sources of data include individuals, sensors, and businesses.
Individuals generate huge amounts of information, like social media transactions, foot traffic, payment information, website interactions, and more. These datasets can be difficult to process. Businesses, on the other hand, provide comparatively structured data, which is useful for financial decision-making.
Finally, in the era of the Internet of Things (IoT), sensors can capture and relay important information from one device to another. These include CCTV cameras, parking lot sensors, and POS terminals.
Financial services companies can use alternative data to complement information derived from traditional sources, to serve customers in a better way. But data is mostly unstructured, and without AI and ML, it is not possible to extract meaning from huge volumes of heterogeneous data and derive correlations. For banks and asset managers, AI and ML are making it possible to structure products that promote sustainability.
For instance, ML makes it possible for investment managers to perform ESG (Environmental, Societal, and Governance) assessment from a granular perspective. ESG performance is referred to in different ways in corporate sustainability reports and filings. A uniform ESG taxonomy (derived from ESG standards) can be applied across all assessments, by machine learning models, to evaluate ESG reports clearly and consistently.
Alternative data is being used by fintech firms to evaluate the creditworthiness of loan-seeking candidates, with no prior credit history. Data from disparate sources like cash flow analysis from bank accounts, and non-financial acts like payment history of credit card bills, utility payments, and social media activity can be used for this purpose. Here again, the use of AI and ML is gaining traction. Machine learning algorithms can help to evaluate unstructured data to inform about the customer behaviour of a borrower, and also predict their ability to repay the loan.
Financial institutions and fintech companies have copious amounts of information, and many opportunities to explore these previously untapped data types. Technology is now providing them with models to predict and prepare for the future. Better data strategies can create a more resilient and efficient financial industry.