“These types of strategy have exploded in recent months and by 2020, the industry is projected to be worth USD 350mn, with total alternative data spending in excess of USD 1.7bn.”
Data has become businesses’ most precious resource
Investment professionals across the industry have begun to understand the utility of alternative data and that it is forever changing the investment landscape. But what exactly is it? Alternative data (Alt Data) is information gathered from non-traditional information sources. By analysing it, is possible to gain insight beyond what traditional data sources are capable of providing. Alt Data sources are typically non-financial information that can be used to better assess the future price performance of invested assets. Alt Data is found in a company’s internal data, physical technological installations or, most commonly, via web-scraping tools, an automated method of extracting data from websites. The web-scraped data itself takes many forms, including product pricing, search trends, insights from expert networks and web traffic data of all kinds. Around 2014, a group of largely sophisticated hedge funds started operating in this new data-rich investment world, aggressively seeking information advantages. Since then, many start-ups have jumped into this business, attempting to monetise this extensive availability of data.
Why Alt Data?
We found four reasons. The first driving the need for Alt Data is related to the growth of data availability over the last ten years, thanks to advancements in technology. Roughly 800 datasets across more than 20 categories relevant to the buy-side are available today. The five most popular data types are: social/ sentiment, private company, credit card, supply chain and Web. Examples include satellite imagery used to count the number of cars in shopping centre car parks as a metric for retail sales activity; geospatial analysis for identifying the geographical proximity of competitors; and pricing data analysis for insight on everything from bundle pricing to financial rates monitoring.
The second is the pursuit of buy-side firms to ride the “low latency” world of delivering non-traditional data in order to make faster and better investment decisions, enabling them to capitalise on opportunities early and mitigate potential risks. Using Alt Data, investors can monitor how a business is doing on a weekly or even daily basis (rather than waiting for the monthly or quarterly company updates), giving them an incredible edge over other investors. For example, by employing “quantitative overlays”, such as leveraging credit-card swipe payment information, fundamental analysts can now monitor sales data against earnings estimates and forecast potential share price impact well in advance.
The third reason is based on ROI (return on investment). Alt Data is expensive and failure is often due to spending too much money and time on the wrong data firms. But if you are wise enough in selecting the right vendors, you can get a very good edge and performance ROI from your data spend.
And finally there is the so called “fear of missing out”: buy-side firms do not want to be left out of this party and are keen to glean insight on what is happening around the business. All types of investors are now embracing the use of more data and those who fail to join this revolution are likely to underperform and be left behind.
Implementing Alt Data is not only about benefits
While firms recognise the alpha generation potential of these new datasets, they face challenges like data connectivity, data cleaning, varying quality and ease-of-use. Some of the top challenges when leveraging Alt Data are lack of workflow integration, short histories, collection systems that are prone to change, information integrity and reliability as well as data protection policies. Sufficient regulatory protection for individuals remains elusive at this stage. And some of the data types being procured by hedge funds are not anonymous with respect to personal information. The GDPR (General Data Protection Regulation) in the EU, for example, has recently been adopted to strengthen and standardise the protection (anonymity) of its citizens’ personal data. The main driver behind this regulation lies in the problematic nature of the complex information management, which results in the difficult governance of Big Data.
Data must now be handled appropriately, certified and compliant with local laws, in respect of both privacy and security management. Despite these challenges, we still think there can be substantial benefits to using alternative data. Very recently, some of the leading sources of reference in pricing data and major investment firms have started offering clients a single access point with multiple, market leading alternative data providers, for finding and receiving reliable data, that eliminates costly and lengthy procurement processes. This speeds up time to value, enabling easy and efficient integration to existing systems or databases with quant investors who need only select their preferred programming languages (mainly Python). With this access point, professional investors can browse and examine quality metadata online, trial sample datasets prior to acquisition and immediately put them to use within their organisation.
How quants will start using alternative data practically
According to recent studies and surveys, on average over 80% of funds use or are expected to use alternative data. We believe that 2019 will mark the beginning of a more mature phase of this business that will probably last between five to seven more years, and where the early majority of quants will start incorporating alternative data in their businesses. We think the hottest category of alternative data that we will emerge in the coming years could be consumer transaction data, where the buy side gets the most return on investment. The most prominent emergence will probably be the increase in demand for employment data. The first significant challenge for quants dealing with Alt Data is backtesting - having a mechanism to evaluate the effectiveness of a trading strategy by running it against historical data. Today, backtesting on Alt Data is very difficult, simply because we do not yet have good and sufficiently broad historical data.
This also creates an urgent need for advanced analytics skills and capabilities to process this vast amount of data. As a result, data teams are growing everywhere: the number of Alt Data full-time employees on the buy side - mainly data scientists and analysts - has grown ~450% in last five years.
To fully reap the benefits of this new investment world of abundant data, machine learning will play a central role in identifying patterns and correlations, managing risk and transforming this knowledge into actions that allow buy-side firms to gain a competitive advantage. TensorFlow and Scikit-learn in Python are the prevailing big data analytics used in asset management today. Nearly all major industry players are now filling their quant teams with physicists and data scientists, providing them with access to the data and turning them loose, expecting them to come up with something brilliant. We do not agree with this kind of approach: quants will certainly fail if they do not use a more “conservative approach” that keeps all decisions and management in the hands of experienced finance professionals. Discerning valuable information from noise requires extensive real financial experience – not maths and statistics alone.
The goal then, is not to replace finance professionals with mathematicians, but to evaluate an experienced investor’s hypothesis and test it with machines to realise superior, explainable and more actionable information. Intelligently mining this data is critical to avoid getting lost in machine-made interpretations. As we said last year, it is not mans versus machine, but experienced finance professional with machine.
To conclude, we believe the big data revolution will usher in a new era of investing that will ultimately benefit markets by lowering day-to-day volatility, producing fewer surprises and empowering investor confidence, enhancing market stability. Big data holds incredible promise to facilitate so many investment decisions. Companies capable of extracting value from their data will enjoy a competitive advantage – as long as they ensure they distinguish what is of value from what is not.