Almost every industry — from consumer goods to financial services — is centered on one thing: the commoditization of data. Media streaming platforms deploy algorithms to chew through data created by millions of users worldwide to curate a library of must-see movies and TV shows, tailored to individual's interests. The same principle is now starting to apply in asset allocation.
Through alternative data and machine learning, investors are beginning to deconstruct and recategorize asset choices and traditional asset allocation decisions in order to build customized portfolios. Customization is not a buzzword and does not have to mean more expensive. It's simply the ability to design portfolios from the bottom up, free of traditional constraints, starting with the client's objective.
As the investment landscape becomes less constrained, continuing to categorize securities rigidly in terms of their traditional sectors has begun to feel anachronistic for those investors that seek exposure to trends driven by technology or shifting demographics. This might include allocating capital to the area of robotics and automation — companies in the investible universe that offer products and services through the robotics and automation value chain — or even companies in the "internet of things" ecosystem. Currently, no single sector or industry exists that spans these opportunities, highlighting the need to break down categorization concepts in order to create new investment strategies. To operate outside of the characteristic mapping that occurs in traditional asset allocation decisions, a multidisciplinary approach that draws on expertise in strategic asset allocation, security analysis and selection, and originating insights from alternative data is essential.
Take timber, for example. Traditional classifications would confine it to the materials sector — namely the containers and packaging, and paper and forest products subindustries. Such companies would be involved with the production of paper, paper-based containers and packaging, and the production of timber, lumber and other wood products. However, within these categories, there may be too few companies directly linked to timber from which to construct a portfolio. Specialist fundamental investors with intimate knowledge of the timber value chain would be able to intuit related companies in other sectors, such as home improvement retailers. But when doing this at a global level it is highly possible to overlook opportunities — the MSCI ACWI Investable Market index contains 8,675 constituents. Nowadays, however, investors can overcome this issue through the systematic processing of data at scale.
Alternative datasets offer a different starting point from which to construct a comprehensive universe. Financial statements, earnings call transcripts, news data, industry reports and analyst notes can be parsed by using text mining techniques to understand what words and terms correlate strongly with certain companies. The words and terms that feature often are ranked more highly and then cross-referenced against stocks from other sectors to map connected companies within the timber value chain.
Once the value chain has been defined, the universe is optimized by applying fundamental analysis and quantitative techniques to validate the machine-learning mapping — identifying the causal links that connect the underlying companies and eliminating spurious correlations and false inferences. Fundamental analysis and quantitative inputs also allow for the integration of environmental, social and governance factors through the stock selection process, depending on the investor's requirements. In the case of timber, this could include considerations around timberland management from an environmental and societal perspective, the impact on or of climate change, and whether activities are conservation-based.
Investors need to be mindful that it may be possible to read too much into data. In some areas, the efficacy of certain signals is yet to be tested. But the fact remains that we have access to signals that we didn't 10, five, or even two years ago as data proliferates. Do the linguistic choices of management teams indicate a lack of confidence in business performance? Does a company's reporting tally with crowd sourced reviews? Is a company's sales data consistent with information from satellite images or shipping data?
At a company level, BP PLC, for instance, has heavily invested in technology throughout the entire production process, including at oil rigs where equipment is fitted with sensors to relay real-time information on production and flag potential maintenance issues. Seabed robots are being deployed to collect information and track operations conducted underwater. Such information is valuable and, where it is accessible, it can provide an advantage to investors who are able to analyze it.
Whether an information edge is garnered through on-the-ground assessments of company fundamentals, supply chain dynamics or structural themes — or, indeed through the systemization of insights gathered from diverse alternative data — these inputs collectively serve to build a fuller picture of a company's long-term prospects than any of these factors would if viewed in isolation. Today's world is a complex mesh of rapidly evolving societal, demographic, corporate, technological and geopolitical trends and, consequently, the investment landscape has never relied on a greater number of relevant inputs. We have the tools to process them and access to new data through which to interpret them. Portfolios that rely only on traditional categorizations and data sources are missing the bigger picture and the bigger opportunity.
Jason Williams is a portfolio manager/analyst on Lazard Asset Management's equity advantage team, London. This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines, but it is not a product of P&I's editorial team.
