Companies across industries are leveraging disruptive technologies to innovate and better serve their clients. The investment industry is no different. In this Q&A, Portfolio Manager Peter Carpi and Investment Data Science Director Tom Bok will:
- Explore how data science is increasingly providing investors with differentiated insights on opportunities and risks;
- Discuss the value of robust collaboration among investors and data scientists; and
- Highlight a few examples of applied data science in the small-cap equity space.
How is Data Science Expanding the Tool Kit Fundamental Investors Use to Generate Alpha1 for Clients?
Tom Bok: The intersection of fundamental investing and data science offers investors access to new alternative data sets and a more nuanced understanding of existing data. There is an enormous and expanding amount of data available to try to leverage into an informational edge for our clients. For investors at Wellington, the truly limited resource is time, not data.
Our Investment Science Team, therefore, works to create data science tools that fit into our investors’ processes to help guide their focus. We aim to direct their expertise to where it will be most fruitfully employed through cleaner and more impactful data. We think the value of better interpretation of data in the investment process is underappreciated by much of the market.
Peter Carpi: As an investor, I appreciate how data science offers key inputs into my idea generation and due-diligence processes. It also plays an increasing role in portfolio construction and risk management. For example, as a small-cap2 investor, my universe covers approximately 3,000 stocks between the Russell 2000 Index3 and the Russell Microcap Index.4 My process focuses on inflections in business momentum, and we’ve found that there’s an enormous amount of untapped information in company meeting transcripts. But with over 25,000 transcripts per year, it was impossible to read and analyze them all.
Fortunately, working with our Data Science Team, we were able to build a proprietary Natural Language Processing (NLP)5 tool that we can train to automatically scan tens of thousands of transcripts in a matter of minutes for signals of key inflections. For instance, these 3,000 companies have an average CEO tenure of five years, meaning there are approximately 150 CEO changes every quarter. Our NLP tool can help me see these potential transitions—which represent a possible inflection point for any business—across this large and inefficient universe of stocks. This collaboration scales up my process in a way that would not be possible without data science, creating more opportunities to generate alpha for clients.
Our NLP Tool in Practice
Source: Wellington Management. For illustrative purposes only.
This is an example of an output from the NLP tool that tracks management commentary around a semantic profile for “Adding Capacity.” We are able to drill down into each time the company spoke publicly, such as the commentary examples listed below, to see the context of each mention. “Adding Capacity” is one of roughly 50 semantic profiles we have created to track management commentary around complex concepts with our NLP tool. Other examples include CEO changes, litigation, sales hiring, and environmental-social-governance (ESG).6 Some captured mentions might include the following examples:"
– "...This contractor is negotiating a lease for an additional space needed and plans to install equipment to increase capacity by 50% by the end of 2018…”
– “…We have a growing and diverse product portfolio to support that growth. We have been ramping up our manufacturing capacity for growth in our leading products…”
– “…We see the need and opportunity to use our capabilities and resources to expand into new markets, what we’re calling dynamic opportunities…”
What Differentiates Wellington's Approach to Data Science?
Tom: There are a couple of key differences that make Wellington’s approach to data science distinct. The first is in the depth of our collaboration. Wellington has world-class investors and researchers that know the industry inside and out. So, our team is not just testing every possible data point. We’re being guided by seasoned investors’ perspectives on the market and what is and is not meaningful data.
The second is in the amount of resources we have dedicated to our investment science efforts. Significant advancements in computing power and open-source languages7 have allowed substantial developments in data science that are now accessible to investment firms that have the expertise to exploit them. Importantly, there are many off-the-shelf tools and data sets available to the market. But what differentiates Wellington is our desire, ability, and resources to create custom tools to best fit our investors’ needs. We think firms buying off-the-shelf tools and data sets aren’t going to be able to generate the same informational advantages available through bespoke approaches.
Collaboration between investors and data scientists is critical in making these custom data-science tools as impactful as possible.
Peter: Collaboration between investors and data scientists is critical in making these tools as impactful as possible and in assessing the value of each alternative data set. For instance, we recently analyzed a US Customs and Border Patrol data set that shows shipping manifest data. We thought this may have offered useful insights on company supply chains. But, working together, we identified issues with the data such as missing air-freight and trucking shipments, language challenges, and the use of shell corporations. Though providers are selling this data to the market, together we concluded that the data set was not clean or complete enough to offer compelling investment insights.
Wellington works to combine deep data resources, custom-designed tools, and robust collaboration to create differentiated data-science capabilities and investment insights. Crucially, we think our scale also offers unparalleled management access to try to confirm what these alternative data sets are telling us.
Much of the underlying data may be available to all investors, but we believe valuable investment insights are only possible for those who can turn that data into knowledge. Investors need the right tools to convert raw data into information and the skill sets to combine and analyze like-themed information from multiple independent sources to then extract knowledge from information (FIGURE 2).
Transforming Data Into Knowledge by Combining Like-Themed Information
Source: Wellington Management. For illustrative purposes only.
What Does the Collaboration Between Investors and Data Science Look Like in Practice?
Peter: We often reach out to our investment science partners to work backward and systematize part of our existing investment process to be more efficient and accurate at finding opportunities. For example, I was looking for ways to better capture the hiring of salespeople as I think this signals a company’s belief in market demand, in their confidence to onboard additional known expenses, and that their supply chain is in place to support new volume. I reached out to our data science colleagues hoping to create a custom tool to track salesforce-specific hiring.
Tom: Investors often come to our team with an idea like that to try and automate a time-intensive process that requires a bit of human intuition. In this case, Peter needed to identify and count specific types of sales positions that he hypothesized would reveal important insights into a company’s prospects. However, the job titles for these particular positions are inconsistently labeled and vary by company, region, and industry. So, our team developed this tool with Peter through a theory-driven process that involved us going back and forth to profile the essence of these particular jobs. Together, we added a function to our existing custom tool to extract these job postings from aggregate jobs data, which he can now see in his dashboard.
Peter: As I look to find new investment ideas, I now have an evolving set of tools that track a variety of investment concepts that might signal impending inflections in business fundamentals. These include changes in jobs hiring, regional hotel-occupancy rates, patent filings, competitive dynamics, and new-product launches, among many others.
Which Opportunity Sets Are Best Positioned to Leverage Data Science Tools?
Peter: Data science allows investors to extract new insights from data sets, and I believe it is therefore particularly effective in less-efficient markets such as small-cap equities. If an investor is analyzing new data on one of the world’s largest tech companies, it is unlikely that this will generate a unique insight that many other sell-side or buy-side analysts have not discovered. Small caps have lower research coverage that makes it a less-efficient asset class. With fewer eyes on more companies, we think small cap offers greater opportunities for insights from alternative data sets.
Data science allows investors to extract new insights from data sets, and I believe it is therefore particularly effective in less-efficient markets like small-cap equities.
Importantly, many small-cap investment firms may not have the resources to create custom tools, which we think could make these data inefficiencies more enduring.
Tom: In our view, investors at the firm who cover less-efficient asset classes—such as small caps, ESG, and global impact—stand to benefit the most from alternative data sets. This is similarly true for investors that cover countries and markets where data is more opaque. It’s easy for investors to find data on US consumers or well-known US themes, but there are many less transparent areas where alternative data sets can be extremely useful. This is not to say that our data and tools cannot benefit large-cap US investors. On the contrary, we can evaluate or provide a signal for an investor’s specific theory or hypothesis, helping to enhance or challenge their conviction.
How Does Wellington Work to Expand These Capabilities to New Investors and Markets?
Tom: The data science team conducts outreach to investors through several channels, including email, educational seminars, and more recently, our new Mosaic platform. Wellington is currently in the midst of a firmwide technological transformation centered around Mosaic, a new proprietary desktop software application that will be the centralized portal for all of Wellington’s investment research and dialogue. Here, investors can access the full breadth of our data science capabilities—including raw data sets, clean dashboards, and research notes—in an intuitive and user-friendly interface. We think this will help maximize the value of data science across everything from investment idea generation to portfolio construction and will continue to enhance the collaboration between our teams.
1 Alpha is the measure of the performance of a portfolio after adjusting for risk. Alpha is calculated by comparing the volatility of the portfolio and comparing it to some benchmark. The alpha is the excess return of the portfolio over the benchmark.
2 The term small cap is used to describe companies with a relatively small market capitalization, generally a company with between $300 million and $2 billion market capitalization. Source: Investopedia
3 The Russell 2000 Index measures the performance of the small-cap segment of the US equity universe.
4 The Russell Microcap Index measures the performance of the microcap segment of the US equity market and makes up less than 3% of the US equity market.
5 Natural language processing is a subfield of linguistics, computer science, and artificial intelligence concerned with the interactions between computers and human language, in particular how to program computers to process and analyze large amounts of natural language data. Source: Wikipedia
6 ESG is an acronym for enviromental, social and governance standards used by socially conscious investors to screen potential investments. Source: Investopedia
7 Open-source language refers to computer-programming languages that are generally considered non-proprietary (depending on their open-source license provisions), and can be modified or built upon in a manner that is open to the public.
Important Risks: Investing involves risk, including the possible loss of principal. Security prices fluctuate in value depending on general market and economic conditions and the prospects of individual companies. • Small-cap securities can have greater risks and volatility than large-cap securities. • Focus on investments that involve climate change opportunities or sustainable and environmental initiatives may result in foregoing certain investments and underperformance comparative to funds that do not have a similar focus. ESG factors are not the only factor considered and as a result, certain companies in which the Fund invests may not be considered an ESG company or have a high ESG rating.
The views expressed herein are those of Wellington Management, are for informational purposes only, and are subject to change based on prevailing market, economic, and other conditions. The views expressed may not reflect the opinions of Hartford Funds or any other sub-adviser to our funds. They should not be construed as research or investment advice nor should they be considered an offer or solicitation to buy or sell any security. This information is current at the time of writing and may not be reproduced or distributed in whole or in part, for any purpose, without the express written consent of Wellington Management or Hartford Funds.