AlphaTech Column: The Difficulties of Valuing Research
For both the buyside and sellside, valuing research is both a management and a measurement challenge, considering an estimated 40,000 pieces of research are created each week. MIFID II adds another complexity to this process. This week's column reviews InvisageAlpha, a research and data aggregation platform.
The challenges of information management in the institutional investment process were addressed in last week's AlphaTech column by Bill Stephenson.
A challenge made more daunting with the proliferation of alternative data sources and quantitative approaches to portfolio construction techniques.
In addition, the Street continues to produce large volumes of research daily (40,000 pieces per week by some estimates), which creates its own challenges in terms of consumption, interpretation and analysis - making it increasingly difficult to gain unique insight from all of this content expressed in an alpha-generating portfolio. Recent regulatory measures such as MiFID II also require asset managers to monitor the performance and impact of research on their investment performance to justify the value ascribed to their spend with research providers.
Estimates of the total market for global investment research vary. A recent story by the Financial Times puts it at roughly $14 billion annually, and a recent report by the CFA Institute notes both a reduction in research consumption and associated budgets since MiFID II.
The opacity surrounding quantifying the value of research, ideally incorporating the quantifying of alpha derived from research, adds a measurement challenge to the management challenge described above. This problem persists for both the buyside and sellside.
Applying Innovation to Attribution
There are a number of Research Management Systems in the market helping both buyside and sellside firms deal with these tasks. Yet many tend to focus on the low hanging fruit of counting “clicks,” or interactions, and also embed an inherent popularity-driven “selection bias” in the broker review and vote processes.
At AIR Summit, we think the interesting emerging innovation in this area incorporates data analytics and technologies such as machine reading, Natural Language Processing and AI to extract alpha-generating insight.
The goal is quantifying the relative value of each piece of research individually, as well as in aggregate across various sources [e.g.: an analyst, bank or broker] and groupings such as sector or theme. Integration with the portfolio construction process across various forms of data and research and multiple asset classes is a logical and valuable application.
Invisage: Identifying Research Alpha
InvisageAlpha was a company that presented at AIR Summit 5.0 in 2019 and is a member of our AlphaTech community. Invisage provides a platform for aggregating research and data across multiple sources, and then applies proprietary NLP and analytics models to capture the alpha performance of these sources.
This results in an objective evaluation and attribution of performance, which helps both the buyside and sellside transparently assess research performance. Further, this represents an interesting benchmarking opportunity for users which the company displayed in a recent report where it explored the relative performance of providers during the ongoing Covid-19 pandemic.
Interestingly, they note that during Covid-19 research volume has increased by 240% year-on-year over the analysis period. Clearly, this speaks to the management and measurement challenges.
Broad Data and Asset Coverage Supports In Depth Process Integration
Another key capability is the breadth of data and securities Invisage can cover. They can apply their process to traditional fundamental research as well as others: macroeconomic research, alternative data, quantitative research, analyst calls, etc. And it is multi-asset as well, covering equities, fixed income, commodities and currencies.
Invisage currently supports more than 40,000 listed equities worldwide and a comprehensive universe of government and corporate bonds. “Out-of-the-box,”, the platform includes research coverage of more than 6,000 US stocks provided by 170 brokerage firms and banks.
The combination of applying their machine reading and NLP capabilities to both fundamental and macroeconomic research is also quite compelling. This allows portfolio managers and analysts to connect macro trends to portfolio level holdings and investment strategies.
Finally, this supports a process of moving along a continuum from research extraction to idea extraction. By applying metadata enrichment to the source data, the platform is able to provide a discovery process related to things like conviction, consensus, investment strategies or themes as examples.
Signal or Platform?
As we continuously discover and curate emerging innovation in capital markets at AIR Summit, often times it seems easy to label firms in one of two ways – data/signal providers delivering new forms of raw ore, and platforms looking to bring efficiency and automation to existing legacy processes.
Of course, it is more complicated and nuanced than that and I am not suggesting one is superior or more interesting than the other. But I think it will be interesting to watch the emerging firms that are able to ingest as much of the former while providing the latter.
We know from conversations with our AIR members that the giant step function improvement is nice to ponder but often impractical, and a more incremental approach is often preferred. That sweet spot cross-section of modernizing the investment process while incorporating new and ever-expanding information is an area that is exciting.