Bernstein: Reflexivity supports investment research with continuous monitoring and a knowledge graph

Why does investment research need continuous monitoring?

Jan Szilagyi, Reflexivity’s co-founder and CEO, believes investors face the challenge of information being scattered across different data providers, asset classes and formats, making it difficult to continually filter and connect useful signals.
The platform continuously scans earnings reports, company announcements, news and macroeconomic changes, proactively flagging events that could affect securities or portfolios and analyzing their potential impact.

The platform’s knowledge graph connects approximately 40 thousand assets and more than a dozen financial data sources. It incorporates relationships such as supply chains, industries, regions and macroeconomic exposures into its analysis to track the indirect effects of events on other assets.
For example, after a company reports earnings or encounters operational problems, the platform can analyze the potential impact on suppliers, customers, competitors and related industries. In a power-grid modernization example, Szilagyi said the platform followed the supply chain and company disclosures to identify a Taiwanese transformer manufacturer with constrained capacity and global market exposure.
After users upload their holdings or watchlists, the platform prioritizes relevant updates; recommendations are further adjusted as users conduct research and use the platform.

The platform can identify companies related to a theme and construct baskets of stocks. In an AI infrastructure demonstration, the screening extended to supply-chain, data-center and component companies, with a rationale and source documents for each constituent; users can also refine the results by market capitalization, region and other criteria.
Investors can set scenarios such as interest-rate shocks and commodity price changes to see potential portfolio outcomes and contributions to downside risk. They can also replace holdings and rerun the analysis, or use company fundamentals and micro-level changes to test macroeconomic views.
A standard license costs approximately 6000 US dollars per user per year and includes the datasets integrated into the platform; coverage is strongest in the United States, Europe and Japan. Customized enterprise deployments are priced separately.

Szilagyi believes AI allows investors to assess more opportunities without changing their investment process and may improve risk-adjusted returns through broader opportunity capture and diversification; human judgment remains important when new situations lack historical precedent.
As customers pay more attention to inference costs, the company controls spending by distributing tasks among multiple models and caching repeated computations. The company says query costs have fallen by approximately 70% over the past three months while analysis quality has been maintained.