LinqAlpha’s Thematic Scoring Agent introduces a scalable LLM-based system that quantifies how companies talk about investment themes across earnings calls, enabling institutional investors to capture narrative signals beyond keyword mentions.
For decades, investment themes have guided how institutions allocate capital—shaping portfolios around structural growth stories like AI, decarbonization, or supply chain resiliency. Academic research has validated the role of thematic exposures in explaining alpha and outperforming sector-based groupings.
But thematic signal extraction has remained manual. Analysts read transcripts line-by-line, apply brittle keyword filters, and typically assess one theme at a time.
LinqAlpha’s Thematic Scoring Agent changes that. Built on our new two-stage LLM framework, the agent:
Extracts theme-relevant content from earnings transcripts
Scores sentiment at the sentence level for each theme
Supports simultaneous evaluation of dozens of narratives
This system doesn’t just detect if a theme is mentioned—it quantifies how companies talk about it.
Thematic Scoring Agent will be rolled out across our platform, supporting fund managers who track cross-cutting narratives across thousands of earnings calls. It will also be integrated into our Company Screener as a core product feature.
Our research paper, "Thematic Scoring: Quantifying Contextual Narratives using Language Models," co-authored with Alejandro Lopez-Lira (University of Florida) and Yoon Kim (MIT), is now available on SSRN.
Read the full paper → https://ssrn.com/abstract=4857726
Curious how LinqAlpha’s AI can help your team quantify narratives and surface differentiated investment signals? Contact us at support@linqalpha.com or book a demo.