Narrative tracking

Tracking the “AI bubble” across media, earnings calls, and SEC filings

May 2025
shape black right

Market-moving stories often begin long before the data shows up in earnings reports or stock prices. With Bigdata.com’s Narrative Miner, financial analysts can track the evolution of high-impact narratives, like concerns about an AI investment bubble, across news media, corporate communications, and regulatory filings.

This allows institutional investors, strategists, and risk teams to separate market noise from real narrative momentum and to quantify it across time and source type.

Tracking the “AI bubble” narrative across financial disclosures

Objective

Uncover how the “AI bubble” story emerged in the media, filtered into earnings call language, and eventually appeared in SEC filings, along with when and how intensely each channel picked it up.

Why it matters

Narratives drive sentiment. Sentiment drives capital flow. Knowing when and where a story peaks is critical for:

  • Risk timing and hedging
  • Earnings season prep
  • Portfolio narrative exposure
  • Strategic positioning in thematic ETFs or sector allocations

The solution with Bigdata’s narrative miner

  1. Define a custom narrative theme In this example, the narrative includes statements like:
    • “AI valuations detached from fundamentals”
    • “CEOs acknowledge AI implementation challenges”
    • “VCs funding unproven AI at unsustainable levels”
  2. Mine across three financial document types Run searches and LLM-based labeling on:
    • News media (e.g. MT Newswires)
    • Earnings call transcripts
    • SEC filings (10-Ks, 10-Qs)
  3. Visualize how narratives evolve Bigdata’s miner tracks intensity over time and source. For instance:
    • Media coverage peaked in April 2023
    • CEOs spoke about bubble concerns on earnings calls in August 2024
    • Regulatory filings showed heightened concern by March 2025
  4. Extract insights and timing gaps You can quantify lags—e.g., SEC filings trailed peak news coverage by ~443 days—and isolate leading vs. lagging signals.

What we found

  • Top narrative in news: “Current AI investments may not generate predicted financial returns”
  • Top narrative in earnings calls: “Tech CEOs acknowledge AI implementation challenges amid high expectations”
  • Top narrative in SEC filings: “Current AI investments may not generate predicted financial returns”
  • Total mentions:
    • News media: 12
    • Earnings calls: 24
    • SEC filings: 104
  • Average lag between news peak and filing peak: 443 days

Technical stack

  • Bigdata narrative miner
  • OpenAI GPT-4o-mini for LLM classification
  • Document types: News, transcripts, and filings
  • Tools used: Python, Pandas, Plotly, Gaussian filters for trend smoothing

Ideal for

  • Buy-side research teams tracking emerging investment risks
  • Sell-side analysts preparing for narrative shifts
  • Compliance teams monitoring disclosure trends
  • Strategy desks modeling fund flows around thematic exposures

Practical applications

  • Narrative-driven trading signals Combine narrative intensity with price/volume to trigger alerts.
  • Earnings season prep Get ahead of what management teams are likely to say—based on prior media build-up.
  • Policy & regulatory horizon scanning Spot when speculative hype becomes an SEC concern.

Start your own narrative analysis – follow our tutorial to start tracking your own market themes.