Masters in Business with Barry Ritholtz

Episode: Using AI to Find Investing Stories with Perscient Co-Founder Ben Hunt

Date: January 9, 2026
Guest: Ben Hunt (Co-founder & President, Perscient; Author, Epsilon Theory)

Overview of the Episode

In this episode, Barry Ritholtz sits down with Ben Hunt, an academic turned fund manager and tech entrepreneur, to explore how artificial intelligence and narrative theory are reshaping the investing landscape. Hunt, known for his work at Epsilon Theory and as co-founder of Perscient, discusses how new data tools and AI techniques allow investors to map, measure, and even anticipate market-moving narratives in real time. The conversation moves from Hunt’s unconventional career path to technical aspects of narrative analysis, the evolution of storytelling in markets, and practical uses for investors and institutions.

Key Discussion Points & Insights

Ben Hunt’s Unconventional Path to Investing (03:18–07:20)

• Academia was a “way station” for Hunt—he pursued a PhD and was a tenured political science professor but always had an entrepreneurial “bug.”
• Started companies during grad school and professorship: “It is a bug, it’s not a feature. You can’t help yourself.” (03:32)
• The world of investing appealed to Hunt’s love of games and problem-solving. “Markets—it’s the biggest game in the world, for sure.” (07:20)

Narrative Theory and Its Influence (07:50–12:17)

• Hunt’s academic focus on game theory and narrative theory shaped his approach to investing.
• The moving from structured numerical analysis to analyzing unstructured narrative data is central in both political science and finance.
• “Words and the stories and the narratives that are told to us... politicians have known this forever.”
• “This is at the heart of all of the generative AI and the AI.” (11:51)

The Rise of Narrative in Financial Markets (13:00–22:17)

• Post-Great Financial Crisis, Ben Bernanke and the Fed engaged in “forward guidance”—using words as a powerful policy tool.
• “We started using our words and coordinating our words to change the market, to change market behavior.” (19:58)
• [On CEOs:] “What makes for a good CEO is can you tell the story, can you tell the narrative of your company to get a multiple? Because a multiple is a narrative. A multiple is a story.” (21:17)
• Storytelling’s centrality: from Fed policy to CEOs like Benioff (Salesforce), Jobs, Hastings, and Ellison.

The Modern Media & Narratives Ecosystem (24:15–30:11)

• 24/7 news cycles and mobile technology amplify storytelling and narrative’s influence.
• “There’s not enough hard news to fill the time…so what takes the space? Opinion. Story takes the place.” (28:29)
• “It is absolutely a neurotransmitter addiction... it is so important to keep that from our kids.” (29:11)

[Segment Break at 30:11]

Next Segment: How Perscient Maps Narratives with AI

How Perscient Uses AI to Identify Narratives (32:21–44:57)

• Global Scale and Data Sources:
  • Accesses “everything that's published publicly in the world” (Dow Jones, LexisNexis, etc.), in multiple languages.
  • “We’ve processed over 200 billion tokens... [in recent months].” (34:20)
• From Math to Meaning:
  • “All the LLMs are linguistic calculators... calculations here are not particularly complicated, but you have to do them at enormous scale.” (34:11)
• Identifying and Tracking Narratives:
  • It’s not just “bullish on financials”—they track specific, fine-grained stories and how their prominence waxes and wanes.
  • “There are only about a dozen stories for why you’re bullish on something... management change, top line growth, opportunity, upcoming catalyst…” (39:03)
• “The stories, we think that they’re amorphous and variable. The fact is... their meaning is amazingly constant over time.” (40:46)

Methodology: AI with Human Context Engineering (42:26–45:51)

• Don’t trust LLMs for open-ended answers. The key: context engineering—humans set the boundaries, the “scaffold” and the dataset.
• “The secret to using AI successfully is to take this magic genie… and you stuff it into that bottle. You do not let it out, you constrain it dramatically.” (42:50)
• AI is used “as an operating system”—with controlled prompts, input data, and output validation for consistency.

Descriptive vs. Prescriptive Narratives (46:08–48:22)

• “We tell descriptive stories... and we also tell prescriptive stories. The Fed should be hawkish, the Fed should cut rates.”
• Perscient focuses on identifying stories designed to move public opinion: “We can boil that down into... the stories that are trying to tell the reader how they should think or how policy should go. That we find has a lot of predictive capability to it.” (47:16)

Practical Applications: For Investors & Beyond (51:34–65:33)

• For Investors & Fund Managers:
  • “My goal is… here is data that I think you’ll find useful. The efficacy… is up to you.”
  • Product for financial advisors: helps anticipate client concerns, matches stories to their portfolios, and arms advisors with responses.
• Beyond Markets—Policy, Politics, Brands:
  • Used the system to track Russian media narratives before the Ukraine invasion, predicting a full-scale war before it hit Western coverage. (62:00)
  • Policymakers and corporations use narrative data; e.g., sports teams on stadium campaigns or brands gauging public response.
• “Once you’ve got a tool where you can actually measure them and visualize them... [it’s] like when they invented the microscope…” (65:08)

Evolution of Narrative Analytics Technology (53:58–58:39)

• Early methods (“handcrafting” language models) had limited coverage; modern LLMs expanded the net 100x over.
• Now process global, multi-language data, distinguishing between domestic and international narrative variants (e.g., Chinese news about luxury demand vs. Western coverage).

Creating Alpha & Distinguishing Signal from Noise (58:01–61:49)

• “Not sentiment analysis”—different from high-frequency, word-counting approaches.
• Track underlying stories that drive behavior rather than only word frequency or sentiment shifts.
• “We’re able to track the actual stories that drive behavior.” (59:58)
• Hunt: “My goal is not to do the trades. My goal is… I want to sell the picks and shovels…” (60:17)

Narrative Dynamics in Politics & Society (70:08–85:52)

• Example: “The end of Pax Americana” emerged domestically and internationally as narrative data shifted post-2016.
• Nowcasting, not forecasting: “I want to nowcast is what I want to do, not forecast.” (74:48)
• Immigration: unexpected pro-immigrant narrative growth (post-Oct 2025), despite policies and loud anti-immigrant rhetoric (76:02–77:25).
• “This country is actually quite pro-immigrant. I know that sounds...” —Ben Hunt (77:25)
• Macro stories tracked: de-dollarization, gold as insurance, melting “iceberg” of foreign capital in Treasuries.
• “The story that America has raised the world’s wealth... that story is non-existent today.” (85:18)

Emerging & Dormant Narratives: Credit Risk Example (87:28–89:55)

• “Every day you have a new bank CEO come out and say, no, no, everything's fine. For every interview that says it's fine, you're getting two articles… saying ain't fine.”
• Private credit risk stories now at “an order of magnitude” higher than at any time in last decade, “rarely see” narrative momentum fade once established.

Notable Quotes & Memorable Moments

• On narrative as a market force:
  “What makes for a good CEO is can you tell the story, can you tell the narrative of your company to get a multiple? Because a multiple is a narrative. A multiple is a story.” (21:17, Ben Hunt)
• On narrative analysis vs. truth:
  “I don’t know what the truth is... and I don’t think it matters.” (50:27, Ben Hunt)
• On using AI responsibly:
  “The secret to using AI successfully is to take this magic genie… and you stuff it into that bottle... You do not let it out.” (42:50, Ben Hunt)
• On narrative survival:
  “Stories have, they never die. They’re waiting for a new force to… give them the next version.” (85:52, Ben Hunt)
• On predictive vs. present focus:
  “I’m not predicting, I’m observing... I want to nowcast, not forecast.” (74:43, Ben Hunt)
• On research process evolution:
  “We’d look back... how did we miss all these other fish? And the answer was... the small language model we were constructing. We’d made a small net.” (55:30, Ben Hunt)
• On advice for young professionals:
  “Build your intellectual capital [early]. Once you enter, particularly the investment world… you are spending your intellectual capital, you’re not gaining intellectual capital.” (95:58, Ben Hunt)
• On the danger of seeking the capital-A ‘Answer’:
  “I was looking for the answer with a capital A as opposed to a process with a capital P… There’s real magic in understanding that.” (97:34–98:54, Ben Hunt)

Timestamps for Important Segments

• 03:18 – Ben Hunt’s early career, entrepreneurial bug, and transition to finance
• 07:50 – Game theory and narrative theory influences
• 13:00 – The Fed’s use of forward guidance; narrative’s new financial power
• 21:17 – The rise of CEO storytelling as corporate skill
• 24:15 – Transformation of media, social, and mobile amplifying narrative
• 32:21 – Perscient’s approach: ingesting global data for narrative analysis
• 39:03 – Identifying and cataloging the finite set of investing stories
• 42:26 – Context engineering: Using AI for measured, human-directed narrative analysis
• 47:16 – Descriptive vs. prescriptive stories; predictive use
• 53:58 – Leap in language model capability with AI
• 58:01 – Distinction from sentiment analysis; seeking true “story” signal
• 62:00 – Policy and corporate applications (e.g., Ukraine invasion, brand narratives)
• 70:08 – Using narrative data for global and domestic politics; Pax Americana
• 76:02 – American attitudes on immigration and their media reflection
• 87:28 – Early warning in private credit risk stories
• 95:58 – Advice to young professionals: build, then spend, your “intellectual capital”
• 97:34 – The shift from seeking “the answer” to trusting process and probabilities

Takeaways for Listeners

• Narratives are now central to both financial markets and politics: Technology and media cycles have made narrative control a dominant tool for policymakers and CEOs.
• Modern AI approaches allow meaningful, real-time mapping of how these stories move: But human direction is always required for effective, reliable analytics ("context engineering").
• The future of investing, public policy, and even brand management increasingly relies on understanding narrative momentum.
• Hunt’s final advice? Prioritize building your intellectual capital early, never risk “all in,” and focus on process—not perfection.