SPONSORS

To keep AI Street growing, I’m bringing on sponsors that align with what we cover: AI, markets, and finance.

If you want your brand in front of professionals at JPMorgan, Citadel, BlackRock, Skadden, and McKinsey, get in touch for sponsorship details: [email protected].

INVESTING

JPMorgan Bets $500M on Crowdsourced AI Hedge Fund

JPMorgan Asset Management is allocating up to $500 million to Numerai, a San Francisco hedge fund that doesn’t look or operate like a typical quant hedge fund.

It’s the biggest institutional bet Numerai, which uses crowdsourced AI models to trade stocks, has landed in its 10-year history. The investment from the largest U.S. bank by assets shows that investors are willing to invest in money managers with unusual business models.

Numerai doesn’t have a traditional research team. Instead, it runs a public tournament where thousands of data scientists around the world submit anonymized model predictions. The most accurate models are rewarded; inaccurate ones lose part of their stake. The best signals are blended into a single machine-learning model that trades global equities.

The fund returned 25.45% net last year, with a Sharpe ratio of 2.75, reflecting excellent risk-adjusted returns. The $500 million will be deployed over the next year, potentially more than doubling Numerai’s current AUM, which is roughly $450 million.

“Traditional quant funds have to specialize in their way of doing things,” Richard Craib, the fund’s founder, said in a press release. Numerai is permanently open to new ways — tree ensembles, transformers, or signals sparked by LLM reasoning. That openness is our edge.

Numerai pays contributors in Numeraire (NMR), a crypto token it launched in 2017. Data scientists submit predictions and back them with NMR. If the model performs, they earn more. If not, they lose part of their stake. The setup gives contributors skin in the game and tries to keep incentives aligned across an anonymous global network.

After the allocation news broke, NMR surged nearly 40% intraday. Numerai later announced a $1 million token buyback and shared that it had recently hired engineers from Meta and Voleon.

The firm launched back in 2015 and has taken a long-game approach, building its system with backing from Union Square Ventures, Howard Morgan, and Paul Tudor Jones.

Takeaway: It’s hard to overstate how unusual it is to have a hedge fund leverage the wisdom of crowds to make stock bets and pay contributors in cryptocurrency. This isn’t Ben Graham’s money manager.

As AI tools get cheaper and more widespread, Numerai may benefit from a growing pool of skilled data scientists / amateur investors willing to submit models to its tournament.

NYC

AI Street Meetup

As mentioned last week, I’m in New York Sept. 19 to attend Cornell’s Future of Finance & AI conference.

I’m putting together an AI Street drinks meetup on Sept 18 at 7:00. If you’d like to attend, please add your email to this google form and select what part of town works best for you. Hope to see you then!

RESEARCH

The AI + Finance Papers Making the Rounds on Wall Street

Surprisingly, I get the most LinkedIn traction with write-ups of academic finance papers.

Last week, I highlighted research from BlackRock on building AI agents for stock picks, which garnered a lot of engagement.

A few readers have asked for more of this kind of coverage. Here’s a sampling of a few I’m reviewing:

• DBOT: Artificial Intelligence for Systematic Long-Term Investing

• FinReflectKG: Agentic Construction and Evaluation of Financial Knowledge Graphs

• Language Model Guided Reinforcement Learning in Quantitative Trading

• (Deep) Learning to Trade: An Experimental Analysis of AI Trading and Market Outcomes

If you’re a researcher working in this space, I’d love to hear from you. Reach out: [email protected]

With the news cycle quiet this final week of August, I wanted to resurface some of the academic work I’ve highlighted over this year:

AI Asset Pricing Models

The tech behind ChatGPT can uncover hidden drivers of stock returns, according to new academic research.

Traditionally, researchers proposed theories, wrote rules, and deployed code. With large-scale AI, the model can sift through mountains of data, learn context, and generate predictions.

This approach challenges factor models (like Fama-French). Instead of deciding up front what matters, AI scours the entire cross-section for what drives returns. That’s the core idea behind the new paper, “Artificial Intelligence Asset Pricing Models.”

I had a fun chat with one of the authors, Semyon Malamud, who broke it down well:

“𝘐𝘯 𝘵𝘩𝘦 𝘰𝘭𝘥 𝘥𝘢𝘺𝘴, 𝘱𝘦𝘰𝘱𝘭𝘦 𝘸𝘰𝘶𝘭𝘥 𝘫𝘶𝘴𝘵 𝘤𝘰𝘶𝘯𝘵 𝘸𝘰𝘳𝘥𝘴... 𝘣𝘶𝘵 𝘵𝘩𝘦𝘯 𝘱𝘦𝘰𝘱𝘭𝘦 𝘳𝘦𝘢𝘭𝘪𝘻𝘦𝘥 𝘺𝘰𝘶 𝘤𝘢𝘯'𝘵 𝘵𝘳𝘦𝘢𝘵 𝘸𝘰𝘳𝘥𝘴 𝘰𝘶𝘵 𝘰𝘧 𝘤𝘰𝘯𝘵𝘦𝘹𝘵. 𝘠𝘰𝘶 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘶𝘯𝘥𝘦𝘳𝘴𝘵𝘢𝘯𝘥 𝘵𝘩𝘦 𝘤𝘰𝘯𝘵𝘦𝘹𝘵𝘶𝘢𝘭 𝘦𝘯𝘷𝘪𝘳𝘰𝘯𝘮𝘦𝘯𝘵 𝘰𝘧 𝘦𝘷𝘦𝘳𝘺 𝘨𝘪𝘷𝘦𝘯 𝘸𝘰𝘳𝘥. 𝘞𝘦 𝘦𝘹𝘵𝘦𝘯𝘥𝘦𝘥 𝘵𝘩𝘪𝘴 𝘪𝘥𝘦𝘢 𝘵𝘰 𝘧𝘪𝘯𝘢𝘯𝘤𝘪𝘢𝘭 𝘮𝘢𝘳𝘬𝘦𝘵𝘴 - 𝘺𝘰𝘶 𝘤𝘢𝘯'𝘵 𝘶𝘯𝘥𝘦𝘳𝘴𝘵𝘢𝘯𝘥 𝘛𝘦𝘴𝘭𝘢 𝘰𝘳 𝘎𝘰𝘰𝘨𝘭𝘦 𝘰𝘳 𝘔𝘪𝘤𝘳𝘰𝘴𝘰𝘧𝘵 𝘢𝘭𝘰𝘯𝘦. 𝘠𝘰𝘶 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘵𝘳𝘦𝘢𝘵 𝘪𝘵 𝘪𝘯 𝘤𝘰𝘯𝘵𝘦𝘹𝘵, 𝘥𝘦𝘵𝘦𝘳𝘮𝘪𝘯𝘦𝘥 𝘣𝘺 𝘰𝘵𝘩𝘦𝘳 𝘴𝘵𝘰𝘤𝘬𝘴." Read more

Humans Create Bubbles. AI Doesn’t: Study

Large Language Models make more rational pricing decisions than humans, according to new academic research.

In the experiment, both human and AI participants traded a financial asset over 30 rounds in a controlled market environment. All traders knew the asset’s fixed value was 14 units, yet humans still inflated prices to 2-3 times its worth before a crash.

Most AI models stayed disciplined, trading near the real value—even when researchers encouraged bubble-like behavior.

"Humans almost always generate a bubble using this paradigm and LLMs almost never," said Thomas Henning, one of the paper's coauthors. Read more 

AI That Reads and Listens to Earnings Calls Delivers Better Stock Predictions: Study

AI that analyzes both what executives say and how they say it during earnings calls makes better stock predictions than text-only analysis, according to researchers from Stevens Institute of Technology.

The study introduces the "ECC Analyzer," a framework that leverages large language models (LLMs) to dissect earnings conference calls (ECCs) by integrating audio recordings, transcripts, and targeted financial queries. The system reduced prediction errors by 27.7% compared to current methods, particularly improving short-term volatility forecasts (3- and 7-day windows).

"Different types of data reveal different insights," Yupeng Cao, a Stevens Institute doctoral candidate and co-lead author, tells me. "Text gives us clear statements and numbers, while audio patterns might reveal additional context not captured in transcripts alone." Read more

ChatGPT Relies on Memory, not Math for Financial Predictions

New research shows that Large Language Models struggle with basic accounting and rely on memorization rather than true numerical reasoning, according to a study from University of Chicago’s Bradford (Lynch) Levy.

For example, if you ask an LLM to add any two numbers between 0 and 100, it’s generally correct. But if you ask it to add any two numbers between 0 and 10,000, accuracy plummets.

Levy created a novel test, described in his paper 𝘊𝘢𝘶𝘵𝘪𝘰𝘯 𝘈𝘩𝘦𝘢𝘥: 𝘕𝘶𝘮𝘦𝘳𝘪𝘤𝘢𝘭 𝘙𝘦𝘢𝘴𝘰𝘯𝘪𝘯𝘨 𝘢𝘯𝘥 𝘓𝘰𝘰𝘬-𝘢𝘩𝘦𝘢𝘥 𝘉𝘪𝘢𝘴 𝘪𝘯 𝘈𝘐 𝘔𝘰𝘥𝘦𝘭𝘴, by changing the least significant digit in a company's accounting results (for instance, changing $7.334 billion to $7.335 billion) to show that GPT-4's accuracy in predicting earnings changes dropped dramatically - from 60% to no better than random chance. This suggests the model isn't actually analyzing financial data but simply matching patterns it has memorized.

"If you look at what's happening under the hood, these models aren't thinking through financial statements - they're remembering patterns from their training data," Levy said. Read more

ICYMI

Using AI to Quantify Intangible Value

AI Stack with Sparkline Capital’s Kai Wu

Quantitative investors have historically relied on accounting data and price metrics.

Kai Wu thinks they're missing the soft factors that drive stock performance today.

As Sparkline Capital founder and CIO, he uses AI to analyze patents, corporate communications, and other unstructured data to identify what he calls "intangible value"—the intellectual property, brand strength, and human capital that he believes traditional financial statements understate.

He started his career at GMO working on Jeremy Grantham's $40 billion asset allocation team, helping manage a $2.5 billion global macro hedge fund. In 2014, he co-founded Kaleidoscope Capital, a quantitative hedge fund in Boston that grew to $350 million in assets before selling his stake in 2018.

He founded Sparkline Capital that same year, spending time exploring where the investment industry was headed and discovering large language models—well before they became mainstream. He launched his first ETF in 2021 and has since built a suite of active ETFs centered on his intangible value framework.

In our conversation, Wu explains how he applies AI on centuries of patent data and culture indicators, why he thinks the line between quantitative and fundamental investing is blurring, and why transfer learning made text-based factor investing viable. He also shares his view on what investors can learn from Renaissance Technologies’ use of unstructured data.

This interview has been edited for clarity and length. 

Tell me about Sparkline Capital

The main business at Sparkline is asset management through ETFs. We’re still trying to create alpha using quantitative techniques but in terms of structure we are trying to skate to where the puck is going. A lot of assets and investor interest are moving into ETFs, specifically active ETFs.

Historically, ETFs were synonymous with index funds. But due to a variety of changes, we’re now seeing more active strategies put into ETF wrappers. That provides efficiency, operational benefits, and tax advantages compared with traditional hedge funds. There’s a lot of interest in that category.

I launched my first fund four years ago, a second one about a year ago, and now I’m building out a suite of products centered on the concept of intangible value. I believe that if value investing, in the Ben Graham and Warren Buffett sense, is going to thrive in the digital economy, then we need to adapt the definition of intrinsic value to include intangible assets.

The techniques we use—LLMs and unstructured data—are what make this possible. If you just look at accounting data, you’re missing out on the most valuable information on intangible assets. There’s simply not enough information. Why wouldn’t you also look at the 80-plus percent of data that’s unstructured? And why wouldn’t you use the latest tools to analyze it?

Nobody I know is really trying to solve this problem.

Read the full interview here

WHAT ELSE I’M READING

• AI Makes It Harder for Entry-Level Coders to Find Jobs, Study Says (Bloomberg)

• AI Data Center Boom Fuels Economic Growth (NYT)

• These AI-Skilled 20-Somethings Are Making Hundreds of Thousands a Year (WSJ)

• Anthropic Says Criminals Used its Claude AI Technology (Bloomberg)

• A high-frequency trading firm is offering its interns $425k to come back as OpenAI looms (eFinancialCareers)

• AI giants race to scoop up elusive real-world data (Restofworld)

CALENDAR

Upcoming AI + Finance Conferences

• AI in Financial Services (Arena) – Sept 9–10, 2025 • London

  Focused on AI strategy, implementation, and ROI in finance.

• Cornell Financial Engineering Manhattan 2025 Future of Finance & AI Conference – Sept 19, 2025 • New York (I’ll be attending)

  A one-day forum on AI, quantitative finance, and hedge-fund strategies, attracting leading quants and industry practitioners.

• Bloomberg-Columbia ML in Finance Conf – Sept 25, 2025 • New York

  Academic–industry event hosted by Columbia University and Bloomberg, focused on ML applications in finance including asset pricing, market forecasting, and LLM risk.

• GAIIM Conference 2025 – Sept 30, 2025 • New York

  Forum on practical applications of AI in investing, featuring tools for research, valuation, and portfolio workflows.

• AIFin Workshop at ECAI 2025 – October 26, 2025 • Bologna, Italy

  One-day academic workshop on AI/ML in finance, covering trading, risk, fraud, NLP, and regulation.

• AI in Finance 2025 – October 27–30, 2025 • Montréal

  Academic event covering ML in empirical asset pricing and risk.

• ACM ICAIF 2025 – November 15–18, 2025 • Singapore

  Top-tier academic/industry conference on AI in finance and trading.

• AI for Finance – November 24–26, 2025 • Paris

  Artefact’s AI for Finance summit, focused on generative AI, future of finance, digital sovereignty, and regulation 

• NeurIPS Workshop: Generative AI in Finance – Dec. 6/7 • San Diego One-day academic workshop at NeurIPS focused on generative AI applications in finance, organized by ML researchers.