At 007 Venture Partners, we are drawn to founders who build with conviction, clarity and ambition. They aren't just building products. They are taking on the hardest problems in the biggest arenas in the world.
Few arenas are bigger than global markets. Trading firms generated hundreds of billions of dollars in profit last year, and Jane Street alone booked a record $39.6B in trading revenue in 2025. Yet the whole quant industry runs on a scarce input. There are only around 30,000 quant researchers in the world, and building each new strategy takes a firm three to six months.
The problem is not a shortage of capital. Allocators want to put an estimated $2T into quant strategies. The problem is that there aren't enough great researchers to deploy it, and hedge funds are turning money away.
Where It Started
Prodigy Research began with two brothers who had seen both sides of the problem up close. Michael had worked on a Jane Street trading desk and then trained frontier models at Google DeepMind. Yuhua had built large-scale AI infrastructure at Apple and shipped Salesforce's flagship AI platform.
From where they stood, the bottleneck was obvious. The best trading firms in the world were limited by how many researchers they could hire, while AI was quickly learning to do the kind of research those people did. So they walked away from Jane Street, DeepMind and Apple to build something new together.
That is what Prodigy Research is here to do. We are proud to share that 007 Venture Partners has invested in Prodigy Research (YC S26), a San Francisco-based frontier AI trading research lab training what it believes is the world's best foundation model for quantitative finance.
A New Kind of Quant Firm
At Prodigy, AI does the research itself, rather than assisting the people who do it.
Every strategy a quant fund runs starts as an idea that a human researcher has to come up with, test and prove. Prodigy hands that whole process to swarms of AI agents, with no human in the loop:
- Ideate: agents generate and screen new strategy ideas continuously.
- Backtest: each candidate is tested against historical market data.
- Validate: strategies are stress-tested for robustness before any capital is committed, guarding against patterns that only look good on paper.
- Deploy: the strategies that survive go into live trading with real money.
Because no step needs a person, discovery scales with compute rather than headcount. Prodigy can explore more than 100 times as many ideas as a team of human quants, and do months of incumbent research in days.
Underneath the agents is Prodigy's own foundation model, trained specifically for quantitative finance rather than built on top of a general-purpose chatbot. The company reports that it outperforms leading frontier models, including Claude Fable and GPT-5.6 Sol, on financial and quant research tasks.
Why Live Trading Is the Whole Game
In quant finance, a model is only as good as its results in real markets. Live P&L is the number Prodigy is built around.
Over the course of its YC batch, Prodigy's model generated $700K in profit from live trading, a +140% return with a 3.9 Sharpe ratio. Over the same period, the S&P 500 returned +1.01% and the Nasdaq fell 3.67%. The company reports it never had a down week.
Those returns came from delta-neutral strategies, which are designed to make money regardless of which way the market moves. In other words, this was not a bet on a rising market. It was the model finding edge in a flat-to-down one.
The results are early, company-reported and drawn from a short period on a modest capital base. But it is rare to see a seed-stage AI company whose product is measured in real money rather than pilots and demos.
From Strategies to a Foundation Model for Markets
What excites us most is the loop this creates.
Every strategy Prodigy deploys produces something general-purpose AI labs don't have: ground-truth feedback from real markets. An idea goes in, a trade is made, and the market delivers a verdict. Repeated across thousands of strategies, that becomes a proprietary training signal that sharpens the model with every cycle.
Over time, the same model can extend across asset classes, frequencies and geographies, adding capacity without adding headcount. That gives Prodigy several paths to build value: its own trading book, managing capital for institutions that can't find room in today's funds, and potentially licensing its research agents to other firms.
Read in order, this is a story about scale rather than any single strategy. The quant firms that defined past generations were limited by how many great researchers they could hire. Prodigy is built to be limited only by compute.
Why Now
Every generation of trading firm has hit the same wall. Renaissance (founded 1978), Jane Street (1999) and XTX Markets (2015) each built a new kind of firm, and each was eventually limited by how many great researchers it could find.
What has changed is that AI agents can now reason, write code and run experiments well enough to do quant research themselves. Incumbents such as Man Group and Bridgewater are starting to bring AI into their research, but they are fitting it into organisations built around human researchers.
Meanwhile, the demand is clear. Allocators want to put an estimated $2T into quant strategies that existing funds can't absorb, and the company estimates more than $200B a year in trading profit is left uncaptured. What has been missing is a firm built AI-native from day one. Prodigy sits in that gap.
The Founders
Michael Wang and Yuhua Wang are brothers.
Michael, Co-founder and CEO, joined Jane Street full-time at age 20 and worked on a trading desk generating billions in annual P&L. He then became a Senior Researcher at Google DeepMind, training frontier foundation models.
Yuhua, Co-founder and CTO, built large-scale AI infrastructure, models and agents at Apple, shipped to billions of users. Before that, he shipped Salesforce's flagship AI platform.
Between them, they cover what this company needs: how elite trading desks work from the inside, how frontier models are trained, and how to run AI systems reliably at scale. As brothers, they also bring a level of trust and alignment that most founding teams take years to build. That combination, and the conviction behind it, is rare.
Why We Invested
What stood out to us from day one was Michael and Yuhua, and the scale of their ambition. Two brothers who walked away from roles at Jane Street, Google DeepMind and Apple to build a frontier AI trading research lab together is a rare story, and the conviction behind it is even rarer. Global markets are one of the biggest arenas in the world, and they have set out to use AI to solve them.
What moved us from interest to conviction was velocity. Within a single YC batch:
- The model went from research to live, profitable trading with real money
- Its delta-neutral strategies returned +140% with a 3.9 Sharpe ratio while major indices were flat or down
- Its launch reached around 1.9 million views on X and sparked a wide industry debate about the future of AI in trading
It is rare to see a two-person team get this far this fast. It is rarer still when the product's results are measured directly in the market.
We are proud to invest alongside Y Combinator, Paul Graham, Andy Fang and Stanley Tang (co-founders of DoorDash), Rick Rubin, Walter Kortschak, Coinbase Ventures, Multimodal Ventures, Amino Capital, Transpose Platform and K5 Global.
The Road Ahead
The next stage is about turning a strong start into an institutional track record. Over the coming year, the team plans to extend its live results across more market conditions, scale the capital behind its strategies, and broaden the model across new asset classes and markets.
The company's mission also aligns with UN Sustainable Development Goals 8 and 9. Liquid, efficiently priced markets lower the cost of capital for companies and improve outcomes for the pension funds, endowments and long-term savers who invest through them. And Prodigy's autonomous research loop of hypothesis, test, validate and deploy is a template for AI-accelerated discovery well beyond finance.
We are proud to support Michael, Yuhua and the Prodigy Research team on this journey.
The founders' ambition is to build the world's first trillion-dollar trading firm. When research is no longer limited by how many people a firm can hire, you don't just build a better hedge fund. You change what a trading firm can be.
And that is what Prodigy Research is building.
