YC's new startup claims AI quant made 108% in two months, outperforming top 10% of Jane Street traders

According to Data Beating monitoring, YC S26 incubator Prodigy Research has officially launched, claiming to train the world's most powerful quantitative financial base model. CEO Michael Wang declared that their in-house AI has outperformed the top 10% of traders at Jane Street, and their internal financial assessment has even surpassed Claude Fable and GPT-5.6 Sol.

The live trading performance boasts even more impressive results: a Delta-neutral strategy with a 108% return over two months, without a single losing week, while the major stock indices barely moved during the same period.

YC is also endorsing them. Partner Brad Flora stated that they have been live trading with Agent throughout this YC batch and have been "making more and more money." However, only self-reported results are currently public: Prodigy has not disclosed their capital, leverage, strategy capacity, complete trading records, or undergone third-party auditing. They also have not explained how they precisely determined the comparison with the "top 10% of Jane Street."

The comments section has raised a direct question: if they can double their returns in two months without any losing weeks, why do they still need to participate in YC's funding?

The two founders are brothers. Michael Wang previously worked as a trader at Jane Street before joining Google DeepMind; Yuhua Wang has AI experience from Apple.

In fact, they have been working on AI trading for a while. Their previous project, Prodigy AI, focused on autonomous trading agents for prediction markets like Polymarket and Kalshi, claiming superior performance in-house over Claude Opus 4.7 and GPT-5.5. Now, they have expanded their scope to the entire field of quantitative finance.