Implied CEO Ying Hua joined Ethan Kho on Odds on Open to discuss the intersection of fundamental and quantitative investing, the role AI plays in a portfolio manager's process, and the evolving landscape of multi-manager hedge funds. Ying's contrarian view is that more AI in the market creates new sources of alpha, increasing the demand for exceptional talent. When every fund runs the same automations, creative human judgment becomes the differentiator.
"AI by definition, next-token prediction, is pattern matching. So it's just another wave of quant, but this time less on number data and more on word data."
Like alternative data before it, AI will move from a source of alpha to table stakes as adoption grows. The edge will then come from how each fund puts it to work. Doing that well requires understanding the limits of LLMs. A few that came up in the conversation:
- LLMs are solely pattern matching; they don't understand a company the same way a human can
- Training data matters. Models trained on public context tend to regurgitate what they "know" rather than applying it to the current market regime
- Without sufficient domain knowledge, AI cannot produce grounded predictions. Constructing objective domain knowledge is harder than it sounds
You can also watch the episode on YouTube.