Writingabout codeand everyday life.
Written slowly, and only when there's something worth saying.
Recent writing
- August 16, 202601
Cross-Sectional Stock Selection: Where Returns Come From and Why Models Fail
Cross-sectional models rank stocks against one another rather than forecast the market outright. This article explains where their edge can come from, how to evaluate it with rank IC, portfolio sorts, signal decay, and trading costs, and how regime shifts, research bias, portfolio constraints, and target drift can make that edge disappear.
- May 18, 202602
Why AI Agents Need Agent-Level Isolation
An agent profile can change how an agent talks, but it cannot change what the agent can actually reach: environment, tools, secrets, memory. What really needs isolating is the trust domain and the runtime domain, not the persona. This post is about why a single profile is not enough, why stuffing every skill into one general agent gets out of hand, when a separate Soul and memory are actually unnecessary, and how the isolation design in Hermes agent could become lighter.
- February 24, 202503
My vision of AIPC and AI Phone
I believe the future of AIPC and AI Phone won’t just depend on powerful chips. Instead, built-in AI models will be the heart of these devices. These models will act as the “brain,” working together with hardware and operating systems to make one complete system. Competition will shift from hardware specs to how well the AI models perform and improve the user experience.
- July 13, 201904
First Steps with F# - Fetching Real-Time Stock Quotes
If you're going to obsess over the market, you've got to keep an eye on it. But are you really going to pull out your phone or open a browser tab at work? A self-respecting retail investor uses the command line: low-key, and infinitely customizable. And since I make my living in the .NET ecosystem, my tool of choice is F#.
Projects
