Insights
Ideas for AI-native trading workflows
Long-form pieces on process, positioning, and using models without outsourcing judgment.
Editorial note
- Drafts and outlines can be accelerated with LLMs; claims and compliance-sensitive language stay human-reviewed.
- Articles are markdown-backed—easy to extend as your publishing workflow matures.
We ran 40+ battle-tested strategies through autoresearch
Classic setups are hypotheses. A Karpathy-style experiment loop on strategy code and benchmarks surfaces second-order improvements—filters, exits, and risk—that the textbook version rarely explores.
Why LLM-native trading tools will beat static dashboards
Markets move faster than menu-driven tools. The winning stack is shifting from fixed dashboards toward systems that can reason over context, explain tradeoffs, and adapt the workflow in real time.
The new edge is not faster data. It is a tighter AI research loop.
Most traders do not lose because they lack charts. They lose because their research loop is slow, fragmented, and emotionally inconsistent. AI changes the loop more than the signal.
How to use AI for trading without being reckless
The right use of AI in trading is not blind automation. It is controlled acceleration: let AI generate hypotheses, score setups, and explain scenarios while you keep the risk controls hard-coded.