Hello, I’m Nathan

About Me

I am currently a high school junior from Oklahoma and have been learning Python for a little less than a year. Outside of programming, I enjoy running, playing basketball, and reading. I spend most of my time researching AI and its potential applications in finance. All of my major projects are open-source to encourage transparency and external contributions.

About My Work

Most of my work lives at the intersection of:

• Empirical evaluation,
• Flexibility > abstraction,
• Explicit failure > silent corruption,
• Transparency above all else.

I care about honest methodology, conservative claims, and building tools that make exploring easier for everyone.

---

Major Projects

LIBB — LLM Investor Behavior Benchmark

A lightweight internal framework for logging and analyzing LLM trading behavior:

• trade-level records
• portfolio state over time
• archived model decision text
• reproducible metric computation

Built to support evaluation and analysis, not optimization or monetization.

---

Evaluating ChatGPT as a Portfolio Decision-Maker in Micro-Cap Equities

An exploratory evaluation of ChatGPT trading micro-cap stocks over a six month timeframe:

• Detailed methodology and failure modes
• Prompting consisted of daily updates and weekly deep research
• Fully documented scripts, graphing, artifacts, etc.
• Provided blog updates every week on performance

---

LLM Trading Lab

A research-oriented workspace for running and evaluating AI-driven trading experiments.
This repository serves as a baseline platform for future studies, with a focus on behavioral analysis, reproducibility, and transparent documentation rather than performance optimization.

A platform for evaluating LLM trading experiments:

• forward-only experimental design
• transparent logging and artifacts
• research analysis over optimization
• powered by LIBB for all future experiments
• monthly blog updates will document the progress of experiments

---

Notes

• Most projects start exploratory and evolve iteratively.
• I document constraints and limitations aggressively.
• I’m cautious about generalization and prefer scoped conclusions.
• I’m always open to criticism and advice.

---

Contact

If you’re interested in:

• evaluation methodology,
• LLM behavior analysis,
• applied research,
• asking questions,
• simply connecting,

feel free to reach out.

• Email: [email protected]

• Blog: https://nathanbsmith729.substack.com/

• X: https://x.com/NathanS729

---

Thank you for reading!

Collapse