• 

  今日概述

• 

  持有和交易

• 

  研究和观察列表

• 

  数据和维护

Inspiration

As an individual investor and a software tester, I realized that my biggest problem wasn't a lack of market data—it was the lack of a disciplined decision-making process.

Most investment tools focus on providing more information or predicting the next winning stock. I wanted to build something different: a personal investment operating system that helps investors make consistent, explainable decisions based on trusted data and clear rules.

What it does

Money is a local AI-assisted investment decision system that combines portfolio management, research watchlists, risk boundaries, decision rules, and daily investment workflows into a single dashboard.

Rather than generating automatic trading signals, it helps investors understand what deserves attention, what is currently blocked, and why a decision is made.

How I built it

Money is built with Python, JavaScript, HTML, and CSS, running on a lightweight local Python server.

The project was developed in multiple phases. The latest phase focused on redesigning the decision architecture by separating eligibility, ranking, recommendation, and presentation responsibilities. The current implementation completed integration with 55 passing regression tests.

Throughout development, I used OpenAI Codex as an engineering partner for implementation, refactoring, testing, and documentation, while GPT-5.6 helped review architecture, refine product direction, and improve the overall user experience.

Challenges

The biggest challenge was keeping the system consistent as it grew.

Different modules should never produce conflicting investment decisions. I redesigned the architecture so that decision logic exists in one place, while the dashboard and reports simply consume those results. This improved maintainability without changing the existing business logic.

What I learned

This project taught me that building an investment assistant is less about predicting the market and more about designing trustworthy decision systems.

I also learned how effective Codex can be when used as an engineering collaborator instead of a simple code generator, especially for large-scale refactoring and architecture evolution.

Built With

• python?javascript?html?css?openai?codex?gpt-5.6?git?github