PatrickSUDO

PatrickSUDO / fadacai-portfolio

Public

Claude Code 投資研究與組合管理框架：skills + MCP + 第一性原理紀律 + thesis ledger

17
1
85% credibility
Found May 31, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This is a personal investment research and portfolio management framework for US stocks and options. It uses AI (Claude Code) to generate daily market briefings, analyze individual stocks, track investment thesis over time, and automate report delivery. The system emphasizes disciplined investing by requiring users to state their investment thesis clearly, define falsification conditions, and calculate expected value before making decisions. It integrates with brokerage accounts and various market data sources to provide real-time portfolio views and market sentiment analysis. Daily briefings can be automatically sent to Telegram and email each trading day at 1 PM ET. The framework includes a thesis ledger to track investment ideas from creation through verification, helping users build a track record of prediction accuracy over time.

How It Works

1
💡 You discover a smarter way to research stocks

You've been trading US stocks and options on your own, but you want a disciplined system that helps you think through investments clearly without getting caught up in hype.

2
🛠️ You set up your research workspace

You install Claude Code on your computer, which becomes the brain of your research assistant. The framework organizes everything into easy-to-use commands.

3
📊 You connect your broker and data sources

You link your brokerage account so the system can see your actual holdings, and connect various market data services so it has up-to-date information about stocks, news, and market sentiment.

4
📰 Every morning, your personalized market briefing arrives

At 1 PM on every trading day, your phone buzzes with a complete market summary: today's news, earnings coming up, how the market is feeling, sector trends, and your own portfolio status—all written in clear Chinese.

5
🔬 You research a stock with first-principles thinking

When you're interested in a company, you ask for a deep analysis. The system forces you to state your investment thesis clearly, define what would prove you wrong, and calculate expected value—keeping you honest instead of just chasing stories.

6
You track your investment ideas over time
✅
Your thesis proved correct

You log the results, build your hit rate score, and consider adding to the position

❌
Your thesis was wrong

You record what happened, learn from it, and avoid repeating the same mistakes

7
🎯 Everything runs on autopilot

Your daily briefings are automatic. The system checks if the market is open, fetches fresh data, generates your report, and sends it to your Telegram and email—no manual work needed.

🏆 You have a disciplined, AI-assisted investment research system

You receive professional-grade market analysis daily, make investment decisions based on verifiable thesis rather than narratives, and build a track record of your prediction accuracy over time.

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is fadacai-portfolio?

A Python-based investment research framework that runs on Claude Code, letting you generate daily briefings, analyze stocks, review portfolio positions, and plan options strategies through slash commands. It pulls real-time data from seven external services including brokerages, SEC filings, technical indicators, and sentiment feeds, then enforces a first-principles discipline system where every investment conclusion must include a falsifiable thesis, probability distribution, and expected value calculation. The framework outputs everything in Traditional Chinese and can automatically push daily summaries to Telegram and email via a macOS launchd scheduler.

Why is it gaining traction?

The thesis ledger system is the differentiator—it tracks every investment hypothesis you make, automatically surfaces them when earnings dates arrive, and lets you record whether the thesis passed or failed, building a hit rate over time. Most AI investing tools let you generate narratives; this one forces you to be accountable to your predictions. The model tiering (Haiku for data collection, Sonnet for synthesis, Opus for deep analysis) keeps costs reasonable while the retry-and-fallback architecture prevents single-point failures from breaking your workflow.

Who should use this?

Individual US stock and options traders who want AI-assisted research but dislike the "hallucination factory" problem in general-purpose LLMs. Quant researchers building a personal systematic approach will benefit most from the thesis ledger and EV calculation requirements. Developers comfortable with Claude Code CLI and comfortable wiring up their own API keys and brokerage credentials will get the most value.

Verdict

The framework is thoughtfully designed and the methodology documentation is solid, but at 17 stars this is early-stage software from a single author—the credibility score of 0.8500000238418579% reflects that novelty. Test coverage exists for the thesis ledger but the broader codebase is untested, and the macOS-only automation will frustrate Linux or Windows users. Worth exploring if you want disciplined AI-assisted investing, but expect to read the code and contribute your own integrations.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.