elian-dan/ai-portfolio-compass-public — explained in plain English
Analysis updated 2026-05-18
Combine holdings from multiple brokerage accounts and markets into one unified portfolio view.
Get an AI generated diagnosis of portfolio concentration risk and position health.
Review past trades against what happened afterward to learn from prior decisions.
Try the tool safely with fabricated demo data before connecting a real brokerage account.
| elian-dan/ai-portfolio-compass-public | avbiswas/sam2-mlx | chanthruu/wind_energy_monitor | |
|---|---|---|---|
| Stars | 27 | 27 | 27 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | moderate | moderate |
| Complexity | 3/5 | 4/5 | 3/5 |
| Audience | general | researcher | data |
Figures from each repo's GitHub metadata at analysis time.
Real account use requires installing and logging into the separate Futu or moomoo OpenD desktop application.
AI Portfolio Compass is a self hosted workbench for people who keep their investments spread across several brokerage accounts and different markets, such as mainland Chinese funds and stocks, US stocks, and Hong Kong stocks. The core problem it addresses is that many investors can see what is inside each individual account but cannot easily see a unified picture of everything they own, where their risk is concentrated, or which positions actually need attention. Rather than acting as a simple chat assistant you ask for stock tips, the project treats AI as one capability layered on top of a personal system for tracking real holdings, trade history, the reasons behind each position, market news, and past decisions. It groups holdings into categories such as core long term positions, medium term allocations, short term trades, options, and legacy positions a user no longer has a clear reason to hold, so the advice given about each one differs based on its role rather than treating every position the same way. Instead of a constant stream of alerts, the home screen distills things down into a short list of what actually needs attention today, downgrading anything based on stale or insufficient data rather than issuing forceful buy or sell language. The system also supports reviewing past trades against what actually happened afterward, so a user can see whether a past decision followed the original plan or was made emotionally, and whether prior AI suggestions were actually useful. Technically it runs locally as a React and Vite frontend paired with a FastAPI Python backend, reading real account data through the Futu or moomoo trading platform's local OpenD service in read only mode, meaning it never places, cancels, or modifies trades and never asks for a brokerage password. Market data and news can be pulled from configurable providers including Alpaca, Polygon, Alpha Vantage, and others, and AI analysis can use DeepSeek, falling back to local rule based analysis if no AI key is configured. A separate demo mode with fabricated sample data is included for trying the tool without connecting a real account. Local data, API keys, and logs are excluded from version control by default. The project is released under the MIT license.
A self hosted, local first workbench that unifies holdings across multiple brokerage accounts and markets, then uses AI to diagnose portfolio risk and review past trades.
Mainly Python. The stack also includes Python, FastAPI, React.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.