mangoview

# llms.txt — Mangoview

## Project Overview

Mangoview is an investment decision-support tool that answers three questions for any decision: 时机对不对 (Is the timing right? / 天时) · 公司行不行 (Is the company solid? / 地利) · 自己有没有犯傻 (Are you being stupid? / 人和). Built on a 3-layer 6-step analysis engine + signal system + multi-source data collectors + FastAPI + Next.js 14.

Mission: Turn subjective investment reasoning into a structured, auditable pipeline.

## Core Pages

- [README](https://github.com/lj22503/mangoview) — Full documentation (Chinese)
- [README_EN](https://github.com/lj22503/mangoview/blob/main/README_EN.md) — English documentation
- [DESIGN.md](https://github.com/lj22503/mangoview/blob/main/DESIGN.md) — System design
- [SPEC.md](https://github.com/lj22503/mangoview/blob/main/docs/SPEC.md) — Product spec
- [PROGRESS.md](https://github.com/lj22503/mangoview/blob/main/PROGRESS.md) — Project progress

## What it is

**One-liner**: A structured investment decision engine that answers three questions — timing, company quality, and self-bias check.

**Core value**:
- Structure: every analysis goes through 3 layers (天时/地利/人和) × 6 steps
- Multi-source: AKShare / 东方财富 / 天天基金 / 新浪 / tushare
- Bias-aware: dedicated 人和 (self-bias) layer
- Auditable: every decision logged with reasoning chain

**Target users**: Individual investors who want disciplined, structured decision-making.

## 3-Layer × 6-Step Analysis Engine

### Layer 1: 天时 (Timing)
- Macro cycle analysis (Merrill clock / Kondratieff)
- Industry rotation signals
- Market sentiment indicators

### Layer 2: 地利 (Company)
- Fundamentals: revenue / profit / cash flow
- Competitive position
- Valuation (DCF / relative)

### Layer 3: 人和 (Self-bias)
- Loss aversion detection
- Herding / FOMO identification
- Overconfidence check
- Decision history review

## 6-Step Pipeline

1. **Collect** data from multi-source providers
2. **Filter** signal noise
3. **Judge** direction (long / short / wait)
4. **Analyze** company fundamentals
5. **Self-check** bias
6. **Decide** with audit trail

## Tech Stack

| Layer | Tech |
|---|---|
| Backend | FastAPI + Python 3.11+ |
| Frontend | Next.js 14 + Tailwind + Recharts/ECharts |
| Data | AKShare / 东方财富 / 天天基金 / 新浪 / tushare |
| Knowledge | investment-framework-skill (submodule) |
| Test | 86+ unit tests in `backend/core/engine/tests/` |

## Installation

```bash
# Backend
cd backend
pip install -r requirements.txt
cp .env.example .env  # fill in API keys

# Frontend
cd frontend
npm install
npm run dev
```

## Search Keywords

investment decision, 3-layer analysis, 天时地利人和, bias detection, fund analysis, A股分析, market data, AKShare, tushare, signal system, decision audit, FastAPI investment tool, Next.js finance app, quantitative analysis, RAG, MCP.

## Contact

GitHub Issues: https://github.com/lj22503/mangoview/issues