AI Trading Agent Analysis

A sophisticated multi-agent trading system that analyzes stocks using specialized AI agents for technical analysis, fundamentals, sentiment, macro context, and risk management to generate comprehensive trading decisions.

System Overview

This skill helps you analyze stocks using a multi-agent architecture orchestrated by LangGraph. The system employs multiple specialized agents that work together to provide comprehensive stock analysis and trading recommendations.

Prerequisites

Python 3.11+

Poetry (Python dependency management)

Node.js and npm (for frontend interface)

Gemini API key (primary LLM)

Optional: Algogene API credentials for enhanced data

Setup Instructions

1. **Initialize Python Environment**

```bash

poetry install

poetry shell

```

2. **Configure Environment Variables**

Create a `.env` file with:

```

GEMINI_API_KEY=your_gemini_api_key

# Optional OpenAI-compatible API configuration

# Optional Algogene API configuration

```

3. **Setup Frontend (Optional)**

```bash

cd frontend/

npm install

```

Analysis Workflow

The system follows a multi-phase workflow:

Phase 1: Data Collection

**Market Data Agent**: Fetches real-time market data via akshare API

Phase 2: Parallel Analysis (4 Agents)

**Technical Analyst**: Analyzes technical indicators and chart patterns

**Fundamentals Agent**: Evaluates financial statements and metrics

**Sentiment Agent**: Analyzes news sentiment

**Macro News Agent**: Assesses market-wide news impact

Phase 3: Research Synthesis

**Bull Researcher**: Develops bullish investment thesis

**Bear Researcher**: Develops bearish counterargument

Phase 4: Debate & Risk Assessment

**Debate Room**: Synthesizes bull/bear perspectives

**Risk Management**: Assesses risk and position sizing

**Macro Analyst**: Provides stock-specific macro context

Phase 5: Final Decision

**Portfolio Manager**: Makes final trading decision based on all inputs

Usage Instructions

Command Line Analysis (Primary Mode)

**Basic stock analysis:**

```bash

poetry run python src/main.py --ticker STOCK_CODE

```

**Analysis with detailed reasoning:**

```bash

poetry run python src/main.py --ticker STOCK_CODE --show-reasoning

```

**Backtesting:**

```bash

poetry run python src/backtester.py --ticker STOCK_CODE --start-date YYYY-MM-DD --end-date YYYY-MM-DD

```

API Service Mode

**Start backend server:**

```bash

poetry run python run_with_backend.py

```

**Start frontend (in separate terminal):**

```bash

cd frontend/

npm start

```

The backend provides REST API endpoints for:

Agent execution

Analysis results

Workflow monitoring

Run status tracking

Log retrieval

Code Structure

**Agent Implementation**: `src/agents/`

`state.py` - Shared state management and workflow

Individual agent modules for each specialist

`super_node.py` - Enhanced agent with ReAct capabilities

**Tools & Utilities**: `src/tools/`

`algogene_client.py` - Data source integration

`news_crawler.py` - News fetching

`code_interpreter.py` - Code execution

**Backend API**: `backend/`

FastAPI application with CORS

Routers for agents, analysis, workflow, logs, runs

**Frontend**: `frontend/`

React TypeScript with Tailwind CSS

Components: InputForm, AnalysisResult

Development Commands

**Code formatting:**

```bash

poetry run black src/ backend/

poetry run isort src/ backend/

```

**Linting:**

```bash

poetry run flake8 src/ backend/

```

**Testing:**

```bash

poetry run pytest

```

**Frontend testing:**

```bash

cd frontend/

npm test

```

**Production build:**

```bash

cd frontend/

npm run build

```

Key Features

**Multi-agent orchestration** using LangGraph state management

**Real-time market data** from akshare and Sina Finance APIs

**Comprehensive logging** of all LLM interactions and agent executions

**Dual operation modes**: CLI for quick analysis, API for integration

**Workflow tracking** with unique run IDs

**Backtesting capabilities** (currently under testing)

**Risk management** and position sizing recommendations

**Bull/bear debate synthesis** for balanced perspectives

Monitoring & Logs

Agent execution logs: `logs/` directory (organized by agent type)

LLM interaction tracking: `src/utils/llm_interaction_logger.py`

Real-time workflow status via API endpoints

Structured terminal output with summary reports

Important Notes

The system is designed for Chinese stock markets (uses akshare)

Both CLI and API modes can run simultaneously

Agent reasoning can be displayed with `--show-reasoning` flag

Backtesting functionality is available but marked as under testing

All workflow runs are tracked with unique identifiers for debugging

Example Workflow

1. Set environment variables in `.env`

2. Activate Poetry environment: `poetry shell`

3. Run analysis: `poetry run python src/main.py --ticker 600519 --show-reasoning`

4. Review agent reasoning and final trading recommendation

5. Check logs in `logs/` directory for detailed execution traces

For web interface:

1. Start backend: `poetry run python run_with_backend.py`

2. Start frontend: `cd frontend/ && npm start`

3. Navigate to http://localhost:3000

4. Enter stock ticker and view real-time analysis results

AI Trading Agent Analysis

AI Trading Agent Analysis

System Overview

Prerequisites

Setup Instructions

Analysis Workflow

Phase 1: Data Collection

Phase 2: Parallel Analysis (4 Agents)

Phase 3: Research Synthesis

Phase 4: Debate & Risk Assessment

Phase 5: Final Decision

Usage Instructions

Command Line Analysis (Primary Mode)

API Service Mode

Code Structure

Development Commands

Key Features

Monitoring & Logs

Important Notes

Example Workflow

Reviews (0)