A web-based interface for manually marking profitable buy/sell signals on historical market data to generate training test cases for machine learning models.

Overview

This tool allows traders to:

View multi-timeframe candlestick charts
Navigate through historical data
Mark entry and exit points for trades
Generate test cases in realtime format
Train models with annotated data
Simulate inference to measure model performance

Project Structure

ANNOTATE/
├── web/                      # Web application
│   ├── app.py               # Main Flask/Dash application
│   ├── templates/           # Jinja2 HTML templates
│   │   ├── base_layout.html
│   │   ├── annotation_dashboard.html
│   │   └── components/
│   └── static/              # Static assets
│       ├── css/
│       ├── js/
│       └── images/
├── core/                    # Core business logic
│   ├── annotation_manager.py
│   ├── training_simulator.py
│   └── data_loader.py
├── data/                    # Data storage
│   ├── annotations/
│   ├── test_cases/
│   └── training_results/
└── tests/                   # Test files

Installation

# Install dependencies (if not already installed)
pip install dash plotly pandas numpy

# Run the application
python ANNOTATE/web/app.py

Usage

Start the application: Run python ANNOTATE/web/app.py
Open browser: Navigate to http://localhost:8051
Select symbol and timeframe: Choose trading pair and timeframes to display
Navigate to time period: Use date picker or scroll to find market conditions
Mark trades: Click on chart to mark entry point, click again for exit
Generate test cases: Click "Generate Test Case" to create training data
Train models: Select model and click "Train" to run training session
Simulate inference: Click "Simulate" to test model performance

Features

Multi-timeframe synchronized charts (1s, 1m, 1h, 1d)
Interactive trade marking with P&L calculation
Test case generation in realtime format
Model training integration
Inference simulation with performance metrics
Session persistence and auto-save
Dark theme UI

Integration with Main System

This sub-project is designed to be self-contained but can be integrated with the main trading system:

# Import annotation manager in main system
from ANNOTATE.core.annotation_manager import AnnotationManager

# Import training simulator
from ANNOTATE.core.training_simulator import TrainingSimulator

# Use generated test cases in training
test_cases = annotation_manager.get_test_cases()

Configuration

Configuration is loaded from the main config.yaml file. The application uses:

Data provider settings for historical data access
Model paths for training integration
Symbol and timeframe configurations

Development

To add new features:

Update requirements in .kiro/specs/manual-trade-annotation-ui/requirements.md
Update design in .kiro/specs/manual-trade-annotation-ui/design.md
Add tasks to .kiro/specs/manual-trade-annotation-ui/tasks.md
Implement changes following the task list

License

Part of the AI Trading System project.