https://github.com/maximefarre/quantitative-finance-dashboard
Streamlit dashboard for backtesting systematic trading strategies (MA Crossover, Regime Switching) on stocks, indices, forex and commodities — with automatic parameter optimization, trade analytics and daily reports.
https://github.com/maximefarre/quantitative-finance-dashboard
backtesting python quantitative-finance strategies streamlit trading
Last synced: about 2 months ago
JSON representation
Streamlit dashboard for backtesting systematic trading strategies (MA Crossover, Regime Switching) on stocks, indices, forex and commodities — with automatic parameter optimization, trade analytics and daily reports.
- Host: GitHub
- URL: https://github.com/maximefarre/quantitative-finance-dashboard
- Owner: MaximeFARRE
- License: mit
- Created: 2025-11-19T19:55:17.000Z (7 months ago)
- Default Branch: main
- Last Pushed: 2026-04-21T18:45:37.000Z (2 months ago)
- Last Synced: 2026-04-21T20:43:28.813Z (2 months ago)
- Topics: backtesting, python, quantitative-finance, strategies, streamlit, trading
- Language: Python
- Homepage:
- Size: 234 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Quantitative Finance Dashboard
A **Streamlit-based quantitative finance dashboard** for backtesting systematic trading strategies on individual assets and multi-asset portfolios, built as a 4th-year engineering project at ESILV.
---
## 1. Project Context
This project was developed as part of the **Python, Git, Linux for Finance** course.
It simulates the work of a **quantitative research team** in an asset management company, whose role is to support portfolio managers with **quantitative tools**.
The objective is to design and deploy a **professional interactive dashboard** able to:
- retrieve financial market data from a dynamic source (API),
- implement quantitative strategies and backtesting,
- simulate multi-asset portfolios,
- display results and key metrics in a clear and user-friendly way.
The application is built in **Python with Streamlit**, versioned using **Git/GitHub**, and **deployed on a Linux virtual machine** to ensure continuous availability.
---
## 2. Team & Division of Work
This project was completed by **two students**, with a strict separation of responsibilities, as required by the project guidelines.
- **Maxime Farré — Quant A (Single Asset Analysis)** · [@MaximeFARRE](https://github.com/MaximeFARRE)
Responsible for:
- single-asset data loading and preprocessing,
- implementation of quantitative strategies on one asset,
- backtesting logic and performance metrics,
- visualization of asset price vs strategy performance.
- **Emilien Combaret — Quant B (Multi-Asset Portfolio Analysis)** · [@EmilienCombaret](https://github.com/EmilienCombaret)
Responsible for:
- extension to multi-asset portfolios (minimum 3 assets),
- portfolio allocation and rebalancing logic,
- portfolio-level metrics and diversification analysis,
- Streamlit user interface for portfolio configuration and results.
Both modules are integrated into a **single unified Streamlit application**.
---
## 3. Data Sources & API
Market data is retrieved from a **public financial API** through the Quant A data loader
(e.g. *yfinance* or an equivalent public data provider).
Key characteristics:
- daily and intraday data (`1d`, `60m`, `15m`, `5m`),
- OHLCV data handling,
- support for MultiIndex formats,
- automatic refresh approximately every 5 minutes,
- robust handling of missing or invalid data.
---
## 4. Quant A — Single Asset Module
### Objective
Analyze and backtest **one asset at a time** (equities, FX, commodities, etc.).
### Implemented strategies
- **Buy & Hold**
- **Moving Average Crossover**
- **Regime Switching (Trend / Mean Reversion)**
### Metrics
For each strategy:
- cumulative return,
- annualized return,
- annualized volatility,
- Sharpe ratio,
- maximum drawdown.
### Visualization
- main chart showing:
- raw asset price,
- cumulative strategy value (base 100),
- interactive controls for:
- strategy parameters,
- data frequency selection.
---
## 5. Quant B — Multi-Asset Portfolio Module
### Objective
Extend the analysis to a **portfolio of multiple assets** (at least 3 simultaneously).
### Portfolio construction
Two portfolio modes are available:
1. **Fixed weights**
- equal-weight,
- custom user-defined weights.
2. **Strategy-based allocation**
- portfolio weights derived from Quant A strategy signals.
### Allocation & rebalancing
- allocation rules:
- equal-weight,
- inverse volatility,
- rebalancing frequency:
- none,
- daily,
- weekly,
- monthly.
### Portfolio metrics
The module computes:
- portfolio value and returns (base 100),
- annualized volatility and return,
- CAGR,
- Sharpe ratio,
- maximum drawdown,
- correlation matrix,
- annualized covariance matrix,
- diversification ratio,
- effective number of assets,
- risk contributions by asset.
### Visualization
- comparison of individual assets vs portfolio value,
- correlation, covariance and distance heatmaps,
- tables of weights, volatilities and risk contributions.
---
## 6. Application Structure
```text
PROJET/
├── .streamlit/
│ └── config.toml # Streamlit configuration (layout, theme)
│
├── app/
│ ├── quant_a/ # Quant A — Single Asset module
│ │ ├── data_loader.py # Market data retrieval (API)
│ │ ├── strategies.py # Single-asset strategies
│ │ ├── metrics.py # Performance metrics
│ │ ├── optimizers.py # Portfolio/strategy optimizers
│ │ ├── daily_report.py # Daily report generation (cron)
│ │ ├── ui_quant_a.py # Streamlit UI for Quant A
│ │ ├── universe.py # Asset universe definition
│ │ └── readme_quant_a.md # Quant A technical documentation
│ │
│ ├── quant_b/ # Quant B — Multi-Asset Portfolio module
│ │ ├── data_adapter.py # Adapter to reuse Quant A data loader
│ │ ├── strategies.py # Multi-asset strategies
│ │ ├── portfolio.py # Portfolio valuation logic
│ │ ├── backtest.py # Portfolio backtesting & turnover
│ │ ├── metrics.py # Portfolio & diversification metrics
│ │ └── page_quant_b.py # Streamlit UI for Quant B
│ │
│ └── __init__.py
│
├── docs/ # Technical documentation
│ └── api_quant_a.md # Quant A public API reference
├── reports/ # Daily reports generated via cron
├── screenshots/ # Application screenshots
├── main.py # Streamlit application entry point
├── requirements.txt # Python dependencies
├── LICENSE
└── README.md
```
---
## 7. Installation
```bash
# 1. Clone the repository
git clone https://github.com/MaximeFARRE/Projet-Python-Git-A4.git
cd Projet-Python-Git-A4
# 2. Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
# 3. Install dependencies
pip install -r requirements.txt
# 4. Launch the dashboard
streamlit run main.py
```
Open your browser at `http://localhost:8501`.
### Automate daily reports with cron (Linux / macOS)
Add the following line to your crontab (`crontab -e`) to generate a CAC 40 report every weekday at 18:00:
```
0 18 * * 1-5 cd /path/to/project && /path/to/.venv/bin/python -m app.quant_a.daily_report >> logs/cron.log 2>&1
```
---
## 8. Screenshots
*Screenshots will be added here once the application is running.*
---
## 9. Known Limitations
- **Quant B integration**: some features described in section 5 are developed on the `quant_b` branch and require a merge to run end-to-end.
- **No transaction costs**: the backtester does not model bid-ask spreads or commissions. All results are gross of fees.
- **Intraday data depth**: Yahoo Finance limits intraday history to the last 5–7 calendar days depending on the interval.
- **In-sample optimization**: the parameter grid search runs on the full selected period — no walk-forward or out-of-sample validation is performed.
- **ML forecasting**: the linear regression model in Quant A is illustrative only and is not intended for production trading signals.
---
## License
This project is licensed under the MIT License. See [LICENSE](LICENSE) for details.