Stats Lab for Assets

This project is a personal statistical analysis tool for financial assets.

I’m a Computer Engineering student with a strong interest in investing and finance, and I wanted a reliable and transparent way to measure assets statistically, beyond just looking at prices or past returns.

The goal is to:

• understand how different assets behave (risk, skewness, drawdowns, persistence)
• compare assets consistently
• use these statistics to make better portfolio and strategy decisions

This is not a trading bot.
It’s a diagnostic and analysis tool.

What this tool does

Given historical price data for an asset, it computes and visualizes:

Performance & Risk Metrics

• CAGR (annualized growth)
• Mean return (daily & annualized)
• Volatility (annualized)
• Downside volatility (annualized)
• Skewness of returns
• Autocorrelation (lag 1)
• Sharpe ratio
• Return percentiles (tails & median)
• Max and average drawdown

Visual Diagnostics

It automatically generates and saves:

• Price chart
• Equity curve
• Drawdown plot
• Return distribution (with skew)
• Rolling downside volatility

All plots are saved as PNG files, grouped per asset and time window.

Project structure

stats-lab-for-assets/
├── data/                   # CSV price datasets
├── images/                 # Generated plots
├── notebooks/              # EDA notebooks
├── src/
│   ├── data_process.py     # Download and preprocess data
│   ├── metrics.py          # Statistical analysis (Asset class)
│   ├── visualization.py    # Plots
│   └── main.py             # Main
├── environment.yml         # Conda environment
└── README.md               # Documentation

Quick start (Docker)

You only need Docker and Docker Compose installed.

Build the image (one time)

docker compose build

Getting data

Historical price data is downloaded using Yahoo Finance.

To fetch and store data for an asset, run:

python3 src/data_process.py GLD

This will:

• download data
• clean it
• save a csv file into the data/ folder

! You can change the data dates on the function get_data() !

Usage

This project has two main parts:

• metrics.py → computes all statistics for an asset
• visualization.py → generates and saves plots as PNG files

Usage (needs the csv data in the data folder):

docker compose run stats-lab GLD GLD_2000-01-01_to_2020-01-01 (or whatever the years of the csv)

Important to note:

• csv file name without the ending .csv
• plots are saved in folder images
• it prints a summary in the terminal

Example output:

With S&P500 data from 2000 to 2025: