The online demo is available on streamlit here
This fork is an improvement done on the original portfolio optimization, it also condenses everything into a class so it can be ran with a few lines. The main algorithms have not been modified much. It uses yfinance as the provider for historical data and is optimized using scipy.minimize function and Monte Carlo method.
Right now, the library is not hosted on PyPi so you will need to do a local install on your system if you plan to use it in other scrips you use.
From the terminal, run this command
pip install git+https://github.com/wheynelau/portfolio-optimizationThis will install all the dependencies listed in the setup.py file. Once done
you can use the library wherever you want.
Potential issues: As this is my first repo if the setup does not work I may not know how to fix, please do just copy the code if it doesn't work. In the meantime I will also figure out how to improve this.
Example to optimize a portfolio for crypto and stocks
Personal suggestion: Do a quick google for the stock if you can't find it, I didn't know that to look for VWRA I had to input "VWRA.L".
import pandas as pd
from pyopt.client import PyOpt
# Define the symbols: Stocks and/or crypto (Actual input is not case-sensitive)
stocks = ['AAPL', 'MSFT', 'SQ']
crypto = ['BNB','BTC']
# Initialize the client.
pyopt = PyOpt() <--- This can be done before or after defining the symbols
# Add tickers to the optimizer
pyopt.add_stocks(stocks)
pyopt.add_crypto(crypto)
# Run optimizer
pyopt.run()- Missing error handling on some inputs
- Clean up code and make it more readable
- Improve github skills, in terms of formatting, version control, pull requests etc
- Create new exception class for PyOpt
- PySimpleGUI
- Minumum weightage purchase. (Useful if low capital and no fractional shares)
- Print results to html for readability and save it for future reference
- Input using user provided .csv files
- Max weightage option Update: Not specified securities, but for all
- Weightage for individual securities . Would be better with gui or menu based
- Experiment getting input using **kwargs
- Provide more settings
Still a student so creating this was just a hobby. Providing feedback or criticisms will greatly help me. Giving me a star helps too.
I added some TODO or ideas for improvements but due to exams I did not manage fix many of them. If there are issues please post them so I can try to fix them.
Credits to areed1192's github and his youtube for motivating and teaching me the algorithm's for Monte Carlo and scipy.minimize. You can check out his infomative youtube channel here!