This project is a first attempt at using mean-variance optimization to optimize stock portfolio.
This project is interested in learning to optimize the risk and return associated with stocks in a portfolio. This project aims to 1) produce a portfolio that minimizes risk and 2) find the weightage of stock combinations that has the lowest risk for a specific return rate in mind.
First, this project uses the Mean-Variance Portfolio (MVO) Theory to find the stock weightage of a portfolio that minimizes risk for a specific annual return rate. This is studied using blue-chip stocks in the US markets that has been listed on the stock market since 1990s so that there is sufficient data to perform Mean-Variance Optimization.
Subsequently, this project uses the MVO to optimize for stock weightage in a portfolio that minimizes risk on a weekly and monthly basis.
Stock market data is collected using the AlphaVantage API through the AlphaVantage.jl package in Julia. An API key is required to obtain data using the API.
As it is too slow to make an API call and optimize at the same time due to large dataset and many number of API requests, API calls were made and data were stored as JLD2 files. Data files can be found in the data folder.
To obtain data for different stocks, simply obtain the API key and input the desired ticker symbol.
The optimization used is as follow, where the variance (risk) is minimized
where
- the portfolio in this project does not combine high growth and blue chip stocks into the same portfolio, making the individual portfolio rather small
- this project only uses stocks from US market} and this may not provide sufficient diversification
- this project is based on historical data and the past does not necessarily tells us future returns and risks
- the optimization in this project does not constraint weightage of stocks in portfolio based on industry