Financial Modelling using Jupyter Notebook PYTHON

Performing Portfolio Optimisation and Evaluating Value-at-Risk.

You are given a list of securities and you are required to produce a jupyter notebook that downloads data about these securities from either Bloomberg, or yfinance or WRDS and explains:

How the data was downloaded

How to choose the weights of the securities from that list should be in your portfolio

What is the forecasted value at risk of your portfolio for one month at 95% confidence.

The securities are:

Google stock (any flavour of that stock will do)

Amazon stock

Tesla stock

Goldman Sachs stock

DAX total return index

BNDX ETF traded in the New York Stock Exchange

Savings account paying 3.5% per year in USD completely risk free

Your portfolio currency is the US Dollar – remember that the DAX is quoted in Euros, so you will need to account for foreign exchange rate risk in your Value-at-Risk.

To calculate your Value-at-Risk, you should use a normal distribution approximation. Your document needs to explain your rational behind the choice of of mean and standard deviation of your normal distribution.

To gain full marks, the standard deviation must be based on a conditional heteroskedastic variance model – but if you can’t get it running well, you can use a flat variance number, as it will be better than no number 🙂

You should start your project by making a copy of the Week 10 notebook and reusing the code for the yfinance download and return resample.. Only when your notebook is working with yfinance you update it to Bloomberg – so, if you can’t get Bloomberg to work, then at least you have a yfinance version. Also, you can reuse the code in Week 6 + Week 8 to calculate Value at Risk and Portfolio Optimisation.

NOTE: The Bloomberg terminals do not have Python installed and due to licencing arrangements the University is not allowed to install Python on those computers, so if you are downloading data from Bloomberg you will have to save it as a csv file and bring into your notebook using the function read_csv.

Finally – if you simply can’t code anything at all and made no progress; don’t panic. Write only the text section of the Jupyter notebook and explain the entire theoretical part of this notebook. If you write a very good theory with no coding whatsoever, that will give you already 50% of the marks. This assessment will be marked 50% on theory and 50% on practical implementation.