FinProject: COMPASS SM1/SC1 financial data project
Project Brief
Yahoo Finance is one possible source of historical financial time series data. We will mostly focus on the stock price history of large US companies, in the S&P 500 index. The current constituents of the S&P 500 index can be found here, among many other sources. It may be cumbsersome to manually download stock prices of all 500 companies, so we may have to find better data source.
One possible project is porfolio optimisation, i.e. to select the best asset distribution subject to certain constraints so that the expected return is maximized while financial risk is minimized. It is important to first construct the covariance matrix for the rates of return on the assets in the portfolio. The optimisation problem itself may be constrained by many factors, e.g. whether short selling is allowed. Transaction cost is another consideration that will impact on the frequency of adjustment to the portfolio.
Layout
-
Package: access with
install_github("Dom-Owens-UoB/FinProject", subdir="portopt"). Features Principal Components Analysis in S4; Support Vector Machine for Classification; Gaussian Process Fitting; Fitting a Factor Model; Portfolio Optimisation via Analytical Solution (with short selling); Comparing Portfolio Performance;
Plan
-
Factor analysis notes
SM1 Content:
-
VAR: method explanation, application, results
-
Gaussian Process classification: method explanation, application, results
-
forecast, cost comparison
SC1 Content:
-
Package: data, functions, testing
-
Visualisation (time series, forecasts...)
-
Debugging