## Financial Engineering and Machine Learning FINA1147

You will use one market index and four companies’ daily data and these companies should be
from two different sectors. (The required data can be downloaded from Yahoo finance:
http://uk.finance.yahoo.com/). The sample period should be latest and at least 2 years in
length (for example, from Jan 2018 to Jan 2020). For the volatility forecast, the required FX
data can be downloaded from the course Moodle page. For the empirical analysis, you can
use statistical software such as EViews, STATA, or SPSS etc., which has to specify in the
report.
A. Mean-Variance Optimization
1. Briefly explain mean-variance portfolio optimization.
(5 Marks)
2. Estimate the covariance matrix for the selected four companies’ stocks.
(5 Marks)
3. Plot by creating portfolios using the selected four companies and the obtained covariance
matrix. Discussing the results of the portfolio.
(10 Marks)
B. Panel Data Analysis
1. Construct a panel data set using the latest 100 days of the four stock prices. Transfer the
stock prices, market index, and risk-free rate into log returns.
(8 Marks)
2. Verify the CAPM theory using OLS, FE, and RE estimators. Select the appropriate model
(OLS, FE or RE). Discuss the obtained regression results.
(12 Marks)
C. Time-Series Data Analysis
1. Choose one of your stock price series, compute ACF and PACF for the log returns. Discuss
the results.
(4 Marks)
2. Forecast the log returns with AR(5) model and verify the forecasting accuracy by
considering the last 6 months of the data as out-of-sample.
(8 Marks)
3. Estimate the ARMA(3,2) model and comment on the estimations.
(8 Marks)
D. Volatility Analysis
1. Choose one of your stock price series, verify the ARCH effect and estimate GARCH(1, 1)
model.
(10 Marks)
2. Using the data provided, carry out a GARCH(1, 1) volatility forecasting by considering the
last 6 months of the data as out-of-sample. Discuss the results.
(10 Marks)
E. Discussions on Machine Learning Application
1. Discuss the following concepts.
a) Machine Learning
b) Supervised Learning
c) Differentiate between test set and training set
(12 Marks)
2. Explain your understanding on neural network in machine learning. Provide one possible
application of neural network in the financial practice and explain the processes.
(8 Marks)
You need to do all the above tasks and submit your results with detailed discussion on the
tests in a report form (using academic style and minimum 2000 words).
deduction of marks ranging from 10 to 30 will be applied. You will use one market index and four companies’ daily data and these companies should be
from two different sectors. (The required data can be downloaded from Yahoo finance:
http://uk.finance.yahoo.com/). The sample period should be latest and at least 2 years in
length (for example, from Jan 2018 to Jan 2020). For the volatility forecast, the required FX
data can be downloaded from the course Moodle page. For the empirical analysis, you can
use statistical software such as EViews, STATA, or SPSS etc., which has to specify in the
report.
A. Mean-Variance Optimization
1. Briefly explain mean-variance portfolio optimization.
(5 Marks)
2. Estimate the covariance matrix for the selected four companies’ stocks.
(5 Marks)
3. Plot by creating portfolios using the selected four companies and the obtained covariance
matrix. Discussing the results of the portfolio.
(10 Marks)
B. Panel Data Analysis
1. Construct a panel data set using the latest 100 days of the four stock prices. Transfer the
stock prices, market index, and risk-free rate into log returns.
(8 Marks)
2. Verify the CAPM theory using OLS, FE, and RE estimators. Select the appropriate model
(OLS, FE or RE). Discuss the obtained regression results.
(12 Marks)
C. Time-Series Data Analysis
1. Choose one of your stock price series, compute ACF and PACF for the log returns. Discuss
the results.
(4 Marks)
2. Forecast the log returns with AR(5) model and verify the forecasting accuracy by
considering the last 6 months of the data as out-of-sample.
(8 Marks)
3. Estimate the ARMA(3,2) model and comment on the estimations.
(8 Marks)
D. Volatility Analysis
1. Choose one of your stock price series, verify the ARCH effect and estimate GARCH(1, 1)
model.
(10 Marks)
2. Using the data provided, carry out a GARCH(1, 1) volatility forecasting by considering the
last 6 months of the data as out-of-sample. Discuss the results.
(10 Marks)
E. Discussions on Machine Learning Application
1. Discuss the following concepts.
a) Machine Learning
b) Supervised Learning
c) Differentiate between test set and training set
(12 Marks)
2. Explain your understanding on neural network in machine learning. Provide one possible
application of neural network in the financial practice and explain the processes.
(8 Marks)
You need to do all the above tasks and submit your results with detailed discussion on the
tests in a report form (using academic style and minimum 2000 words).