Each of you has to download a time series representing some stock returns (or any other time se?ries like trading volumes,

inflation, GDP) from Yahoo Finance, Google Finance or Datastream. For Datastream you have to go to the computer in X?. Using

the Box-Jenkins methodology, propose a model that best captures the features of your chosen stock returns. Your approach has

to be based on
plot of the series

unit root tests

plots of ACF/PACF

information criteria

tests for serial correlation

tests for heteroskedasticity

F -statistics and t-statistics

nonparametric kernel density estimate

and may include other tests that you know and/or have been discussed in class/lecture. Every time you compute a test

statistic
state the null hypothesis that you are testing.
give the distribution of the test statistic under the null hypothesis (with degrees of freedom if this is the case).
interpret the outcome of the tests either using a critical value or/and a p value. Mention how the p value was computed.
state the assumptions on which the test is based.
Create a table mentioning the competing models and the values of the corresponding infor-mation criteria. Once you have

chosen the ?best? model create a table with the parameter estimates, their standard error or the corresponding t? statistic

and briefly comment on it.
The project has to be in the form of a short article (with introduction, main results, conclusion and other (sub)sections if

you wish) of maximum 10 pages including tables and figures. You can (and are encouraged to) read articles related to the

topic you wish to study in the project.