In this subject, you should replicate the results of Clare and Thomas (1995) for a given country of your choice. Try to use as many stocks as you can (at least 100) and also as many time periods as possible (at least 36 years). Given the discussions we had in the class, use selection/monitoring periods of 12 months and 24 months [you may choose other periods apart from these as you please].
I recommend that you follow these steps when doing this assignment:
a) Read Brooks (2019, section 3.12) carefully.
b) Read Clare and Thomas (1995) carefully. Note the important highlights that help you with understanding the overreaction hypothesis and writing out the introduction section. Further, take notes on the methodology used and collect your notes and write out the methodology section.
c) (Optional but helpful) Check [using google] which other papers have been doing a similar study on the overreaction hypothesis and take a brief look at their methodology, data section and results. This step will expose you to the literature around the topic.
d) Decide which set of stocks [country] you wish to focus on. Explain the data collection procedure briefly and the data itself, for example make a table of the firms you wish to use and note their symbols [you use these in Python to collect data on]; mention which is the time period you have considered and mention the data frequency.
e) Prepare your Python codes and think carefully of which results do you wish to report in which form [you may use tables and/or graphs].
f) Obtain and summarize the results and comment on them.
g) Conclude.
h) (Do not forget to use titles/legends for your Python figures (if you choose to use these), so that they are readable. Add comments in your Python codes, so that it becomes easier to read/evaluate them.
Task 2)
a) Choose 3 stocks (or financial indices) and calculate their simple daily returns. The data should be from 01/01/2002 up to 01/04/2022.
b) Plot the returns.
c) Provide a table including mean, standard deviation, median, 20% and 80% quantiles, skewness and kurtosis for each series.
d) Provide histograms for the return series [superimpose a normal density plot].
e) For each series fit a GARCH(p, q) model and plot the in-sample estimates of the standard deviations provided by the model you choose for each series [you should choose p and q for each series using BIC]. Contrast the estimated standard deviations from your GARCH(p, q) model with (i) rolling window standard deviations [with window sizes of 20 and 50 days] and (ii) with estimates of the standard deviations obtained using EWMA [choose and use two different values for λ].
I recommend that you follow these steps when doing this assignment:
a) Read Brooks (2019, chapter 9) carefully.
b) Explain in your solution what are conditional heteroskedasticity models, in particular focus on GARCH(p,q). Also briefly explain the EWMA and the rolling window estimators for the standard deviation.
c) Decide which set of stocks/indices you wish to focus on. Explain the data collection procedure briefly and the data itself, for example make a table of the firms you wish to use and note their symbols and also mention which is the time period you have considered for data.
d) Produce nice plots [as outlined above] and try to make your plots looking as nice as you can: make use of colors, markers and everything
have learned in the class.
e) Comment on your results.
f) (Do not forget to use titles/legends for your Python figures, so that they are readable. Add comments in your Python codes, so that it becomes easier to read/evaluate them.)