Stephen Thiele, "Modelling the Conditional Distribution of Financial
Returns with Asymmetric Tails", Journal of Applied Econometrics,
Vol. 35, No. 1, 2020, pp. 46-60.

There are two folders containing the data and files needed to
reproduce the main results of the paper.  Specifically, there are
files in the folder "Univariate-files", which produces the in-sample
estimation results, and in the folder "Multivariate-files", which
produces the out-of-sample portfolio results.  The analysis was
performed in Matlab (version 2016a).

All data and program files are ASCII files in DOS format. They are
zipped in the file st-files.zip, which contains the two folders.
Unix/Linux users should use "unzip -a".


Univariate files

The data for this paper were obtained from the Oxford-Man Institute
Realized Library in 2014 (https://realized.oxford-man.ox.ac.uk/).  The
data consists of daily returns for the FTSE 100 (UK), S&P 500 (US),
DJIA (US), NASDAQ (US), Kospi (South Korea) and Bovespa (Brazil)
indices between 03/01/2000 and 07/10/2014. The number of trading days
recorded in the database range from 3685 to 3706, so the returns for
each index are stored in the separate file "INDEX_ret.csv".

Running the file "univar_insample.m" produces the point estimates,
standard errors, diagnostics, and QQ plots.  Note that the functions
used to estimate the AST DCS model or the special cases have starting
values as an input, but also have several alternative starting values
as backups built into the function to try to avoid local optima. 
Furthermore, the "univar_insample.m" file also has a section for each
model that does the estimation with a series of randomly generated
starting values.


Multivariate files

The data consists of a set of files with common trading days for pairs
of indices used in the out-of-sample portfolio construction exercise.
The filenames follow the convention "INDEX1_INDEX2_rets.csv", where
the order of the indices in the file name is consistent with the order
of the columns in the file (ie - INDEX1 relates to column 1).

Running the file master_oos_run.m will sequentially call the file
"oos_index1_index2.m" to analyse each pair of indices. Results are
saved down into an excel file for each pair.


Note: Both folders contain a file "jae_data.csv" that contains all six
indices side by side with dates and missing/non-trading days.  This
file does not read directly into any of the code, but is there as a
base file from which the other clean data files can be constructed.