Bollerslev, T. (2022). Realized semi (co) variation: Signs that all volatilities are not created equal. Journal of Financial Econometrics, 20(2), 219-252.
Data collection is done using the scripts taq_scrape.sas and taqmsec_scrape.sas. These scripts can be modified to collect clean five-minute prices for any set of stocks (by permno, cusip, or symbol) for any particular day. For data from 2014 or earlier, use taq_scrape.sas. These scripts come with some placeholder arguments, but can be modified programmatically as in get_taq_data_simple.ipynb; this notebook references metadata for CRSP that is too large to be included in this repository. However, its fairly easy to input your own list of tickers/permnos/cusips. The notebook generates SAS scripts that collect data for the given stocks for each day in the given dates. These scripts can be run on WRDS using autogenerated shell scripts; the command to run all the scripts in parallel is also generated by the notebook. The scripts output csv files which go through some minor cleaning/reformatting before being saved as parquet files. These processed files should be saved locally in a new folder called data/taq/prices. Additionally, CRSP data is also collected for adjusted overnight returns. This is obtained using the get_crsp_daily.ipynb notebook; this notebook also preprocesses the raw SAS output.
Data cleaning is done by merging the TAQ data with CRSP data. The majority of the cleaning is done by clean_prices.ipynb. The output is saved to data/proc/clean_prices.
Data Analysis is done in three notebooks. The main notebook loads all the data, does some additional data processing and computes volatility measures. Most of the processed dataframes are saved to the data folder; the dataframe containing high-freq prices for all tickers is not included in the repo due to its size. This notebook concludes with the SHAR regressions. The return regressions are done in spy_return_regs.ipynb while the semicovariance/semibeta results are computed in semitables.ipynb. All figures/tables generated by the notebooks are saved to the exhibits folder.