import finsets as fds
finsets
Download and process datasets commonly used in finance research
Each module handles a different data source. Almost all submodules (other than utility ones) have a get_raw_data
function that downloads the raw data and a process_raw_data
function that processes the data into a pandas.DataFrame
having, as index, either:
- A
pandas.Period
date reflecting the frequency of the data (for time-series datasets), or - A
pandas.MultiIndex
with a panel identifier in the first dimension and apandas.Period
date in the second dimension (for panel datasets).
The period date in the index will be named following the pattern Xdate
where X is the string literal representing the frequency of the data (e.g. Mdate
for monthly data, Qdate
for quarterly data, Ydate
for annual data).
Install
pip install finsets
How to use
or
from finsets import fred, wrds, papers
Below, we very briefly describe each submodule. For more details, please see the documentation of each submodule (they provide a lot more functionality than presented here).
WRDS
Downloads and processes datasets from Wharton Research Data Services WRDS.
Each WRDS module handles a different library in WRDS (e.g. compa
module for the Compustat Annual CCM file, crspm
for the CRSP Monthly Stock file, etc.).
Before you use any of the wrds
modules, you need to create a pgpass
with your WRDS credentials. To do that, run
from finsets.wrds import wrds_api
= wrds_api.Connection() db
This will prompt you for your WRDS username and password. After you enter your credentials, if you don’t have a pgpass
file already set up, it will ask you if you want to do that. Hit y
and it will be automatically created for you. After this, you will never have to input your WRDS password.
You will still have to supply your WRDS username to functions that retrieve data from WRDS (all of them have a wrds_username
parameter). If you don’t want to be prompted for the username for every download, save it under a WRDS_USERNAME
environment variable:
- On Windows, in a Command Prompt:
setx WRDS_USERNAME "your_wrds_username_here"
- On Linux, in a terminal:
echo 'export WRDS_USERNAME="your_wrds_username_here"' >> ~/.bashrc && source ~/.bashrc
- On macOS, since macOS Catalina:
echo 'export WRDS_USERNAME="your_wrds_username_here"' >> ~/.zshrc && source ~/.szhrc
- On macOS, prior to macOS Catalina:
echo 'export WRDS_USERNAME="your_wrds_username_here"' >> ~/.bash_profile && source ~/.bash_profile
The functions in the wrds_
modules will close database connections to WRDS automatically. However, if you open a connection manually, as above (with wrds.Connection()
) make sure you remember to close that connection. In our example above:
db.close()
Check the wrds_utils
module for an introduction to some of the main utilities that come with the wrds
package.
FRED
Downloads and processes datasets from the St. Louis FRED.
To use the functions in the fred
module, you’ll need an API key from the St. Louis FRED.
Get one here and store it in your environment variables under the name FRED_API_KEY
Alternatively, you can supply the API key directly as the api_key
parameter in each function in the fred
module.
= fred.fred.get_raw_data(['GDP']) gdp
'info'] gdp[
id | realtime_start | realtime_end | title | observation_start | observation_end | frequency | frequency_short | units | units_short | seasonal_adjustment | seasonal_adjustment_short | last_updated | popularity | notes | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0 | GDP | 2023-11-15 | 2023-11-15 | Gross Domestic Product | 1947-01-01 | 2023-07-01 | Quarterly | Q | Billions of Dollars | Bil. of $ | Seasonally Adjusted Annual Rate | SAAR | 2023-10-26 07:55:01-05 | 92 | BEA Account Code: A191RC Gross domestic produ... |
'Q'] gdp[
GDP | |
---|---|
1947-01-01 | 243.164 |
1947-04-01 | 245.968 |
1947-07-01 | 249.585 |
1947-10-01 | 259.745 |
1948-01-01 | 265.742 |
... | ... |
2022-07-01 | 25994.639 |
2022-10-01 | 26408.405 |
2023-01-01 | 26813.601 |
2023-04-01 | 27063.012 |
2023-07-01 | 27623.543 |
307 rows × 1 columns
PAPERS
Downloads and processes datasets made available by the authors of academic papers.
Each papers
module handles a different paper. The naming convention is that the module’s name is made up of the last names of the authors and the publication year, separated by underscores. If more than two authors, all but the first author’s name is replaced by ‘etal’. For example, the module for the paper “Firm-Level Political Risk: Measurement and Effects” (2019) by Tarek A. Hassan, Stephan Hollander, Laurence van Lent, Ahmed Tahoun is named hasan_etal_2019
.
papers.hassan_etal_2019.list_all_vars().head()
name | |
---|---|
0 | gvkey |
1 | date |
2 | PRisk |
3 | NPRisk |
4 | Risk |