edgar-10k-mda

This repo contains some python code I used to download form10k filings from EDGAR database, and then extract the MDA section from the downloaded form10k filings heuristically

Installation

I used python3.6

#python36
pip install -r requirements.txt

Quick Start

Specify the starting year and end year and the directory to save outputs. By default, indices, forms and mdas will be saved to ./data

# Downloads and parses MDA section from 2016 to 2016 quarter 1 and 2, and saves to `./data/`
python edgar.py --start_year 2016 --end_year 2016 --quarters 1 2 --data_dir ./data/

Usage

usage: edgar.py [-h] -s START_YEAR -e END_YEAR [-q QUARTERS [QUARTERS ...]]
                [-d DATA_DIR] [--overwrite] [--debug]

optional arguments:
  -h, --help            show this help message and exit
  -s START_YEAR, --start_year START_YEAR
                        year to start
  -e END_YEAR, --end_year END_YEAR
                        year to end
  -q QUARTERS [QUARTERS ...], --quarters QUARTERS [QUARTERS ...]
                        quarters to download for start to end years
  -d DATA_DIR, --data_dir DATA_DIR
                        path to save data
  --overwrite           If True, overwrites downloads and processed files.
  --debug               Debug mode

Workflow

The code runs the extraction in the following steps

1. Download indices for form 10k to ./data/index
2. Combines all indices into a single csv ./data/index/combined.csv
3. From Step2 combined csv, downloads all form 10k to ./data/form10k
4. Parses the html forms with BeautifulSoup to ./data/form10k.parsed
5. Parses MDA section to ./data/mda

Notes

• MDA section is parsed heuristically, and may not work for all forms. You'll probably need to modify the find_mda_from_text function for coverage.
• You also might need to modify normalize_text function for MDA parsing.