quantlab.dataset.config.MassiveTenKChangesConfig#

class quantlab.dataset.config.MassiveTenKChangesConfig(*, zarr_file_path, start_date=None, end_date=None, symbols=None, kwargs=None, resample_freq=None, resample_how=None, name=None, raw_data_dir_path, market='us_equity', frequency='1d', vendor='massive', table='sep', permatickers=None, roster_universe=None, category_filter='default', massive_dir, stale_after_days=400, workers=4)[source]#

Bases: SharadarDatasetConfig

Config of the 10-K text-change panel on the permaticker axis.

Massive’s 10-K sections (the REST raw tier rest/10k_sections under massive_dir) are compared with the same filer’s previous annual report, keyed by SEC CIK and placed on every SEP security whose TICKERS secfilings carries the CIK. raw_data_dir_path is Sharadar’s raw tier: SEP’s dates are the calendar and SEP’s TICKERS rows the universe, whose fields work as for a price panel (the default is domestic common stock).

Examples

>>> cfg = MassiveTenKChangesConfig(
...     zarr_file_path="/data/zarrs/massive_10k_changes_1d.zarr",
...     raw_data_dir_path="/data/downloads/sharadar",
...     massive_dir="/data/downloads/massive",
... )
>>> cfg.vendor, cfg.table, cfg.stale_after_days, cfg.workers
('massive', 'sep', 400, 4)

massive_dir#: Massive’s raw tier (<download-dir>/massive), holding rest/10k_sections.

stale_after_days = 400#: Days after its filing date a report stops being shown when no newer one has replaced it: a little over a year, so a skipped report shows nothing.

table = 'sep'#: Always SEP: its TICKERS rows give the universe and the CIKs, its dates the calendar.

vendor = 'massive'#: Massive’s text; the calendar and the universe are Sharadar’s.

workers = 4#: Processes the documents are vectorized on (joblib).