quantlab.acquisition.base.Capability#

class quantlab.acquisition.base.Capability(market, frequency, data_type=None, earliest_available=None, entitlement=None, dataset_cls=None, symbol_batch_cls=None, config_factory=None)[source]#

Bases: object

One (market, frequency, data_type) combination a vendor serves.

A descriptor lists its capabilities one by one rather than as every pairing of its markets and frequencies, because vendors serve irregular combinations. For example, Alpaca serves tick data for US equities only, and splits it into quotes and trades. Anything that differs between combinations, such as the earliest available date or the dataset class that converts the data, is stored here rather than on the descriptor. The class is frozen; extend it by adding optional fields.

Parameters:

marketMarket: The market, for example "us_equity".
frequencyFrequency: The bar frequency, for example "1d" or "tick".
data_typestr or None, default None: "bars", "quotes", "trades" and so on, or None when the vendor makes no such distinction. None means “no data type”, not “any data type”.
earliest_availablestr or None, default None: ISO date of the earliest data the vendor serves, when known. It is informational; nothing checks against it.
entitlementstr or None, default None: The subscription this capability needs, when the vendor sells several.
dataset_clstype[MarketDataset] or None, default None: The dataset class that converts this capability’s raw tier into Zarr, or None when no dataset can hold the data. Tick data, for example, arrives at irregular times and does not fit a regular (timestamp, symbol) grid. convert() refuses a capability whose dataset_cls is None, so adding a conversion means filling this field, not adding a branch to convert().
symbol_batch_clstype[SymbolBatchAcquisition] or None, default None: The class that downloads this capability, or None to use the descriptor’s own symbol_batch_cls. One vendor account can serve several products through different classes; WRDS, for example, serves both TAQ quote data and CRSP daily data.
config_factorycallable or None, default None: The function that builds an AcquisitionConfig for this capability, or None to use the descriptor’s own config_factory.

Notes

The class fields are direct class references, never dotted import paths. They are annotated with the abstract base classes so that the top of this module imports no vendor module.

Examples

>>> from quantlab.acquisition.base import Capability
>>> cap = Capability(market="us_equity", frequency="1d", data_type="bars")
>>> cap.data_type
'bars'
>>> cap.dataset_cls is None
True