ML4T Data¶
Market data acquisition, storage, and update workflows for machine learning for trading.
ml4t-data provides one interface for acquiring provider data, validating observations, writing
local datasets, and updating those datasets incrementally. Provider adapters cover exchange,
vendor, public, and synthetic data sources. Each provider documents its own credentials, rate
limits, licensing, and optional dependencies.
-
First successful use
Install the wheel and generate deterministic OHLCV data without credentials or network access.
-
Choose a provider
Compare supported asset classes, authentication requirements, and provider capabilities.
-
Maintain datasets
Configure storage, detect gaps, and update existing data without replacing valid history.
-
Use the public interface
Inspect provider protocols, configuration models, storage interfaces, and exceptions.
Quick example¶
from ml4t.data.providers import SyntheticProvider
provider = SyntheticProvider(seed=42)
data = provider.fetch_ohlcv("SYNTH", "2024-01-01", "2024-01-10", "daily")
assert not data.is_empty()
The synthetic provider is local and deterministic. Live providers require network access and may require an account, credentials, paid data, or an optional package.
Start here¶
- Installation describes supported Python versions and extras.
- Quickstart teaches the first offline workflow.
- User guide contains task-oriented storage, update, and validation guides.
- Provider guide records provider capabilities and external-service boundaries.
- Migration guide maps former QLDM names to current interfaces.
- Contributing explains development setup and quality checks.