Data Sources

QuantEx separates market data ingestion from strategy logic. All data providers implement the abstract DataSource interface.

Class	Use-case
CSVDataSource	Local OHLCV CSV files.
ParquetDataSource	Columnar Parquet files (fast).
(abstract) BacktestingDataSource	Contract for fixed-length sources.

CSV Example

from quantex.sources import CSVDataSource

ds = CSVDataSource("prices.csv", symbol="AAPL")
print(len(ds))        # number of rows
print(ds.peek_timestamp())  # first timestamp

Parquet Example

from quantex.sources import ParquetDataSource

ds = ParquetDataSource("btc.parquet", symbol="BTC-USD")
bar = ds.get_current_bar()
print(bar.close)

Required Columns

All OHLCV sources must provide: timestamp, open, high, low, close, volume.

Custom Sources

Implement:

class MySource(BacktestingDataSource):
    def __len__(self): ...
    def get_raw_data(self): ...  # pandas.DataFrame indexed by timestamp
    def get_current_bar(self): ...
    def get_lookback_data(self, n): ...
    def peek_timestamp(self): ...

Return types mirror the built-ins so the engine remains agnostic.