adapters

Concept Overview

The Rust core returns results as plain dicts and lists. The adapters module converts these into typed Polars DataFrames with proper datetime parsing, column naming, and validation. This is the standard way to move data between the Rust computation engine and Python analysis/visualization code.

SignalStreamBuffer provides an incremental append interface for streaming workflows where signals arrive in chunks — common in live research notebooks or paper-trading loops.

When to Use

Use adapters whenever you receive output from the Rust core or pipeline module and need DataFrames for analysis, visualization, or further processing. The pipeline module’s _frames variant calls these adapters internally.

Alternatives: Manual Polars DataFrame construction from dicts, but you lose validation and timestamp parsing.

Usage Examples

Python

Convert pipeline outputs to typed DataFrames

from openquant.adapters import (
    to_polars_signal_frame,
    to_polars_weights_frame,
    SignalStreamBuffer,
)

# Signal frame from raw timestamps + values
signals = to_polars_signal_frame(
    timestamps=["2024-01-02T09:30:00", "2024-01-02T09:31:00"],
    signal=[0.5, -0.3],
    side=[1.0, -1.0],
    symbol="CL",
)

# Streaming buffer for incremental signal updates
buf = SignalStreamBuffer()
buf.append(timestamps=["2024-01-02T09:32:00"], signal=[0.1])
buf.append(timestamps=["2024-01-02T09:33:00"], signal=[-0.2])
all_signals = buf.frame()  # concat into single DataFrame

API Reference

Python API

adapters.to_polars_signal_frame
adapters.to_polars_event_frame
adapters.to_polars_backtest_frame
adapters.to_polars_weights_frame
adapters.to_polars_indicator_matrix
adapters.to_polars_frontier_frame
adapters.SignalStreamBuffer
adapters.to_pandas

Key Functions

to_polars_signal_frame
to_polars_event_frame
to_polars_backtest_frame
to_polars_weights_frame
SignalStreamBuffer

Implementation Notes

All adapter functions validate input length alignment before constructing frames.
SignalStreamBuffer supports incremental append for streaming research notebooks.
to_pandas() is available for downstream tools that require pandas; requires pandas to be installed.