opendeviationbar 13.4.0

Python bindings for opendeviationbar: Non-lookahead open deviation bar construction for cryptocurrency trading

[//]: # SSoT-OK

opendeviationbar-py

High-performance open deviation bar construction for quantitative trading, with Python bindings via PyO3/maturin.

Resource	URL
PyPI	https://pypi.org/project/opendeviationbar/
Repository	https://github.com/terrylica/opendeviationbar-py
Performance Dashboard	https://terrylica.github.io/opendeviationbar-py/
API Reference	docs/api/INDEX.md
Issues	https://github.com/terrylica/opendeviationbar-py/issues

Installation

pip install opendeviationbar

Pre-built wheels: Linux (x86_64), macOS (ARM64), Python 3.13. Source build requires Rust toolchain and maturin.

Quick Start

from opendeviationbar import get_open_deviation_bars

# Fetch data and generate open deviation bars in one call
df = get_open_deviation_bars("BTCUSDT", "2024-01-01", "2024-06-30")

# Use with backtesting.py
from backtesting import Backtest, Strategy
bt = Backtest(df, MyStrategy, cash=10000, commission=0.0002)
stats = bt.run()

Output: pandas DataFrame with DatetimeIndex and OHLCV columns, compatible with backtesting.py.

API Overview

Function	Use Case
get_open_deviation_bars()	Date-bounded, auto-fetch
get_n_open_deviation_bars()	Exact N bars (ML training)
process_trades_polars()	Polars DataFrames (2-3x faster)
process_trades_chunked()	Large datasets (>10M trades)
populate_cache_resumable()	Long ranges (>30 days)
run_sidecar()	Real-time streaming sidecar

# Count-bounded (ML training)
from opendeviationbar import get_n_open_deviation_bars
df = get_n_open_deviation_bars("BTCUSDT", n_bars=10000)

# Polars (2-3x faster)
import polars as pl
from opendeviationbar import process_trades_polars
bars = process_trades_polars(pl.scan_parquet("trades.parquet"), threshold_decimal_bps=250)

# With microstructure features (57 columns: OFI, Kyle lambda, Hurst, etc.)
df = get_open_deviation_bars("BTCUSDT", "2024-01-01", "2024-06-30", include_microstructure=True)

# Real-time streaming sidecar
from opendeviationbar import run_sidecar, SidecarConfig
config = SidecarConfig(symbol="BTCUSDT", threshold_decimal_bps=250)
run_sidecar(config)

Designed for Claude Code

This repository uses a CLAUDE.md network that provides comprehensive project context for AI-assisted development via Anthropic's Claude Code CLI.

npm install -g @anthropic-ai/claude-code
cd opendeviationbar-py
claude

Claude Code reads the CLAUDE.md files automatically and understands the full architecture, API, build system, and development workflow.

Development

git clone https://github.com/terrylica/opendeviationbar-py.git
cd opendeviationbar-py
mise install          # Setup tools (Rust, Python, zig)
mise run build        # maturin develop
mise run test         # Rust tests
mise run test-py      # Python tests

Requirements

Runtime: Python >= 3.13, pandas >= 2.0, numpy >= 1.24, polars >= 1.0

Build: Rust toolchain, maturin >= 1.7

License

MIT License. See LICENSE.

Citation

@software{opendeviationbar-py,
  title = {opendeviationbar-py: High-performance open deviation bar construction for quantitative trading},
  author = {Terry Li},
  url = {https://github.com/terrylica/opendeviationbar-py}
}