factorium 0.4.0

A quantitative factor analysis library for financial research

Factorium

Factorium is a Polars-first factor research & backtesting toolkit.

• Data pipeline: BinanceDataLoader + AggBar → multi-symbol OHLCV panel in one line.
• Factor engine: Factor with rich TS/CS/math ops and expression parsing.
• Analysis: FactorAnalyzer + FactorAnalysisResult for IC, quantile returns, and plots.
• Backtest: VectorizedBacktester (exposed as factorium.backtest.Backtester) with Polars-vectorized PnL.
• Research workflow: ResearchSession + FactorReport to tie everything into a notebook-friendly API.

For a Chinese introduction, see README_zh.md.

---

Installation

# Recommended
uv add factorium

# Or with pip
pip install factorium

Development setup:

git clone https://github.com/novis10813/factorium.git
cd factorium
uv sync --dev

---

Quick example

from factorium import ResearchSession

session = ResearchSession.from_parquet("data/btc_1h.parquet")

close = session.factor("close")
momentum = (close.ts_delta(20) / close.ts_shift(20)).cs_rank()

print(session.quick_report(momentum, periods=1))

More complete guides live under docs/:

• docs/getting-started/quickstart.md
• docs/user-guide/bar.md
• docs/user-guide/factor.md
• docs/user-guide/analyzer.md
• docs/user-guide/backtest.md

---

License

MIT – see LICENSE.