pyqrt 0.0.3

Quant Research Tools

Quant Research Tools (qrt)

QRT is an umbrella library of quantitative research tools

Install:

uv add qrt

Use:

import qrt as q

# leakage-safe splits
splits = q.splits.walk_forward(X, n_splits=5, embargo="5D")

# features
X["sma_20"] = q.feat.qta.sma(prices, 20)

# backtest + tearsheet
result = q.bt.run(signal, prices)
q.tearsheet.report(result.returns, benchmark="SPY")

Features (planned / in progress)

• Model wrappers — thin, opinionated wrappers around PyTorch models for training, checkpointing and inference on financial time series (q.models)
• Data splitting — leakage-aware splits: walk-forward, purged K-fold and combinatorial purged CV with embargo (à la López de Prado) (q.splits)
• Tearsheets — performance reports for return streams: Sharpe/Sortino/Calmar, drawdowns, rolling stats, monthly heatmaps, benchmark comparison (q.tearsheet)
• Portfolio analysis — attribution, exposure, turnover and risk decomposition (q.portfolio)
• Backtesting — event-driven backtesting of model signals, connected to the master securities database (DuckDB) (q.bt)
• Feature engineering — feature submodules under one namespace: hand-rolled primitives (q.feat.qta.sma(), q.feat.qta.lags()), all TA-Lib indicators (q.feat.talib.RSI(ohlc)) and all pandas-ta-classic indicators (q.feat.pandas_ta.bbands(ohlc)) with a pandas-friendly interface (q.feat)

Libraries used

Notable libraries qrt is built on and/or wraps:

Library	Used for	Docs
pandas	DataFrames/Series as the common data format throughout	docs
TA-Lib	technical indicators, wrapped in q.feat.talib	docs
pandas-ta-classic	technical indicators & candlestick patterns, wrapped in q.feat.pandas_ta	docs
PyTorch	model training and inference (q.models)	docs
DuckDB	In process database (q.data, q.bt)	docs
yfinance	Yahoo Finance market data (q.vendors)	docs
matplotlib	plotting in tearsheets and reports	docs

Project layout

qrt/
├── qrt/
│   ├── models/      # pytorch wrappers
│   ├── splits/      # CV, embargo, purging
│   ├── tearsheet/   # performance reports
│   ├── portfolio/   # portfolio analysis
│   ├── bt/          # backtesting engine
│   ├── feat/        # feature engineering
│   └── data/        # DuckDB securities-master access
├── tests/
├── examples/        # notebooks
└── pyproject.toml