pyqrt 0.0.15

Quant Research Tools

qrt — Quant Research Tools

One consistent import qrt as q API over the fragmented quant Python ecosystem — market data, technical indicators, return statistics, and interactive plotting, wired together so the output of one is the input of the next.

import qrt as q

aapl = q.data.sources.yfinance.read("AAPL", "2024-01-01", "2025-01-01", "1d")
spy = q.data.sources.yfinance.read("SPY", "2024-01-01", "2025-01-01", "1d")

strategy = aapl["close"].pct_change().rename("AAPL")
benchmark = spy["close"].pct_change().rename("SPY")

q.stats.benchmark_stats(strategy, benchmark)   # alpha, beta, Sharpe, tracking error, ...
q.plot.plot(strategy, benchmark=benchmark)      # interactive equity + drawdown report

Why qrt

• No more juggling five libraries with five conventions. TA-Lib and pandas-ta-classic indicators (q.feat.talib, q.feat.pandas_ta), Yahoo Finance/Binance/DuckDB market data (q.data.sources), and 30+ risk/return metrics inspired by quantstats (q.stats) all speak the same plain pandas DataFrame/Series OHLCV and return-stream layout — chain them freely, no glue code, no format conversion.
• A canonical trades format, not just return streams. One row per round-trip trade (entry/exit price & time, direction, MAE/MFE, free-form feature snapshots) is a first-class citizen: q.stats.trade_stats, q.stats.trades_to_returns, and q.plot.trades/mae_mfe/ trade_distribution all consume it directly.
• Built-in robustness checks, not just a backtest score. Bootstrap Monte Carlo, forward win-rate variance testing, and noise-sensitivity testing ship as first-class q.stats/q.plot functions, not an afterthought — ask "does this edge survive a different order of draws / a worse win rate / noisier data?" in one call.
• Interactive by default. Every chart is a real Plotly figure — zoom, hover, range-select — exportable to standalone HTML or PNG with q.plot.show.
• Works offline. Bundled sample OHLCV data (AAPL, SPY, BTC-USD) and demo strategy trade logs mean you can try every function with zero network calls or API keys.

Library layout

Module	Purpose
q.data	local parquet/csv I/O, market data sources (Yahoo Finance, Binance, DuckDB), bundled sample datasets
q.feat	feature engineering: own indicators + TA-Lib + pandas-ta-classic
q.stats	return-stream & trade-level statistics: performance, alpha/beta, Monte Carlo, variance/noise testing
q.plot	interactive Plotly charts and performance reports, for both return streams and trade logs
q.model	ML model utilities (PyTorch helpers via torchinfo)
q.model.selection	leakage-aware CV splits (walk-forward, purged K-fold, embargo) (planned)
q.bt	event-driven backtesting (planned)
q.portfolio	portfolio analysis (planned)

Warning ⚠️

Still in early alpha — APIs may change without notice. Track progress on the Roadmap.

Install

uv add pyqrt

Docs

Full documentation, tutorials, and API reference: https://quantbert.github.io/qrt/