New Python quant toolkits

Quant traders are getting a bigger open-source Python toolbox, with new lists, libraries, and backtesting workflows surfacing in public this month. (x.com) (pypi.org) In systematic trading, a backtest is a rehearsal with old market data: a strategy makes paper trades, and the code measures returns, drawdowns, and turnover before real money is used. Quant Science has been publishing Python-first training material around that workflow, including a six-step framework that starts with idea generation and moves through data, testing, and live trading. (hillsdaleinv.com) (quantscience.io) Quant Science said it had compiled a list of more than 343 open Python projects for quantitative finance and algorithmic trading, and separately posted a demo of a four-line open-source “financial advisor.” The posts point readers toward reusable code rather than a single new product release. (x.com 1) (x.com 2) A second project now drawing attention is QF-Lib, a Python package released on the Python Package Index on April 8, 2026 as version 4.0.6. Its documentation says the library includes an event-driven backtester, portfolio-construction tools, indicators, analysis modules, and connectors for data vendors including Bloomberg, Quandl, Haver Analytics, and Portara. (pypi.org) (qf-lib.readthedocs.io) Event-driven means the software reacts to market moments one by one, like the open, the close, or a new trade, instead of treating a day as one lump. QF-Lib says that design lets users add commissions, slippage, and other trading frictions that can make a paper strategy look worse, but more realistic. (qf-lib.readthedocs.io) (github.com) That focus on realism is showing up in higher-speed trading tools too. HftBacktest, another Python framework, says it models feed latency, order latency, and queue position, using full order-book and trade data to replay how a market-making strategy might actually get filled. (hftbacktest.readthedocs.io) The testing methods being discussed alongside these tools are also getting stricter. A 2024 paper by researchers including Marcos Lopez de Prado grouped robust validation into walk-forward tests, resampling methods, and Monte Carlo simulations, and warned that weak testing can produce strategies that fail out of sample. (hillsdaleinv.com) Walk-forward testing means training a strategy on one slice of history, then moving the window ahead and testing it on the next slice, repeating the process across time. Monte Carlo simulation means reshuffling or perturbing assumptions many times to see whether the result survives different paths, costs, and market conditions. (hillsdaleinv.com) (quantconnect.com) The result is a more public quant workflow: curated code lists to find libraries, small demos to show how little code can wire up an idea, and heavier backtesting stacks for traders who need execution and data engineering details. The common thread is that more of the infrastructure that used to sit inside firms is now visible in Python packages and docs anyone can inspect. (x.com) (pypi.org) (hftbacktest.readthedocs.io)