vectorbt

vectorbt

Vectorized backtesting library that packs thousands of strategy configurations into NumPy arrays for parallel execution. Apache-2.0 with Commons Clause (source-available, not open source).

vectorbt is a vectorized backtesting library: it packs thousands of strategy configurations into NumPy arrays and accelerates them with Numba (and an optional Rust engine). Its docs say this makes it possible to 'test many thousands of strategies in seconds.'

Good for

• Rapid prototyping of trading strategies
• Parameter optimization across large grids
• Portfolio backtesting with rebalancing

Verdict

Powerful for research and backtesting at scale. Licensed Apache 2.0 with Commons Clause: you can use, read, modify and redistribute the code, including inside a company. The license withholds one right: you may not "Sell" vectorbt, meaning offer a paid product or service (including hosting, consulting or support) "whose value derives, entirely or substantially" from it. That restriction makes it source-available ("fair-code"), not OSI open source. VectorBT PRO (`vectorbtpro`) is the author's separate paid successor. Membership starts at $25/mo ($240 for 12 months, $500 lifetime) and gives access to a private repo with extra features: limit orders, leverage, parallelization and cross-validation tools.