VectorBT PRO

What it is

VectorBT PRO is the paid continuation of vectorbt by the same author, distributed as a private GitHub repository you are added to as a collaborator rather than as a package on PyPI. The model is unchanged — strategies as multidimensional arrays, thousands of parameter combinations simulated in one call — and v2026.9.5 adds native Rust simulators, streaming indicators that update on each incoming bar, and portfolio continuation, which resumes a simulation from a saved state instead of replaying it. Python 3.11 to 3.14.

It is also the backtesting engine here that is built for coding agents. The vendor lists a "Built-in MCP server and CLI commands, function calling, LLM-friendly documentation, and agent support", and the Typer-based CLI lets you chat with, search and invoke those MCP tools from a terminal.

Pricing

$25 a month month-to-month, $240 for twelve months paid in one go (list $300, struck through on the page, roughly $20 a month), or $500 once for lifetime access. There is no trial; the vendor's answer is that the open-source package is the trial. Prices are described as a limited-time offer for early contributors that will rise with each milestone.

Data & coverage

PRO ships no data of its own, only connectors. Named classes cover TradingView, Polygon.io, Alpha Vantage, Nasdaq Data Link, Databento, Alpaca, Binance, Bybit, Bitfinex and KuCoin, plus findatapy for Bloomberg, Eikon, Quandl and Dukascopy. Local stores are CSV, HDF5, Parquet, DuckDB and any SQLAlchemy database, with savers that append on a schedule. What you can backtest is whatever your own data subscriptions already cover. A utils.telegram module wraps python-telegram-bot and pairs with that scheduler, so an unattended run can push a message to a chat; there is no rule builder, and the condition is code you write.

Limitations

• The licence bars commercial use, and this is the sharp difference from the free package. It defines Private Use as use "where no direct or indirect monetary compensation, commercial advantage, or for-profit purpose is obtained", and allows a for-profit organisation only "internal, in-house usage for purposes such as evaluation, data analysis, prototyping, or research and development (R&D), provided it does not result in a commercial product, service, or outcome". Trading a firm's capital is an outcome. The free vectorbt's Commons Clause forbids only selling to third parties and does not reach that far, so on the question a commercial reader actually asks, the free package is the permissive one.
• No live trading and no paper account, at any price. Nothing routes an order.
• Nothing is public: no PyPI package, no readable source before purchase, no issue tracker, and full release notes are members-only. You buy on the marketing pages and the Discord.
• A vectorised simulator still fills at bar prices, so intrabar sequencing and queue position remain approximations.

Alternatives

The free vectorbt is the same engine one generation back, and it has shipped its own optional Rust backtesting backend since v1.0.0 in April 2026, so a compiled core is not what the money buys — PRO's native Rust simulators, which step bar by bar with callbacks, are. The licence comparison above is the reason to weigh the free package seriously rather than as a demo. NautilusTrader is the open-source choice when the same code must place real orders, QuantConnect rents a research environment and data rather than selling a library, and Backtrader is the slow, readable, event-driven baseline.

Compare it against the rest of the backtesting frameworks.