Blueshift

What it is

Blueshift is a hosted Python environment for systematic strategies: a browser IDE (with a drag-and-drop block editor alongside it for people who would rather not type Python), an event-driven engine with bundled minute data, and a launch button that runs the same code at a supported broker. Nothing to install and no data to source — that is the pitch against a local framework. The API descends from Zipline (initialize, scheduled handlers, a data object, Pipeline), so Quantopian code reads as familiar; Zipline itself was deprecated in Blueshift 2.0 and the current engine is 4.0.0, last changelogged in May 2025.

QuantInsti is a trading-education company and Blueshift is the free end of it. The paid business is the EPAT programme ($7,799–$9,499 a batch, depending on how early you enrol) and white-labelling the same engine to brokers: Spring by ICICI Direct is Blueshift under another name.

Pricing

Free, with no paid tier to move to — the application has no billing or upgrade screen anywhere in it. Research and backtesting are free outright; live deployment is documented as free "for a limited period", with no stated end and no announced price after it.

Data & coverage

The documented backtest datasets are a liquidity-filtered top-1000 universe of US equities and ETFs, ten FX pairs (both from July 2008) and nine coins quoted against USDT and against INR, plus USDT/INR itself, from January 2018 — minute bars, refreshed once a day after the close and so running about a day behind. The US universe is re-ranked quarterly; the crypto set is static, and the docs call it experimental and exposed to survivorship bias.

Indian data is a gap in the documentation rather than in the product. The datasets page lists no NSE feed, yet the platform's own introduction promises minute data across US and Indian equities markets, a docs tutorial runs its backtest on the NSE Minute dataset, the engine's changelog describes an Indian execution model with a cash and a futures segment, and the shipped application still maps NSE-DAILY and NSE-MINUTE bundles and points at an NSE dataset carrying data into August 2026. Read the Indian universe, its history start and its adjustment policy as undocumented, not as absent. Live runs take their data from the broker, not from Blueshift.

Integrations

Two brokers, and only two: Alpaca for US equities, paper accounts included, and Spring by ICICI Direct for NSE cash and F&O. QuantInsti also publishes a developer kit on GitHub under AGPL-3.0 — PEP 561 type stubs for the API, a VS Code snippet extension and LLM prompt guides — first committed in November 2025, last committed on 19 March 2026, with no tagged release. That repository is the entire open-source surface and it contains no engine. The docs have said since 2021 that the core engine is slated to be made open source; it has not been, and the licence on the helper repo says nothing about the platform you are actually running on.

Limitations

• Imports are restricted to a whitelist of 21 packages, your own code must be pure Python (no C extensions), and you cannot upload your own dataset.
• No debugger, and introspection such as type and dir is disabled for security. The documented workaround is print.
• A live run needs an end date no later than the coming Friday's close, and cannot be launched on a weekend at all.
• The F&O side is undocumented. The application ships an option wizard that backtests NSE option structures, but no options or futures dataset appears on the datasets page, so there is no published statement of what history it runs on.
• Closed-source engine on QuantInsti's servers; nothing here is self-hostable, and the AGPL repository is a helper kit, not the platform.

Alternatives

Zipline-reloaded is the same lineage run locally, with data you supply. QuantConnect is the closer comparison — also hosted and broker-connected, but with paid tiers and far wider data. Backtrader or VectorBT if a notebook and a CSV were all you wanted.