Python backtesting frameworks, compared.
backtesting.py, Backtrader, and vectorbt cover most Python backtesting needs, and a managed platform is a different category entirely. Here's how they line up, side by side.
Frameworks and a platform, compared.
| Periscøpe | backtesting.py | Backtrader | vectorbt | |
|---|---|---|---|---|
| Type | Managed platform | Library | Library | Library |
| Approach | Event-driven | Event-driven | Event-driven | Vectorized |
| Backtesting | ✓ | ✓ | ✓ | ✓ |
| Paper trading | ✓ | — | ~ | — |
| Live trading | Same code, on request | — | ~ | — |
| Managed compute | ✓ | — | — | — |
| Built-in diagnostics | ✓ | — | — | — |
| Actively maintained | ✓ | ✓ | — | ✓ |
Library you build on, or platform that runs it.
The frameworks are excellent at what they do: import them, write your strategy, and get a backtest. The work they leave to you is everything after the backtest, paper and live execution, broker integration, logging, dashboards, and the compute to run it all.
Periscøpe includes an event-driven backtester, but it's a platform, not a library: the same code runs in paper and live, on managed compute, with diagnostics built in. Choose a framework when you want a research backtest you control; choose a platform when you want a running system.
Python backtesting frameworks: questions.
Are these the same kind of thing as Periscøpe? +
Not quite. backtesting.py, Backtrader, and vectorbt are Python libraries you import and assemble into your own system. Periscøpe is a managed platform that includes backtesting but also paper and live execution, managed compute, and diagnostics. The honest comparison is 'library you build on' vs 'platform that runs it for you.'
Which framework is best for pure backtesting? +
It depends on your goal. backtesting.py is the simplest for quick single-asset tests; Backtrader is the most flexible event-driven option (though largely unmaintained); vectorbt is fastest for large vectorized research and sweeps.
When would I choose a platform over a framework? +
When the goal is a running system, not just a research backtest. If you want paper and live on the same code, managed compute, and diagnostics without building them, a platform saves the assembly work.
When you want more than a backtest.
We onboard in focused cohorts. Tell us what you trade and how you work. Live access is rolled out to select users on request.