# Pancake vs backtrader

Canonical: https://www.usepancake.com/compare/pancake-vs-backtrader

backtrader is a long-standing event-driven Python backtesting framework for equities, futures, and forex. Pancake is a hosted, agent-native backtester for prediction markets with verified results at permanent URLs.

## Capabilities

| Capability | Pancake | backtrader |
|---|---|---|
| Open-source engine | ✓ Apache-2.0 (batter, Python 3.12+) | ✓ GPL-3.0 (Python) |
| Event-driven multi-asset framework | ✗ prediction-market binary outcomes only | ✓ equities, futures, forex; broker integrations (Interactive Brokers, Oanda) |
| Actively developed | ✓ | △ original development has slowed; community forks carry maintenance |
| Verification boundary doctrine | ✓ explicit 3-tuple in every result | ✗ no structured epistemic scope statement |
| Agent-callable MCP surface | ✓ 10-tool surface | ✗ Python library, no MCP integration |
| Shareable verified result URLs | ✓ /<handle>/<strategy_slug>/v/<version_n> with SHA-256 result hash | ✗ results are local objects and plots |
| Deterministic reproducibility | ✓ byte-stable result hash (canonical JSON, PCG64, Python 3.12+) | ✗ depends on local environment and library versions |

## What's different

backtrader is one of the most widely taught Python backtesting frameworks. Its event-driven Cerebro engine processes bars through user-written Strategy classes with indicators, analyzers, and broker simulation, and a decade of tutorials and Stack Overflow answers make it the default starting point for learning algorithmic backtesting in Python.

Its model assumes a human researcher at a workstation: you own the data feed, the environment, and the interpretation of the output. Reproducing someone else's backtrader result means obtaining their data file, their parameter values, and their library versions — in practice, results are rarely reproduced at all. Original development has also slowed, with maintenance carried by community forks.

Pancake inverts the model for the agent era. The strategy spec is a declared, hashed document rather than arbitrary Python; the evidence dataset is validated and content-hashed; the batter engine re-derives every number; and the output is a verified result at a permanent URL that any agent or human can audit. What backtrader leaves to convention — data provenance, environment pinning, honest caveats — Pancake encodes as structured fields in the result.

## Methodology overlap

Both apply per-trade commission and slippage models and report Sharpe, drawdown, and trade-level statistics (backtrader via its analyzers module). Pancake formalizes the statistics layer with citable formulas, Wilson CI95, bootstrap CI, permutation testing, and small-sample suppression, and pins engine identity in every result.

## When to use Pancake

Use Pancake when the strategy targets prediction markets and the result needs to be independently verifiable — especially when an LLM agent authors the strategy and a permanent, citable URL is the deliverable. Pancake is the right tool when trust in the number matters more than framework flexibility.

## When to use backtrader

Use backtrader when you are backtesting equities, futures, or forex in a local Python environment and value a mature event-driven framework with broad educational material. backtrader is the right tool for hands-on framework users comfortable owning their full stack.

## Citation

backtrader is an event-driven Python framework for backtesting and trading. Homepage: https://www.backtrader.com

---

Markdown twin of https://www.usepancake.com/compare/pancake-vs-backtrader — same content as the HTML page, generated from the same source data. More machine surfaces: https://www.usepancake.com/llms.txt