← Back to the library
Trading and investing

Strategy backtesting

Execute a reproducible trading-strategy backtest using the selected available engine: manifoldbt, vectorbt, or event-driven Backtrader. Requires real historical bars, explicit rules, costs, and an executable host. Returns measured results and assumptions; backtest-expert owns independent result scrutiny. Never substitutes estimated metrics for an actual run.

Works with the context you provideVersion 1.0.0

manifoldbt Backtester

First read and apply `references/chiefos-runtime.md`. Choose the source method below that matches the requested outcome and available evidence; do not execute every alternate provider path.

tradermonty: manifoldbt-backtester

Load `references/tradermonty/manifoldbt-backtester/SKILL.md` for the complete method, required inputs, ordered decisions, and output contract. Resolve its local resources beneath the same directory.

agiprolabs: backtrader

Load `references/agiprolabs/backtrader/SKILL.md` for the complete method, required inputs, ordered decisions, and output contract. Resolve its local resources beneath the same directory.

agiprolabs: vectorbt

Load `references/agiprolabs/vectorbt/SKILL.md` for the complete method, required inputs, ordered decisions, and output contract. Resolve its local resources beneath the same directory.

Alternate source routing

Additional methods: backtrader, vectorbt. Use the applicable source for its provider and output, retaining the same canonical ChiefOS identity. For a source-specific request, load that source method before answering.

For upstream sibling names, resolve the canonical skill and source entrypoint with `references/source-routing.json`. Load sibling resources from that canonical skill; do not assume the original directory name is independently installed.