Vectorised backtesting python library

A compact research library for fast, vectorised backtesting and portfolio construction with clear timing and guardrails.

github repo

A small research library I wrote for fast strategy prototyping. Vectorised backtesting is not perfectly realistic, but it is a good way to quickly tell whether an idea looks promising or probably dead on arrival. I use it as a first pass tool before moving to slower event driven sims or production code.

Why I built it

• I wanted a quick feedback loop for research: test many variations without building a full simulator each time
• I wanted consistent assumptions across experiments: timing, execution lag, costs, and cash handling
• I wanted something simple enough to extend while exploring new signals and portfolio rules

What it includes

Fast vectorised backtests

• Single asset and multi asset backtests built on numpy and pandas
• Clear timing alignment between signals, trades, and PnL, with configurable execution lag
• Cash accounting mode that tracks equity, cash, turnover, and net performance after costs

Portfolio construction helpers

• Turn signals into positions or weights with a clean interface
• Optional constrained optimisation via cvxpy for basic weight and risk constraints

Guardrails for research

• Input checks that prevent silent bad results, including enforcing positive prices
• Safe handling of edge cases like divide by zero by producing NaN instead of misleading numbers
• Result objects with consistent field meanings so analysis code stays simple