backster

Overview Link to heading

Backster is a Rust command-line research bench for experimenting with trading ideas against time-series market data. It focuses on repeatable, config-driven runs rather than notebook-centric workflows: you point the binary at a TOML file, Backster loads a price series, runs a strategy or optimizer, and reports return statistics.

Ambition Link to heading

Provide the speed of Rust-based Polars for historical simulations while retaining the symbolic power of Wolfram for strategy modeling and research.

What’s novel Link to heading

  • Integrated WSTP bindings for seamless data transfer with Mathematica.
  • High-performance backtesting using the Polars query engine.
  • Flexible configuration system for complex simulation parameters.

Highlights Link to heading

  • backtesting long/short strategies over daily close data
  • fitting rolling return distributions and turning them into trade signals
  • running repeatable multi-run experiments for stochastic strategies
  • fetching and caching market data through Wolfram / WSTP
  • prototyping pipeline-style signal generation in TOML instead of hard-coding every idea in Rust

Stats Link to heading

  • Project page: /projects/backster/
  • Primary language: Rust
  • Commits: 31
  • Created: 2026-04-23T19:08:55Z
  • Last updated: 2026-05-03T21:50:20Z