The Modern Backtesting Toolbox

AI-Powered Development

TFire is optimized to be used with Large Language Models like Claude and ChatGPT to help you:

  • Convert trading ideas into working implementations
  • Generate code for custom indicators
  • Debug issues and optimize performance
  • Understand complex features through interactive Q&A
U
Create a mean reversion strategy that: - Buys when price is 2 std below 20-day MA - Takes profit at mean - Uses position sizing based on volatility
AI
I'll help you implement that strategy. Here's the complete code:
collection = setup_collection(spec) settings = Settings() # Add technical indicators settings[LayerSMA][:period] = 20 settings[LayerBB][:std_mult] = 2 collection = add_layer(collection, LayerBB) # Generate trading signals signal_func = (price, mean, lower) -> price < lower ? 1.0 : price > mean ? -1.0 : 0.0

Lightning-Fast

Built on Julia, TFire delivers C++-level performance with Python-like simplicity. Backtest complex strategies across thousands of assets in seconds, not hours.

Ease of Use

TFire comes with all necessities included. No need to set up some clunky Python framework or to learn some proprietary programming language. It includes a convenient REPL – just like Python.

Flexible

TFire is fully extensible with user-built models and modules – all in the Julia programming language. Whatever you want to add, it is possible.