Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

StratCraft

Build C++ Build

The AI-native Quant Trading Platform. Build, backtest, and analyze trading strategies with a high-performance C++ executor and modern desktop UI.

Features

  • Strategy Builder - AI-assisted strategy generation with multiple modes (Regime Detector, Indicator Entry, and more)
  • C++ Backtest Executor - High-performance native backtesting engine with zero-copy Apache Arrow Parquet data
  • Plugin Architecture - Extensible system with official and community plugins via Marketplace
  • Free Data Providers - YFinance and Dukascopy integration for market data download
  • Result Visualization - Interactive charts, equity curves, trade tables, and performance metrics
  • Cross-Platform - Windows, macOS, and Linux

Quick Start

Prerequisites

  • Node.js 20+
  • pnpm 9+
  • CMake 3.20+ (for C++ executor)
  • C++23 compiler (GCC 13+, Clang 16+, or MSVC 2022+)

Optional:

  • ClickHouse (for advanced data providers; free users use YFinance/Dukascopy)
  • vcpkg (auto-detected; downloaded automatically if not present)

Installation

git clone https://github.com/StratCraftsAI/StratCraft.git
cd StratCraft

# Install dependencies
pnpm install

# Run development mode
pnpm dev:desktop

Package for Distribution

cd apps/desktop
pnpm package:win    # Windows
pnpm package:mac    # macOS
pnpm package:linux  # Linux

Architecture (V3)

Renderer (React UI)
    | IPC
Main Process (Electron)
    | spawn + file I/O
stratforge-runner (pure C++23)

Key design: Strategies are compiled C++ .so libraries (ABI v2 factory pattern). The executor (stratforge-runner) compiles strategy code, loads the shared library, reads Parquet data via Apache Arrow, runs the backtest, and writes result.json. No gRPC, no Extension Host, no shared memory.

Project Structure

StratCraft/
  apps/desktop/            # Electron desktop application
    src/
      main/                # Electron main process
      renderer/            # React UI (StratCraftsAI design)
      preload/             # IPC bridge
      shared/              # Types, constants, utils
  packages/
    executor/              # C++23 backtest executor (stratforge-runner)
    builder-templates/     # Strategy code generation and validation tools
    bridge/                # IPC bridge package
    types/                 # Shared TypeScript types
    plugin-verifier/       # Plugin signature verification
    sdk/                   # Plugin SDK
  plugins/
    data-plugin/           # Tier 0: shared data components
    strategy-builder-nexus/  # Tier 1: Strategy Builder
    back-test-nexus/       # Tier 1: Backtest execution

How Strategy Generation Works

StratCraft uses a server-assisted strategy generation model:

  1. You configure your strategy parameters in the desktop app (open-source)
  2. Parameters are sent to our generation server
  3. The server applies advanced prompt engineering and code templates
  4. If using BYOK (Bring Your Own Key), your API key is used only for the LLM call and is never stored on our servers
  5. Generated C++ strategy code is returned to your desktop app
  6. The strategy is compiled to a shared library (.so/.dll) and backtested locally by the C++ executor

Why not call the LLM directly? The generation quality comes from our curated prompt templates and multi-step pipelines, not from a simple LLM call. The server adds significant value beyond proxying API requests.

Your API Key is safe:

  • Transmitted over HTTPS only
  • Used per-request, never persisted
  • Not included in server logs

Plugins

Extend StratCraft with plugins from the built-in Marketplace:

Plugin Type Description
Strategy Builder Bundled AI-assisted strategy code generation
Backtest Bundled Strategy backtesting and result visualization
Data Bundled Multi-source data provider
Quant Lab Marketplace Alpha Factory, Signal Factory, Factor Library

See Plugin Development Guide for creating your own plugins. Use the Plugin Template to scaffold a new plugin project.

Tech Stack

Component Technology
Desktop Electron 28, TypeScript, React
Executor C++23, Apache Arrow, nonabackTrader
Strategies C++23 (compiled .so via ABI v2 factory)
Build pnpm, Turborepo, CMake, vcpkg
UI Tailwind CSS, StratCraftsAI Design

Contributing

We welcome contributions! Please see our Contributing Guide for details.

Security

See Security Policy for vulnerability reporting and API key handling details.

License

Licensed under the Apache License, Version 2.0. See LICENSE.

Links

About

AI-native quant trading platform — AI strategy builder, C++ backtest executor, plugin marketplace. Windows/macOS/Linux.

Resources

Code of conduct

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages