Hawk-Backtester
/ EN
Sign up Sign in
Hawk-Backtester › Intro

Hawk-Backtester — AI Agent Trading Lab for algorithmic strategy research.

Try for free
  • Hawk-Backtester is a web-based platform for algorithmic trading research and experimentation.
  • It integrates the user's local environment as the execution platform, enabling testing of any algorithm including those using external APIs, GPUs, and other computational resources.
  • User-created algorithms can be tested and results viewed without the code being sent to any external server, including the system's own servers, ensuring privacy.
  • Historical data for testing must, in principle, be uploaded by the user, but a synthetic data generation feature based on statistical models is also available.
  • With AI Agent Mode, the user's AI Agent can autonomously run the loop of algorithm development, test execution, result evaluation, and improvement.
Setup
Setup
Result — Time Series ex.1
Result Time Series ex.1
Result — Time Series ex.2
Result Time Series ex.2
Result — Scatter Analysis ex.
Result Scatter Analysis
Result — Exit Analysis
Result Exit Analysis
Setup
Setup
Result — Time Series ex.1
Result Time Series ex.1
Result — Time Series ex.2
Result Time Series ex.2
Result — Scatter Analysis ex.
Result Scatter Analysis
Result — Exit Analysis
Result Exit Analysis

Why Hawk-Backtester ?»

No Restrictions on Algorithms

Freely use external APIs, GPUs, and any libraries.
No execution restrictions commonly found in cloud services.

Research Code Privacy Guaranteed

Algorithms written locally are executed locally, ensuring confidentiality through a design that never sends code to the server.

Autonomous Algorithm Improvement by AI Agent

Provides an environment where AI autonomously iterates on algorithm development, test execution, result evaluation, and improvement through the AI Agent API.
* The AI Agent is provided and contracted by you; it is not included in this service.

High-Speed Execution in Local Environment

The WASM + Rust + WebSocket tech stack enables near-native-speed simulation execution in the browser.

Debug-Level Detailed Performance Analysis

With per-order ticket management, you can deeply track the causes of P&L through time-synchronized timeline, exposure change, and contribution analysis.
(See documentation for analysis details)

Easy to Use via Web Browser

Get started simply by accessing from your browser.
No local installation required; provides a SaaS-equivalent user experience.No local installation required, providing a SaaS-equivalent user experience.

Architecture»

Local

Browser

Your Algorithm Code
hawk-btPython library
WebSocket
Market SimulatorCapital calc, order execution, etc.
Scope of This Service

Quick Start»

  1. 1
    Create Account(Trial plan is free)
  2. 2
    Install hawk-bt
    pip install hawk-bt

    * hawk-backtester on PyPI is a different package

  3. 3
    Upload historical data in the browser, or generate synthetic data
  4. 4
    Write and execute algorithms locally using hawk-bt
    (See the hawk-bt documentation for usage details)
  5. 5
    Press the Connect to Local Strategy button in the browser to start the simulation

Recommended Environment»

The simulation engine of this service runs on WebAssembly (WASM). The following environment is recommended for optimal performance.

Recommended Browser

  • Google Chrome (latest version) — Optimal WASM execution performance via the V8 engine
  • Chromium-based browsers (Edge, Brave, etc.) are also supported

Chrome Performance Settings

Enabling the following settings improves WASM execution speed.

  • WebAssembly Baseline Compiler — Set chrome://flags/#enable-webassembly-baseline to Enabled
  • WebAssembly Tiering — Set chrome://flags/#enable-webassembly-tiering to Enabled
  • Hardware Acceleration — Enable "Use hardware acceleration when available" at chrome://settings/system

For feature requests and improvement suggestions, please feel free to submit them via the Feedback page in the header after logging in.