Skip to content

AlphaForge Documentation

AlphaForge is a local CLI for quantitative research that turns JSON-defined strategies into Pine Script v6 you can run on TradingView right away. Optuna TPE optimization plus walk-forward validation guards against overfitting, while strategy data, trade history, and API keys all stay on your own machine — from research to live execution (via TradingView) in a single pipeline.

This documentation walks through installation, strategy development, and integration with AI coding agents.

Using AlphaForge with an AI agent

Everything about driving AlphaForge from Claude Code / Codex / Cursor now lives in the AI Agent Integration section, which covers all three paths: bundled skills, the MCP server, and the non-interactive CLI. Start with the Agent Quickstart to try agent integration in about 10 minutes.

Two things that set AlphaForge apart

1. JSON strategy → one command → Pine Script v6 → live on TradingView

Write your strategy in JSON, then run alpha-forge pine generate to compile it into TradingView Pine Script v6. Unlike all-in-one Web UI platforms that lock you into a specific server and exchange adapters, AlphaForge lets you reuse your existing TradingView setup for alerts, automation, and chart visualization.

JSON strategy → Backtest + Optimize → Pine Script v6 → TradingView horizontal pipeline

2. Optuna TPE + walk-forward validation to fight overfitting

alpha-forge optimize run triggers Optuna Bayesian optimization (TPE) out of the box, and --split adds walk-forward analysis (WFT) so you can compare in-sample vs out-of-sample performance in one step. This is the optimization and generalization layer that Web-UI-centric platforms typically lack.

Parameter ranges → Optuna TPE optimization → IS/OOS/WFT delta → Optimized strategy branching flow

How AlphaForge compares to other quant tools

Dimension AlphaForge All-in-one Web UI platforms Frameworks (vectorbt / Backtrader / …)
Strategy authoring JSON DSL (easy to version-control) Python classes inside a UI editor Python classes
TradingView integration Auto-generates Pine Script v6 Typically none Typically none
Built-in optimization Optuna TPE + WFT as standard Weak or manual Possible by adding libraries
Data & keys location Fully local Server-resident Local
Runtime form Binary CLI (a single file, a few hundred MB) Docker / SaaS stack Python scripts
AI agent integration Bundled skills + MCP server Partial Roll your own
Live execution path Through TradingView (broker-neutral) Direct exchange / broker Roll your own

Who this is for

  • TradingView users who want trustworthy Pine Script backed by rigorous backtests and optimization
  • Engineers and quant researchers looking for an alternative to Backtrader, vectorbt, and similar frameworks
  • Developers who want to version-control strategies as code (JSON)
  • Users who want to combine Claude Code, Codex, and other coding agents with AlphaForge to autonomously explore and optimize strategies
  • People who want Bayesian optimization and walk-forward validation in a single command

Topics

  • Getting Started — Installation, license activation, and your first backtest (10-minute Free-plan walkthrough included)
  • Use Cases by Goal — Pick the most relevant next page based on your role (TradingView user / Python developer / Quant / Auto-trading / AI agent user)
  • CLI Reference — Every alpha-forge command with parameters and output examples
  • Strategy Templates — Combination strategies like HMM × BB × RSI with concrete JSON
  • AI-Driven Strategy Exploration Workflow — A how-to for autonomous strategy development with Claude Code / Codex × AlphaForge (AI Agent Integration section)
  • Legal & Disclaimers — Disclaimers, EULA, and Privacy Policy

Privacy — no telemetry

AlphaForge sends no telemetry — no usage statistics or behavioral logs — to any external party. Your strategy data, trading history, and API keys all stay local. The Software accesses the network only for user-initiated operations (data fetching / update checks via self version & self update / Whop license verification). See the Privacy and Telemetry section of the EULA for details.