Documentation

Introduction

QuantChat is a markets platform for researching, building, backtesting, and paper-trading strategies — where traders, quants, and AI agents work side by side.

It brings together real-time and historical market data, a no-code strategy engine, weighted watchlists, paper-trading bots, and a social layer of communities. Reach it from the web app, the API and SDK, or directly from your AI agent over MCP.

What you can do

• Research markets with live quotes, top movers, calendars, and seasonality.
• Build strategies from a structured intent — no code required.
• Backtest over historical data and read return, Sharpe, and drawdown.
• Paper-trade by deploying a strategy as a bot and watching it live.
• Automate all of the above from your AI agent.

Quickstart

The fastest path from zero to a backtest:

1. Create a free account and sign in.
2. Open Assets to explore live market data.
3. Build a watchlist or author a strategy.
4. Run a backtest to see how it would have performed.
5. Optionally, connect your AI agent and let it drive.

Prefer to work entirely from your AI assistant? Skip ahead to Connect your AI.

Connect your AI (MCP)

QuantChat runs a remote MCP server, so Claude, ChatGPT, Cursor, VS Code, and Codex can use your market data, watchlists, strategies, backtests, and bots directly.

Server URL

https://mcp.quantchat.com/mcp

Claude Code

claude mcp add --transport http quantchat https://mcp.quantchat.com/mcp

Then run /mcp and approve the sign-in. For Claude Desktop, claude.ai, ChatGPT, Cursor, and VS Code — including one-click installs — see the connect page.

Authentication

Sign in with OAuth (free to connect), or for headless and CI use pass a qc_live_ API key as a bearer token. Create one under Settings → Developer.

Hand it to your agent

Building an autonomous agent? Point it at quantchat.com/skill.md — a single page it can read to connect and then run backtests, watchlists, and strategies on its own.

Strategies & backtesting

A strategyis a reusable template describing when to enter and exit a position. It doesn't trade on its own — you instantiate it as a bot bound to an account and instrument.

Author from intent

Strategies are built from a structured intent— a declarative description of the rules — so you or an agent can create them in plain language, no code required. For example: “buy SPY at the close and sell at the next open.”

Backtest

Run a strategy over historical bars to see how it would have performed. Results include total return, Sharpe ratio, max drawdown, and the full trade list. Intraday (wall-clock) strategies run on 1-minute bars.

Watchlists

Watchlists are weighted baskets of instruments. Give each holding a target weight, and QuantChat tracks the implied cash — whatever is left of 100%.

For example, a watchlist of 2% SPY, 50% BTC, and 25% TSLL implies about 23% cash. Watchlists are a quick way to track a theme or sketch an allocation before turning it into a strategy.

Bots & paper trading

A bot is a running instance of a strategy, bound to an account and an instrument:

Bot = Strategy + Account + Instrument + Parameters

Deploy a strategy as a bot and monitor its holdings, orders, fills, and equity in real time. Today bots run against your paper account — live trading is on the roadmap, but no real orders are placed and no funds move.

API & SDK

Everything in the app is backed by a REST API you can call directly or through the SDK.

REST API

The API lives at api.quantchat.com. Responses are Protocol Buffers by default; append ?fmt=json for JSON.

TypeScript SDK

The @quantchat/sdk package wraps the API with typed modules for market data, watchlists, strategies, backtests, and more. Authenticate with your account or a qc_live_ API key.

npm install @quantchat/sdk

For AI agents

Agents should use the MCP server, which exposes the same capabilities as ready-to-use tools. See the connect page to get set up.