About this project

TradingView MCP is an open-source Model Context Protocol (MCP) server that exposes market data and technical-analysis capabilities to AI assistants such as Claude, ChatGPT, Cursor and Copilot. It is an independent project and is not affiliated with TradingView Inc. What it provides - Real-time quotes and market snapshots via Yahoo Finance, covering stocks, crypto, ETFs, indices, FX and some regional markets. - Technical analysis tools built on TradingView technical ratings, including RSI, MACD, Bollinger Bands and multi-timeframe reads. - Screeners across global exchanges, including top gainers/losers, volume breakout and smart-volume scanners, rating filters and Bollinger scans. - A backtesting engine with nine strategies (RSI, Bollinger, MACD, EMA cross, Supertrend, Donchian, RSI pullback, Keltner breakout, triple EMA), reporting metrics such as win rate, Sharpe, Calmar, max drawdown, profit factor and expectancy, with commission and slippage simulation. A walk-forward variant adds train/test validation and an overfitting verdict. - Sentiment and news tools drawing on Reddit communities and RSS feeds from Yahoo Finance, MarketWatch, CNBC and crypto outlets, plus a combined analysis tool that merges technicals, sentiment and news. - Interactive candlestick charts rendered inside the conversation through MCP Apps on the hosted version. Deployment and integration - Self-hosting is free under the MIT license; installation is via pip or uv, with configuration examples for Claude Desktop and a Codex plugin. - A hosted option is offered as a single connector URL with paid tiers and a trial; the README states self-hosting remains free. - An OpenClaw integration routes Telegram, WhatsApp and Discord messages to an AI agent that calls the server through a Python wrapper. Notes from the README - The server does not log into or automate a TradingView account and requires no TradingView API key; data is fetched server-side from public endpoints. - An optional Marketaux token enables news and sentiment tools. - The README includes a prominent disclaimer that outputs are not financial advice and that market data may be delayed or incomplete. - Structured error envelopes with stable codes and a retryable flag are used by several tools. The README reports 37 tools in total and documents recent additions such as async hot-path tools, a resilience layer with retry and caching, and a rebuilt financial news service.