tradingview-assistant — AI Trading Intelligence MCP Server

Author: DEBARPAN BOSE CHOWDHURY
Repository: https://github.com/DEBARPAN2000/tradingview-mcp
PyPI: tradingview-assistant

The most complete open-source AI trading toolkit for Claude and MCP-compatible clients.
40+ tools · Backtesting · Walk-Forward Validation · Live Sentiment · Yahoo Finance · Paper Trading — no API keys required.

---

Table of Contents

• Why tradingview-assistant?
• Architecture
• Quick Start
• MCP Setup
• All Tools
• Example Prompts
• Docker
• OpenClaw (Telegram / WhatsApp)
• Troubleshooting
• Project Structure
• Contributing
• License

---

Why tradingview-assistant?

Feature	This Server	Typical DIY Setup	Bloomberg Terminal
Setup Time	2 minutes	Hours	Weeks
Cost	Free & Open Source	Variable	~$30k/year
API Keys Required	None	Multiple	N/A
Backtesting	6 strategies + walk-forward	Manual scripting	Proprietary
Live Sentiment	Reddit + RSS news	Separate service	Terminal only
Market Coverage	Crypto + 12 stock markets	Limited	Broad
Multi-Agent Analysis	3-agent AI debate	Not available	Not available
Paper Trading	Built-in SQLite persistence	Not available	Not available

---

Architecture

┌─────────────────────────────────────────────────────────────────┐
│               Claude Desktop / MCP-Compatible Client            │
└──────────────────────────┬──────────────────────────────────────┘
                           │  MCP Protocol (stdio or HTTP)
┌──────────────────────────▼──────────────────────────────────────┐
│               tradingview-assistant MCP Server                  │
│                       (server.py)                               │
│                                                                 │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────────────┐  │
│  │   Screener   │  │   Backtest   │  │  Sentiment & News    │  │
│  │   Service    │  │   Service    │  │      Service         │  │
│  └──────┬───────┘  └──────┬───────┘  └──────────┬───────────┘  │
│         │                 │                      │              │
│  ┌──────▼───────┐  ┌──────▼───────┐  ┌──────────▼───────────┐  │
│  │  TradingView │  │ Yahoo Finance│  │  Reddit API + RSS    │  │
│  │  Screener    │  │     API      │  │       Feeds          │  │
│  └──────────────┘  └──────────────┘  └──────────────────────┘  │
│                                                                 │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────────────┐  │
│  │  Portfolio   │  │  Risk &      │  │   Market Adapters    │  │
│  │  Service     │  │  Correlation │  │  (EGX, NSE, BIST…)  │  │
│  │  (SQLite)    │  │  Tools       │  │                      │  │
│  └──────────────┘  └──────────────┘  └──────────────────────┘  │
└─────────────────────────────────────────────────────────────────┘

Layer overview

Layer	Location	Responsibility
Transport	server.py	MCP tool registry, input validation, response shaping
Services	core/services/*.py	Business logic — screener, backtest, sentiment, EGX, NSE, risk
Adapters	core/services/market_adapters/	Market-specific logic (EGX, NSE)
Static Data	core/data/	EGX and NSE sector / index datasets
Utils	core/utils/	Input validators, error contracts
Portfolio	core/portfolio.py	SQLite-backed paper trading persistence

---

Quick Start

Option 1 — PyPI install (recommended)

pip install tradingview-assistant

Option 2 — uvx (zero-install, always latest)

uvx --from tradingview-assistant tradingview-mcp

Option 3 — Source

git clone https://github.com/DEBARPAN2000/tradingview-mcp.git
cd tradingview-mcp
uv sync
uv run tradingview-mcp

For detailed instructions (Docker, virtual envs, Claude Desktop, troubleshooting) see INSTALLATION.md.

---

MCP Setup

Claude Desktop (claude_desktop_config.json)

macOS / Linux:

{
  "mcpServers": {
    "tradingview": {
      "command": "/Users/YOUR_USERNAME/.local/bin/uvx",
      "args": ["--from", "tradingview-assistant", "tradingview-mcp"]
    }
  }
}

Windows:

{
  "mcpServers": {
    "tradingview": {
      "command": "uvx",
      "args": ["--python", "3.13", "--from", "tradingview-assistant", "tradingview-mcp"]
    }
  }
}

macOS note: GUI apps may not have ~/.local/bin in PATH. Use the full absolute path to uvx.
Windows note: Pin --python 3.13 to use prebuilt wheels and avoid the 60-second initialization timeout.

HTTP / Streamable-HTTP transport

tradingview-mcp streamable-http --host 0.0.0.0 --port 8000

Metrics endpoints:

Endpoint	Format	Description
GET /metrics/rate_limit	JSON	Rate limiting stats
GET /metrics/observability	JSON	Tool usage and latency
GET /metrics/prometheus	Text	Prometheus scrape target

---

All Tools

The server currently exposes 69 MCP tools across 13 categories. For the full parameter reference with types, defaults, and example prompts for every tool, see TOOLS.md.

Screener Tools

Tool	Description
top_gainers	Top gaining assets on any exchange, ranked by % change
top_losers	Top losing assets on any exchange, ranked by % change
bollinger_scan	Bollinger Band squeeze scanner — finds low-volatility setups before breakouts
rating_filter	Filter assets by TradingView BB rating (-3 Strong Sell to +3 Strong Buy)

Asset Analysis

Tool	Description
coin_analysis	Full TradingView indicator suite for a single asset (oscillators, MAs, sentiment)
multi_agent_analysis	Three-agent AI debate (Technical + Sentiment + Risk) with a final confidence-scored decision
multi_timeframe_analysis	Trend alignment across five timeframes simultaneously (Weekly through 15m)
combined_analysis	Flagship tool — technical + Reddit sentiment + news in one confluence-scored response

Candle Pattern Tools

Tool	Description
consecutive_candles_scan	Scans for assets with N consecutive growing (bullish) or shrinking (bearish) candles
advanced_candle_pattern	Multi-timeframe candle acceleration analysis — detects momentum building

Volume Scanner Tools

Tool	Description
volume_breakout_scanner	Detects simultaneous price breakouts and volume spikes
volume_confirmation_analysis	Checks whether current volume confirms or contradicts price movement for a symbol
smart_volume_scanner	Volume anomaly scanner with RSI market-condition filter

EGX (Egyptian Exchange) Tools

Tool	Description
egx_market_overview	Full EGX snapshot — top gainers/losers, volume leaders, sector summary
egx_sector_scan	Scan stocks in a specific EGX sector (leave blank to list all sectors)
egx_sector_scanner	Sector rotation scanner — ranks hottest/coldest EGX sectors with top stock picks
egx_index_analysis	Constituent analysis for EGX30, EGX70, EGX100, SHARIAH33, EGX35LV, TAMAYUZ
egx_stock_screener	Composite-scored stock ranking engine for the full EGX universe
egx_trade_plan	Complete trade plan for an EGX stock — entry, stop-loss, target, R:R
egx_fibonacci_retracement	Fibonacci retracement levels using swing high/low for EGX stocks

NSE (India) Tools

Tool	Description
nse_sector_list	Lists all NSE sectors
nse_market_overview	High-level NSE overview — sectors, Nifty indices, trading hours
nse_sector_scanner	Sector rotation scanner for NSE with top stock picks per hot sector
nse_stock_screener	Composite-scored stock ranking engine for NSE — supports NIFTY50/BANK/IT index filters
nse_trade_plan	Complete trade plan for an NSE stock — entry, stop-loss, target, R:R
nse_index_analysis	Constituent analysis for Nifty 50, Nifty Bank, Nifty IT, Nifty 100
nse_fibonacci_retracement	Fibonacci retracement levels for NSE stocks

BSE (India) Tools

Tool	Description
bse_market_overview	Full BSE snapshot — top stocks per category and sector summary
bse_sector_scanner	Sector rotation scanner for BSE with top stock picks per hot sector
bse_stock_screener	Composite-scored stock ranking engine for BSE — supports SENSEX/BSE100 filters
bse_trade_plan	Complete trade plan for a BSE stock — entry, stop-loss, target, R:R
bse_index_analysis	Constituent analysis for SENSEX, BSE100, and other BSE indices
bse_fibonacci_retracement	Fibonacci retracement levels for BSE stocks

Sentiment & News Tools

Tool	Description
market_sentiment	Real-time Reddit sentiment score for any asset (crypto, stocks)
financial_news	Aggregated financial news from Reuters, CoinDesk, FT, and other RSS feeds

Backtest Tools

Tool	Description
backtest_strategy	Single-strategy backtest on Yahoo Finance data — Sharpe, Max DD, Win Rate, equity curve
compare_strategies	Runs all 6 strategies simultaneously and returns a ranked leaderboard
walk_forward_backtest_strategy	Walk-forward validation with N folds to detect overfitting

Supported strategies: rsi, bollinger, macd, ema_cross, supertrend, donchian

Yahoo Finance Tools

Tool	Description
yahoo_price	Real-time price for any Yahoo Finance symbol (stocks, crypto, FX, indices, ETFs)
market_snapshot	Global macro snapshot — major indices, top crypto, FX rates, commodity ETFs

Portfolio & Paper Trading

Tool	Description
get_user_portfolio	Current portfolio state — open positions, total value, unrealised P&L
execute_portfolio_trade	Execute a paper trade with risk controls (no real money)
get_portfolio_trade_history	Recent paper trade executions for a user

Signals, Alerts & Execution

Tool	Description
save_trade_signal	Persist a BUY/SELL/HOLD signal with confidence and optional metadata
list_trade_signals	List stored trade signals
set_user_alert_preferences	Configure alert delivery channel (webhook, email) and confidence threshold
dispatch_trade_alert	Dispatch an alert using the user's stored preferences
execute_order	Unified order entry — paper mode now, live brokerage integration in future

Risk & Position Sizing

Tool	Description
get_kelly_position_size	Half-Kelly position size fraction from win rate and R:R ratio
get_trade_levels	Compute stop-loss and take-profit price levels from entry and percentages
assess_portfolio_trade_risk	Pre-trade risk assessment checking sector concentration and leverage limits
assess_trade_risk_full	Comprehensive risk assessment across EGX/NSE/BSE — symbol validity, position sizing, R:R, account impact
assess_liquidity_for_trade	Liquidity assessment for a proposed position size
kelly_position_size	Full Kelly Criterion sizing with configurable Kelly fraction and account balance
risk_based_position_size	Position size from fixed % risk per trade rule
portfolio_position_size	Position size respecting max open positions and concentration limits
compute_portfolio_correlation	Pairwise Pearson correlation matrix from historical return series

Live Market & Trade Analytics

Tool	Description
get_live_price	Real-time price snapshot for any symbol — price, 24h change, volume, bid/ask
get_multi_price	Live prices for multiple symbols in one call
get_global_market_overview	Global market overview with major indices, crypto, and ETFs
create_price_alert	Create an "above" or "below" price alert for any symbol
get_active_price_alerts	List all active price alerts, optionally filtered by symbol
record_trade	Record a trade entry (open or completed) for analytics tracking
close_trade	Close an open trade at exit price and compute realised P&L
get_user_trade_history	Trade history with optional symbol and status filters
calculate_trade_metrics	Performance metrics — win rate, profit factor, Sharpe, max drawdown, recovery factor

Strategy Optimization

Tool	Description
recognize_market_pattern	AI-powered pattern recognition from OHLCV candles and indicators — detects reversal, breakout, consolidation, and more
optimize_strategy	AI-driven parameter optimization suggestions from current metrics and market regime
detect_market_anomalies	Statistical anomaly detection (z-score) in performance metrics history

---

Example Prompts

# Screening
"Show top gainers on KuCoin 15m"
"Run a Bollinger squeeze scan on Binance 4h"
"Find NASDAQ strong buy signals"

# Analysis
"Full technical analysis of BTCUSDT on Binance 4h"
"Multi-timeframe analysis of ETHUSDT on KuCoin"
"Run a 3-agent debate on AAPL NASDAQ daily"
"Combined technical + sentiment + news for TSLA"

# EGX
"EGX market overview today"
"Scan EGX banks sector"
"Generate a trade plan for COMI"
"EGX30 index analysis"

# Backtesting
"Backtest RSI strategy on AAPL for 1 year"
"Compare all 6 strategies on BTC-USD 2 years"
"Walk-forward validate MACD on SPY — is it overfitted?"

# Real-time data
"What is Apple's current price?"
"Global market snapshot"
"Latest crypto news"

# Portfolio
"Paper buy 0.1 BTC at 45000 for user alice"
"Show my portfolio for user alice"

# Risk
"Kelly position size: 55% win rate, 2:1 reward/risk"
"Trade levels for long at 45000, 3% stop, 9% target"

---

Docker

# Pull and run
docker pull debarpan2000/tradingview-mcp:latest
docker run -p 8000:8000 debarpan2000/tradingview-mcp:latest

# Or with docker compose
docker compose up -d

Claude Desktop config for Docker:

{
  "mcpServers": {
    "tradingview": {
      "command": "curl",
      "args": ["-s", "http://localhost:8000/mcp"]
    }
  }
}

---

OpenClaw (Telegram / WhatsApp)

Route Claude through Telegram, WhatsApp, Discord, and 20+ messaging platforms with OpenClaw:

Telegram message -> OpenClaw agent -> trading.py -> tradingview-assistant -> result

# Install the tool
uv tool install tradingview-assistant

# Install the skill and wrapper
mkdir -p ~/.agents/skills/tradingview-mcp ~/.openclaw/tools
curl -fsSL https://raw.githubusercontent.com/DEBARPAN2000/tradingview-mcp/main/openclaw/SKILL.md \
  -o ~/.agents/skills/tradingview-mcp/SKILL.md
curl -fsSL https://raw.githubusercontent.com/DEBARPAN2000/tradingview-mcp/main/openclaw/trading.py \
  -o ~/.openclaw/tools/trading.py && chmod +x ~/.openclaw/tools/trading.py

openclaw gateway install && systemctl --user start openclaw-gateway.service

See OPENCLAW.md for the full setup guide, multi-provider AI model options, and common mistakes.

---

Troubleshooting

Windows: MCP timeout on first launch

Symptom: MCP error -32001: Request timed out in Claude Desktop logs.

Cause: uvx compiles pandas from source on Python 3.14 (no prebuilt wheel), exceeding the 60-second timeout.

Fix — pin to Python 3.13:

{
  "mcpServers": {
    "tradingview": {
      "command": "uvx",
      "args": ["--python", "3.13", "--from", "tradingview-assistant", "tradingview-mcp"]
    }
  }
}

Or pre-warm the cache once:

uv tool install --python 3.13 tradingview-assistant

macOS: uvx: command not found

Use the absolute path in your config: /Users/YOUR_USERNAME/.local/bin/uvx

No symbols returned for an exchange

Check the exchange code is uppercase and supported: KUCOIN, BINANCE, BYBIT, MEXC, OKX, EGX, NASDAQ, NYSE, BIST, etc.

TradingView rate limiting

Space out scans by a few seconds. The server has a built-in request guard with configurable rate limits.

---

Project Structure

tradingview-mcp/
├── src/tradingview_mcp/
│   ├── server.py                        # MCP tool registry (40+ tools)
│   ├── coinlist/                        # Per-exchange symbol lists
│   └── core/
│       ├── services/
│       │   ├── backtest_service.py      # 6-strategy backtesting engine
│       │   ├── egx_service.py           # Egyptian Exchange tools
│       │   ├── correlation_service.py   # Portfolio correlation
│       │   ├── risk_service.py          # Kelly, trade levels, risk checks
│       │   ├── portfolio_service.py     # Paper trading execution
│       │   ├── alert_service.py         # Alert delivery
│       │   ├── data_source.py           # TradingView screener integration
│       │   ├── http_client.py           # Shared HTTP client
│       │   └── market_adapters/         # EGX and NSE adapters
│       ├── data/                        # Static datasets (sectors, indices)
│       ├── utils/                       # Validators and error contracts
│       └── portfolio.py                 # SQLite persistence layer
├── tests/
│   ├── unit/                            # Unit tests
│   └── integration/                     # Live integration tests
├── scripts/
│   └── repo_hygiene_check.py            # Security hygiene checker
├── openclaw/
│   ├── SKILL.md                         # OpenClaw skill definition
│   └── trading.py                       # OpenClaw tool wrapper
├── Dockerfile
├── docker-compose.yml
├── TOOLS.md                             # Complete tool reference
├── INSTALLATION.md                      # Detailed installation guide
├── OPENCLAW.md                          # OpenClaw integration guide
├── CONTRIBUTING.md                      # Contribution guide
└── SECURITY.md                          # Security policy

---

Contributing

See CONTRIBUTING.md for the full guide.

Quick steps:

1. Fork and clone the repo
2. uv sync to install all dependencies
3. uv run pytest tests/unit/ to verify your setup
4. Create a feature branch, make changes, run tests
5. Open a PR against main with a clear description

---

License

MIT © 2025 DEBARPAN BOSE CHOWDHURY — see LICENSE.