Telegram trading bot assistant | LLM powered trading bot for crypto/forex

Purpose & Audience This n8n workflow telegram bot is designed for crypto, forex and stock traders, quantitative analysts, and AI enthusiasts who want to explore how different Large Language Models (LLMs) perform in real-world trading scenarios. Inspired by nof1.ai's Alpha Arena, where AI models compete with real capital to prove their trading capabilities, this template lets YOU run your own AI trading experiments—testing which "brain" generates the most profitable signals for BTCUSD (it can do technical analysis on any forex/crypto with any of the 5000+ assets available at Twelve Data) and other assets.

Whether you're a day trader seeking automated technical analysis, a researcher comparing LLM performance, or simply curious about AI-driven trading—this workflow provides a sophisticated, customizable trading assistant powered by your choice of AI model.

What It Does? 1. Generates disciplined trading signals (BUY/SELL/NOSIGNAL) with precise entry price, stop loss, and take profit levels for crypto assets (default: BTCUSD) 2. Performs real-time technical analysis using 200 hourly candles—calculates RSI, MACD, Bollinger Bands, ATR, support/resistance, Fibonacci levels, and more 3. Enforces strict risk management with 1.5x ATR stop losses and minimum 1:2 risk-reward ratios 4. Acts as an intelligent telegram trading assistant that answers questions about indicators, market conditions, strategies, and provides educational guidance 5. Supports multiple AI models (LLM swapping)—test Google Gemini, OpenAI GPT-4/5, Claude Sonnet/Opus, DeepSeek, Qwen, Groq, or local models (Ollama) 6. Includes detailed reasoning for every signal with 10-15 sentences citing specific data points, confidence scores, and exit plans 7. Handles context-aware conversations with mongo db memory node recent chat context or follow-up questions.

🧪 Experimental Nature - Important Disclaimer This is an EXPERIMENTAL trading tool. Performance varies significantly across different LLM models. Based on nof1.ai's live trading competition results (at the time of template creation): 1. DeepSeek V3.1 showed +47% returns (highest performer) 2. Qwen 3 Max achieved +35% returns 3. Claude Sonnet 4.5 delivered +13% returns 4. GPT-5 experienced -74% returns (lowest performer)

Your mileage will vary. The same prompt can produce drastically different results depending on: 1. Which AI model you use (the "brain" matters!) 2. Market conditions during testing 3. Your risk management discipline 4. Assets volatility and liquidity

We strongly recommend: 1. Test with paper trading first (simulated trades, no real money) 2. Experiment with multiple LLMs to find which performs best for your strategy 3. Compare signals across models before executing trades 4. Always validate signals manually—AI is a tool, not a replacement for judgment 5. Never risk more than you can afford to lose

This template is for educational/informational purposes. Trading involves substantial risk of loss. Past AI performance does not guarantee future results.

Who Is It For? - Crypto & Forex traders who want AI-assisted technical analysis and signal generation for BTC, ETH, SOL, and other assets - Quant researchers exploring how different LLMs handle quantitative trading tasks - AI enthusiasts curious about practical applications of models like DeepSeek, Qwen, Claude, and GPT in financial markets - Algorithm developers who want a foundation to build more sophisticated trading systems - Educators & students learning about algorithmic trading, technical analysis, or AI agent workflows