Analyzer Bot

Production-grade XAU/USD trading signal bot for Telegram, powered by FastAPI + Next.js dashboard. Generates scalp trading signals using signal generator logic across 4 timeframes, logs trades to Firebase, and includes multi-model AI screenshot verification (Nvidia Vision + Llama).

Features

• Signal Engine — 4-timeframe analysis (1H/4H trend, 15M S/R, 5M pullback, 1M entry confirmation)
• Telegram Bot — 6 commands: /signal, /analyze, /log_trade, /stats, /dashboard, /help
• Screenshot Verification — Upload chart screenshots for dual-verified signals via Nvidia Vision AI
• REST API — Endpoints for signals, trades, and trading statistics
• Web Dashboard — Next.js 14 dashboard with charts, trade table, AI trade analysis (Nvidia Llama 3.3 70B), learning Q&A (Nvidia), and screenshot upload
• Firebase Firestore — Persistent storage for signals and trade logs
• TradingView Integration — Real-time OHLCV candle data via RapidAPI

Tech Stack

Bot: FastAPI · python-telegram-bot · Firebase Admin · Pandas · aiohttp · Uvicorn
Dashboard: Next.js 14 · Tailwind CSS · Recharts · Firebase Web SDK
AI: Nvidia Vision (Llama 3.2 11B) · Nvidia Strategist (Llama 3.3 70B) · Nvidia Processor (Llama 3.1 8B)

Getting Started

Bot

cd bot
pip install -r requirements.txt

Fill in bot/.env with your API keys, then:

uvicorn main:app --reload

Dashboard

cd web
npm install
npm run dev

Fill in web/.env.local with API keys and Firebase config.

Project Structure

bot/
├── main.py                  # FastAPI entry point
├── config.py                # Environment configuration
├── app/
│   ├── models.py            # Pydantic schemas
│   ├── tradingview_client.py
│   ├── signal_generator.py
│   ├── firebase_manager.py
│   ├── telegram_handler.py
│   └── services/
│       ├── nvidia_vision_analyzer.py
│       └── multi_model_pipeline.py
├── routers/
│   ├── signals.py
│   ├── trades.py
│   └── telegram.py
└── utils/
    ├── logger.py
    ├── validators.py
    └── helpers.py

web/
├── app/
│   ├── layout.tsx, page.tsx
│   ├── dashboard/
│   │   ├── page.tsx              # Overview
│   │   ├── trades/page.tsx       # Trade log
│   │   ├── analytics/page.tsx    # Charts & AI
│   │   └── learning/page.tsx     # Q&A chatbot
│   └── api/                      # Next.js API routes
│       ├── trades/route.ts
│       ├── stats/route.ts
│       ├── analyze-trade-nvidia/route.ts
│       ├── learn/route.ts
│       ├── analyze-screenshot/route.ts
│       └── verification-history/route.ts
├── components/
│   ├── layout/                   # Header, Sidebar
│   ├── dashboard/                # StatsCards, PerformanceChart
│   ├── trades/                   # TradeTable, TradeFilters
│   ├── ai/                       # TradeAnalyzer, LearningBot,
│   │                             # ScreenshotAnalyzer, VerificationDashboard
│   └── ui/                       # Card, Button, Tabs, Spinner
├── lib/                          # firebase, api-client, formatters
└── types/                        # TypeScript interfaces

API Endpoints (Bot)

Method	Path	Description
GET	/	Root info
GET	/health	Health check
POST	/api/signal	Generate trading signal
GET	/api/signal/{id}	Get signal by ID
POST	/api/trades	Log a trade
GET	/api/trades	Get all trades
GET	/api/stats	Trading statistics

AI Models Used

Model	Purpose
Llama 3.2 11B Vision	Chart screenshot analysis (support/resistance, patterns)
Llama 3.3 70B Instruct	Trade analysis & strategic insights
Llama 3.1 8B Instruct	Data processing & level identification

All models run via Nvidia NIM (free tier) — no Claude/OpenAI costs.

---

Live Dashboard: analyzer-dashboard-kohl.vercel.app

---

Built for AOT — portfolio project.