📈 QuantTrader Lite

AI-Powered Crypto Trading Signal Agent — Lablab.ai AI Trading Agents Hackathon 2026

---

Landing Page

Dashboard Preview

Feature Highlights

---

🧠 What is QuantTrader Lite?

QuantTrader Lite is a fully working, end-to-end AI crypto trading signal agent. It fetches live prices from multiple sources, asks Gemini 2.0 Flash for a BUY / SELL / HOLD signal with a reason, and executes paper trades via Kraken CLI — all visible live on a clean Streamlit dashboard.

Live Price  →  Gemini AI  →  Signal + Reason  →  Kraken CLI (sandbox)  →  Dashboard

🔗 Try it live: https://quant-trader-lite.streamlit.app/

🏗️ Architecture

.

---

✨ Features

Feature	Description
📡 Multi-source price data	CoinGecko → Binance → CryptoCompare fallback chain — never fails silently
🤖 Trend-aware AI signals	Gemini 2.0 Flash analyzes live price + session trend for smarter signals
🟢🔴🟡 BUY / SELL / HOLD	Color-coded signals with full AI reasoning visible on every trade
📊 4 coins supported	Bitcoin, Ethereum, Solana, BNB — each with chart tab and Kraken pair
📋 Full trade log	Every signal logged to JSON with timestamp, price, signal, and order result
🔄 Auto-refresh	Toggle 60-second auto-refresh for hands-free live agent operation
⚙️ Kraken CLI sandbox	Paper trades via Kraken CLI — no real money, full execution flow

---

🚀 Quickstart

1. Clone the repo

git clone https://github.com/your-team/quanttrader-lite
cd quanttrader-lite

2. Install dependencies

pip install -r requirements.txt

3. Set your API keys

export GEMINI_API_KEY=your_gemini_key_here
export KRAKEN_API_KEY=your_kraken_key        # optional for sandbox
export KRAKEN_API_SECRET=your_kraken_secret  # optional for sandbox

4. Run the dashboard

streamlit run streamlit_app.py

Open http://localhost:8501 and click ⚡ Run Agent Now.

---

🗂️ File Structure

quanttrader-lite/
├── streamlit_app.py     # Main app — price fetch + Gemini AI + Kraken + dashboard
├── requirements.txt     # All Python dependencies
├── trade_log.json       # Auto-generated trade history (gitignored)
├── index.html           # Landing page for the project
└── README.md            # This file

---

🛠️ Tech Stack

Layer	Tool	Why
AI Signal Engine	Gemini 2.0 Flash	Fast, reliable, free API tier available
Fallback AI	Groq (llama3)	Ultra-fast inference backup
Market Data (primary)	CoinGecko API	Free, no key needed, reliable
Market Data (fallback 1)	Binance Public API	No key needed, high uptime
Market Data (fallback 2)	CryptoCompare	Free tier, good historical data
Trading Execution	Kraken CLI	Pre-built binary, full sandbox support
Dashboard UI	Streamlit	Python-only, no frontend skills needed
Language	Python 3.11+	Simple, fast, well-supported

---

🔄 Workflows

Agent Pipeline (every run)

User clicks ⚡ Run Agent Now
        │
        ▼
[1] Fetch Price
    Try CoinGecko → if fail → Try Binance → if fail → Try CryptoCompare
        │
        ▼
[2] Compute Trend
    Load trade_log.json → analyze last 10 prices for this coin
    → Direction: UPTREND / DOWNTREND / SIDEWAYS
    → Session High / Low / Avg / % Move
        │
        ▼
[3] Ask Gemini AI
    Send: coin, price, 24h change, trend context
    Receive: SIGNAL: BUY/SELL/HOLD | RISK: LOW/MEDIUM/HIGH | REASON: ...
        │
        ▼
[4] Execute Trade (Kraken CLI Sandbox)
    BUY  → kraken order buy  <pair> <amount> --sandbox
    SELL → kraken order sell <pair> <amount> --sandbox
    HOLD → no order placed
        │
        ▼
[5] Log Trade
    Append to trade_log.json:
    { timestamp, coin, price, change_24h, signal, trade, source }
        │
        ▼
[6] Update Dashboard
    Metrics · Trend panel · AI reasoning box · Charts · Trade table

Auto-Refresh Workflow

Toggle ON "🔄 Auto-refresh every 60s"
        │
        ▼
    Dashboard sleeps 60 seconds
        │
        ▼
    st.rerun() triggers full page reload
        │
        ▼
    Agent pipeline runs again automatically
        │
        └──── loops indefinitely until toggle OFF

Price Fallback Workflow

Request price for coin
        │
        ▼
    CoinGecko API ──── success ──▶ return (price, change, "CoinGecko")
        │
      fail
        │
        ▼
    Binance API ────── success ──▶ return (price, change, "Binance")
        │
      fail
        │
        ▼
    CryptoCompare ──── success ──▶ return (price, change, "CryptoCompare")
        │
      fail
        │
        ▼
    RuntimeError: "All price sources failed" → show error in dashboard

---

📊 Signal Logic

Gemini AI follows these rules when generating signals:

Condition	Signal
Trend = UPTREND and 24h change > +1%	🟢 BUY
Trend = DOWNTREND and 24h change < -1%	🔴 SELL
Sideways, uncertain, or mixed signals	🟡 HOLD
Any condition	Max 2% portfolio risk enforced

---

🪙 Supported Coins

Coin	CoinGecko ID	Binance Pair	Kraken Pair
Bitcoin	bitcoin	BTCUSDT	XBTUSD
Ethereum	ethereum	ETHUSDT	ETHUSD
Solana	solana	SOLUSDT	SOLUSD
BNB	binancecoin	BNBUSDT	BNBUSD

---

🔑 API Keys Needed

Service	Required	Where to get
Gemini API	✅ Yes	aistudio.google.com
Groq API	Optional (fallback AI)	console.groq.com
CryptoCompare	Optional (price fallback)	cryptocompare.com/cryptopian/api-keys
CoinMarketCap	Optional (extra data)	coinmarketcap.com/api
Messari	Optional (market intel)	messari.io/api
CoinGecko	❌ Free, no key	Built-in
Binance	❌ Free, no key	Built-in
Kraken CLI	For real trades	kraken.com/u/security/api

---

✅ Submission Checklist

• Working Streamlit app (AI + Kraken CLI + price feeds)
• Multi-source price fallback (CoinGecko → Binance → CryptoCompare)
• Gemini AI signal with trend context and reasoning
• Kraken CLI sandbox trade execution
• Trade log (JSON) with full history
• Live dashboard with charts, metrics, signal distribution
• GitHub repository (public)
• Live deployment: quant-trader-lite.streamlit.app
• Demo video (2–3 min screen recording)
• Social media post (X / LinkedIn)
• Register on early.surge.xyz
• Submit on Lablab.ai before April 12

---

👥 Team

Member	Role
Member 1	Python Dev — AI agent, Kraken CLI, price APIs
Member 2	Dashboard — Streamlit UI, charts, signal display
Member 3	Presenter — Demo video, README, pitch

---

📄 License

MIT License — free to use, modify, and distribute.

---

QuantTrader Lite · Lablab.ai AI Trading Agents Hackathon 2026 · Kraken CLI Challenge

🚀 Live Demo · 🏆 Hackathon Page · 📋 Prize Registration