Kraken Meme Coin Trading Bot 🤖

An automated cryptocurrency trading bot for the Kraken exchange. Trades meme coins and altcoins using AI-powered news analysis, volume spike detection, and new listing monitoring — running 24/7 on a VPS server.

Features

• News-based signals — Uses Claude AI (claude-opus-4-6) to analyze crypto news headlines from 6 RSS feeds and generate buy/sell signals
• Pump detector — Identifies coins with 3x+ volume spikes vs. normal (early pump detection)
• New listing monitor — Watches Kraken's blog RSS feed and buys coins the moment they list
• Risk management — Daily loss limit, per-trade size caps, dry run mode
• Kill switch — touch KILL in the repo root halts all trading instantly (next cycle becomes a no-op); rm KILL resumes — no restart, no SSH-to-systemctl
• 24/7 operation — Runs as a systemd service on a Linux VPS

How It Works

Every 5 minutes the bot:

1. Checks your CAD balance and daily loss limit
2. Scans Kraken blog for new coin listings → buys immediately
3. Detects volume spikes across all tradable coins
4. Fetches 60 latest crypto headlines and sends them to Claude AI
5. Combines signals and places market orders on Kraken

Architecture

Three independent signal sources feed a single decision loop. The trader merges them, applies risk caps and exit logic, and routes orders through one Kraken REST client. Full strategy details live in STRATEGY.md.

flowchart LR
    subgraph external["External sources"]
        RSS["RSS feeds<br/>crypto news + ~50 Nitter accounts"]
        BLOG["Kraken blog RSS"]
        TICK["Kraken Ticker API"]
        CLAUDE["Anthropic Claude<br/>claude-opus-4-6"]
    end

    subgraph signals["Signal layer"]
        NF["news_fetcher.py"]
        MM["market_matcher.py"]
        LM["listing_monitor.py"]
        PD["pump_detector.py"]
    end

    TRADER["trader.py<br/>cycle orchestrator"]
    POS[("positions.json<br/>trades.csv")]
    KC["kraken_client.py"]
    KEX["Kraken Exchange<br/>REST API"]

    RSS --> NF
    NF --> MM
    MM <--> CLAUDE
    BLOG --> LM
    TICK --> PD

    MM --> TRADER
    LM --> TRADER
    PD --> TRADER

    TRADER --> KC
    KC --> KEX
    TRADER <--> POS

Loading

Setup

Prerequisites

• Python 3.8+
• Kraken account with API key (trading permissions)
• Anthropic API key

Install

git clone https://github.com/YOUR_USERNAME/kraken-bot.git
cd kraken-bot
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Configure

Create a .env file (never commit this):

KRAKEN_API_KEY=your_kraken_api_key
KRAKEN_PRIVATE_KEY=your_kraken_private_key
ANTHROPIC_API_KEY=your_anthropic_api_key
MAX_TRADE_AMOUNT=40.0
MIN_CONFIDENCE=0.65
RUN_INTERVAL_MINUTES=5
DAILY_LOSS_LIMIT=100.0
DRY_RUN=true

Set DRY_RUN=false to go live.

Run locally

source venv/bin/activate
python3 main.py

Deploy to VPS (Linux/Ubuntu)

# Upload files
scp -r . root@YOUR_SERVER_IP:/root/kraken-bot

# SSH in and install
ssh root@YOUR_SERVER_IP
cd /root/kraken-bot
python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt

Create /etc/systemd/system/kraken-bot.service:

[Unit]
Description=Kraken Meme Coin Trading Bot
After=network.target

[Service]
Type=simple
User=root
WorkingDirectory=/root/kraken-bot
ExecStart=/root/kraken-bot/venv/bin/python main.py
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

systemctl daemon-reload
systemctl enable kraken-bot
systemctl start kraken-bot

Check logs:

journalctl -u kraken-bot -n 50 --no-pager

Project Structure

kraken-bot/
├── main.py              # Scheduler — runs every N minutes
├── trader.py            # Main trading logic
├── kraken_client.py     # Kraken API wrapper
├── market_matcher.py    # Claude AI news analysis
├── pump_detector.py     # Volume spike detection
├── listing_monitor.py   # New listing monitor (Kraken blog RSS)
├── news_fetcher.py      # RSS feed fetcher (6 sources)
├── config.py            # Loads .env settings
└── requirements.txt

Daily Iteration

This repo follows a daily iteration discipline. Each day a small, scoped improvement lands as a commit so the bot keeps compounding.

• DAILY_ITERATIONS.md holds the 30 task backlog (docs, tests, refactors, observability, features, ops).
• JOURNAL.md records what shipped each day.
• Run ./scripts/daily.sh from the repo root to see today's task.

Risk Warning

This bot trades real money. Crypto is extremely volatile. Use DRY_RUN=true to test before going live. Set conservative DAILY_LOSS_LIMIT and MAX_TRADE_AMOUNT values. Past performance does not guarantee future results.

Tech Stack

• Python 3
• krakenex — Kraken REST API
• Anthropic Claude API — AI news analysis
• feedparser — RSS ingestion
• schedule — job scheduling