Deploy Always-On

A t3.micro (2 vCPU, 1 GB RAM) handles AlgoDeploy with headroom to spare. No GPU needed.

1. Create an AWS account

Sign up at aws.amazon.com/free. New accounts get 12 months of free tier, which includes 750 hours/month of t3.micro — enough to run one instance continuously for a full year at no compute cost.

2. Launch an EC2 instance

• Go to EC2 → Launch Instance
• Name: algodeploy
• AMI: Ubuntu Server 24.04 LTS (free tier eligible)
• Instance type: t3.micro
• Key pair: Create a new key pair (download the .pem file — you'll need it to SSH in)
• Security group: Allow SSH (port 22) from your IP
• Storage: 8 GB gp3 (default is fine)
• Click Launch Instance

3. Connect via SSH

Once the instance is running, connect from your terminal:

# Make your key file read-only (required)
chmod 400 algodeploy-key.pem

# Connect
ssh -i algodeploy-key.pem ubuntu@YOUR_INSTANCE_PUBLIC_IP

On Windows, use PowerShell or WSL. The chmod step isn't needed on PowerShell — just run the ssh command directly.

4. Install Python and dependencies

sudo apt update && sudo apt upgrade -y
sudo apt install -y python3 python3-venv python3-pip

5. Upload AlgoDeploy

From your local machine, copy the AlgoDeploy directory to your instance:

# From your local terminal (not SSH)
scp -i algodeploy-key.pem -r ./algodeploy ubuntu@YOUR_INSTANCE_IP:~/algodeploy

Then on the instance:

cd ~/algodeploy
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

6. Configure environment

Create your .env file on the instance:

nano ~/algodeploy/.env

Add your broker credentials:

ALPACA_API_KEY=your_api_key
ALPACA_SECRET_KEY=your_secret_key
ALPACA_PAPER=true

7. Test it manually first

Before setting up auto-start, verify everything works:

source ~/algodeploy/venv/bin/activate
algodeploy run my_strategy.yaml

Watch the output for a few minutes. Confirm it connects to Alpaca, scans for signals, and logs activity. Ctrl+C to stop.

8. Install systemd services for continuous operation

AlgoDeploy ships with systemd service files that auto-start the trader and dashboard on boot:

# Copy the service files
sudo cp ~/algodeploy/services/trader.service /etc/systemd/system/
sudo cp ~/algodeploy/services/dashboard.service /etc/systemd/system/

# Create the logs directory
mkdir -p ~/algodeploy/logs

# Enable and start
sudo systemctl daemon-reload
sudo systemctl enable algodeploy-trader
sudo systemctl enable algodeploy-dashboard
sudo systemctl start algodeploy-trader
sudo systemctl start algodeploy-dashboard

9. Verify it's running

# Check service status
sudo systemctl status algodeploy-trader

# Watch live logs
tail -f ~/algodeploy/logs/trader.log

# Check both services
sudo systemctl status algodeploy-trader algodeploy-dashboard

Update your strategy

Edit your config locally, then push the update:

# From your local machine
scp -i algodeploy-key.pem my_strategy.yaml ubuntu@YOUR_IP:~/algodeploy/

# SSH in and restart the trader
ssh -i algodeploy-key.pem ubuntu@YOUR_IP
sudo systemctl restart algodeploy-trader

View logs

# Live trader log
tail -f ~/algodeploy/logs/trader.log

# Last 50 lines of errors
tail -50 ~/algodeploy/logs/trader_error.log

# Systemd journal (includes start/stop/crash events)
sudo journalctl -u algodeploy-trader --since "1 hour ago"

Stop / restart

# Stop trading
sudo systemctl stop algodeploy-trader

# Restart (picks up config changes)
sudo systemctl restart algodeploy-trader

# Disable auto-start on boot
sudo systemctl disable algodeploy-trader

Access the dashboard remotely

The dashboard runs on port 8004. To access it securely from your browser, use an SSH tunnel instead of opening the port to the internet:

# From your local machine
ssh -i algodeploy-key.pem -L 8004:localhost:8004 ubuntu@YOUR_IP

Then open http://localhost:8004 in your browser. The dashboard is now accessible through the secure tunnel.

• SSH key only — never enable password authentication
• Restrict SSH access — set your security group to allow port 22 from your IP only
• Don't open port 8004 — use SSH tunneling for dashboard access (shown above)
• Keep your .env file secure — it contains your broker API keys
• Update regularly — run sudo apt update && sudo apt upgrade -y periodically
• Use paper mode first — validate on the instance before switching to live

AWS publishes a 99.5% uptime SLA for single EC2 instances (99.99% for multi-AZ deployments). On a single instance, that translates to roughly 4.4 hours of potential downtime per year — significantly more reliable than a laptop or desktop machine.

For additional resilience, consider:

• Notifications — configure Telegram, Discord, or webhook alerts so you know immediately if the trader stops or encounters errors
• Systemd auto-restart — the included service files automatically restart the trader within 15 seconds of a crash (up to 3 retries per 5 minutes)
• AWS CloudWatch — set up a free basic alarm to alert you if your instance becomes unreachable