My Setup - Singapore Trading Agent

🖥️ Hardware & Infrastructure

Machine

N150 Mini PC

Intel Alder Lake-N, 4 cores, up to 3.6GHz

RAM

8GB DDR5

7.5GB usable - sufficient for OpenClaw + scripts

Storage

128GB Hanstor M.2 SSD

119GB usable - monitor space if logging heavily

Internet

Singtel Gigabit

~939 Mbps symmetrical

OS

Ubuntu 24.04.4 LTS

Latest LTS, easy cron management

Power

~15W idle

Runs 24/7, ~$5/month electricity

💡 Why This Hardware Works

The N150 is perfect for trading automation because:

Low power: Can run 24/7 without worrying about electricity costs
Quiet: No fan noise, fits anywhere in the home
Enough power: Handles OpenClaw + Python scripts + browser automation easily
DDR5 RAM: Faster memory helps with multitasking
Not for HFT: Don't expect microsecond latency—and that's fine for swing trading

⚠️ Storage Management (128GB)

With 119GB usable, monitor disk space:

Rotate logs weekly: find ~/logs -name "*.log" -mtime +7 -delete
Avoid storing large datasets locally—use APIs or cloud storage
Consider external SSD or cloud sync for historical data

🤖 Agent Configuration

Agent Name

DaNiu 大牛🐮

Model

Anthropic Claude

Primary Channel

Integration

Notion API

SOUL.md Snippet (Trading-Specific)

# Trading Agent Rules ## Risk Controls - NEVER execute trades without explicit human approval (until Level 5) - ALWAYS verify prices against at least 2 sources before reporting - FLAG any data that is delayed or estimated - RESPECT position limits: max 5% per position, max 20% total portfolio ## Market Hours (Singapore Time) - SGX: 9:00 AM - 5:00 PM SGT (Monday-Friday) - US: 9:30 PM - 4:00 AM SGT (Sunday-Thursday evening) - Crypto: 24/7 ## Data Verification - Yahoo Finance: 15-min delay for SGX, use for swing only - SEC EDGAR: Real-time filings, trusted source - Cross-check critical data before reporting ## Communication Style - Be concise, use bullet points - Include timestamps on all data - Cite sources for all numbers - Flag uncertainty clearly

✅ Recommended Trading Automations (For This Setup)

🎯 Start With These 3 (Low Power, High Value)

These run efficiently on the N150 and provide maximum value:

SGX/US Morning Briefing (7am SGT daily, 2-3 min runtime)
Watchlist Price Alerts (every 15 min during market hours, 30 sec runtime)
Weekly Portfolio Report to Notion (Friday 6pm SGT, 1-2 min runtime)

What Works Well on Low-Power Hardware

✓ Scheduled tasks (cron jobs) that run briefly
✓ Monitoring and alerting (no continuous polling)
✓ Daily/weekly reports and summaries
✓ Browser automation for occasional data fetching
✓ API calls with reasonable rate limits

What to Avoid on This Hardware

✗ Real-time tick-by-tick analysis (too much CPU/network)
✗ High-frequency trading (latency too high anyway)
✗ Multiple concurrent browser sessions (8GB RAM limit)
✗ Running 24/7 Python loops (use cron instead)
✗ Local LLM inference (use API-based models)
✗ Storing large datasets locally (128GB SSD fills fast)

⏰ Optimal Cron Schedules (SGT Timezone)

Market Hours Reference

SGX Market

9:00 AM - 5:00 PM

SGT (Mon-Fri)

US Market

9:30 PM - 4:00 AM

SGT (Sun-Thu evening)

Crypto

24/7

Global markets

My Crontab Configuration

# Edit with: crontab -e

# === Morning Briefing (Weekdays 7:00 AM SGT) ===
0 7 * * 1-5 cd ~/.openclaw/workspace/skills/market-briefing && python3 market_briefing.py | openclaw send --to telegram --message "stdin"

# === SGX Price Alerts (Every 15 min during SGX hours) ===
*/15 9-16 * * 1-5 cd ~/.openclaw/workspace/skills/price-alerts && python3 price_alerts.py 2>&1 | grep -v "No alerts" | openclaw send --to telegram --message "stdin"

# === US Market Alerts (Every 15 min during US hours, SGT) ===
*/15 21-23 * * 0-4 cd ~/.openclaw/workspace/skills/price-alerts && python3 price_alerts.py 2>&1 | grep -v "No alerts" | openclaw send --to telegram --message "stdin"
*/15 0-3 * * 1-5 cd ~/.openclaw/workspace/skills/price-alerts && python3 price_alerts.py 2>&1 | grep -v "No alerts" | openclaw send --to telegram --message "stdin"

# === Portfolio Sync to Notion (Friday 6:00 PM SGT) ===
0 18 * * 5 cd ~/.openclaw/workspace/skills/portfolio-sync && python3 sync_notion.py >> ~/logs/portfolio-sync.log 2>&1

# === Earnings Check (Daily 8:00 AM SGT) ===
0 8 * * * cd ~/.openclaw/workspace/skills/earnings && python3 check_earnings.py 2>&1 | openclaw send --to telegram --message "stdin"

# === Log Rotation (Sunday 1:00 AM SGT) ===
0 1 * * 0 find ~/logs -name "*.log" -mtime +30 -delete

✅ Pro Tips

Redirect output to logs for debugging: >> ~/logs/briefing.log 2>&1
Use grep -v "No alerts" to avoid spam when no alerts
Test cron jobs manually first before scheduling
Check cron logs: grep CRON /var/log/syslog

📓 Notion Integration

I use Notion for portfolio tracking, watchlist management, and trade journal.

Database Structure

Database	Purpose	Update Frequency
Holdings	Current positions, quantity, avg cost, current value	Daily (6pm SGT)
Watchlist	Stocks to monitor, target entry/exit, thesis	As needed
Trade Journal	Executed trades, entry/exit, P&L, lessons learned	Per trade
Dividend Tracker	Dividend stocks, ex-dates, payment dates, yield	Weekly

Notion API Setup

# 1. Create integration at: https://www.notion.so/my-integrations
# 2. Copy "Internal Integration Token"
# 3. Share your databases with the integration
# 4. Store token in environment variable:

export NOTION_API_KEY="secret_xxxxx"

# 5. In your Python scripts:
import os
from notion_client import Client

notion = Client(auth=os.environ.get('NOTION_API_KEY'))

⚠️ Security Note Never commit your Notion API key to git. Use environment variables or a .env file (added to .gitignore).

🇸🇬 Singapore-Specific Considerations

SGX Trading

Trading hours: 9:00 AM - 5:00 PM SGT (pre-market 8:30-9:00, after-hours 5:00-5:15)
Data delay: Yahoo Finance has 15-min delay. Not suitable for intraday.
Popular brokers: POEMS, DBS Vickers, IBKR SG, Moomoo SG
Best for automation: Interactive Brokers Singapore (full API access)

US Trading from Singapore

Trading hours: 9:30 PM - 4:00 AM SGT (Sunday evening to Friday early morning)
Settlement: T+1 (one business day)
Currency: SGD-USD conversion fees apply (use Wise or multi-currency account)
Tax: No capital gains tax in Singapore for personal trading

MAS Regulations (Personal Trading)

Personal trading is generally unregulated (you're trading your own money)
Crypto trading is legal but unregulated (MAS has issued investor warnings)
If managing others' money, you need a Capital Markets Services License
No Pattern Day Trader rule (unlike US)

💡 Recommended Setup for Singapore Traders

Broker: Interactive Brokers Singapore (SGX + US + global markets, full API)
Data: Yahoo Finance (free, 15-min delay) for swing trading
Focus: SGX dividend stocks + US tech growth
Timezone advantage: Monitor US market overnight while you sleep, review alerts in morning

💰 My Monthly Costs

Item	Cost (SGD)	Notes
Hardware (N150 mini PC)	$250 (one-time)	Amortized over 3 years = ~$7/mo
Electricity	~$5/mo	15W idle, 24/7 operation
Anthropic API (Claude)	$20-40/mo	Varies with usage
Market Data	$0	Yahoo Finance (free)
Notion	$0	Free tier sufficient
Telegram	$0	Free
Total (monthly)	~$32-52/mo	Plus one-time hardware

Comparison: This is cheaper than TradingView Pro ($60/mo) and vastly cheaper than Bloomberg Terminal ($2,000/mo), while providing similar monitoring and alerting capabilities for my use case.

🚀 What's Next for My Setup

Planned upgrades and improvements:

□ Add Polygon.io for real-time US data ($29/mo)
□ Implement paper trading bot with Alpaca
□ Add earnings transcript summarization
□ Build multi-agent setup (researcher + executor)
□ Integrate with SGX API for real-time data (if available)

Last updated: 2026-03-29 | ← Back to Home | Starter Guide →