I Built an AI bot that trades. Finally profitible

I'm a finance broker by trade, and I got tired of manually watching charts. So I built an automated system using AI (OpenClaw + Python) that:

• Scans every USDT pair on Binance with >$5M volume
• Uses 4H klines with RSI(14), 20-period breakout, and volume spike confirmation
• Enters with market orders, 40% position sizing, max 2 concurrent
• Exits with layered take-profits: 20% at +30% (stop to breakeven), 20% at +50% (20% trail), 20% at +100% (10% trail)
• Kill switch at 50% drawdown, daily loss limit 20%
• Adaptive learning: adjusts entry thresholds every 10 trades based on win rate

Runs 24/7 on a $6/mo VPS in Sydney. Sends alerts for every action.

The interesting part is the adaptive learning — if win rate drops below 35%, it tightens entry filters. Above 60%, it loosens them. Simple feedback loop but it keeps the bot aligned with market conditions.

I wrote up the full strategy, code, and deployment process:

Happy to answer questions about the approach.