Descriptions:
Prediction market trading on Polymarket requires more than intuition — it demands systematic research, liquidity awareness, and disciplined risk management. This tutorial by Sharbel A. walks through how to build a complete AI-assisted trading system from scratch using the Bullpen CLI alongside Claude Code and Open Claw, covering installation through five distinct live trading strategies.
Bullpen, which sponsors the video, provides a structured command layer over Polymarket’s API — abstracting away market slugs, wallet management, and position handling into JSON outputs that AI agents can reason over directly. Using this foundation, the tutorial builds working agents for five specific approaches: smart money tracking, crypto short-window markets, weather markets, arbitrage on correlated markets, and event-driven news trading. The video is notably transparent about the real build process, including a live debugging moment where Claude Code identifies and corrects an error — catching that the wrong Bullpen command was being used for market discovery and rewriting the relevant code automatically.
Beyond the initial builds, the tutorial covers how to wire agents to alert systems (Telegram, Discord, or WhatsApp), how to construct a backtesting harness that replays strategies against historical Polymarket settlement data before risking real capital, and how to configure three-condition safety gates that must all pass before any live order executes. A thorough, code-first guide for developers who want to apply modern AI agent frameworks to structured prediction market workflows rather than trading manually.
📺 Source: Sharbel A. · Published May 25, 2026
🏷️ Format: Hands On Build
