Complete Guide to building an AI Trading Bot with Claude for Beginners

Summary

Ray Fu demonstrates how to build and schedule a basic stock-trading bot using Claude Code and an Alpaca paper-trading account. The bot uses a moving-average crossover strategy and is designed to place trades automatically when its rules trigger.

The walkthrough reports a successful dry run, but provides no backtest results or evidence of live-trading performance.

Strategy and workflow

• Moving averages: Calculate 20-day and 50-day averages from daily stock prices.
• Signals: Buy when the 20-day average crosses above the 50-day average; sell or exit when it crosses below. The bot checks for a new crossover between consecutive days. Otherwise, it can hold.
• Data and trade size: The example uses Yahoo Finance data for the most recent 100 days and trades one share at a time.
• Automation: The bot is scheduled to check the stock Monday through Friday at 9:30 a.m. Eastern, execute a trade if there is a signal, and log its activity. The computer must remain on for the scheduled run.

Assets and tools mentioned

• Apple: Selected for the automated example; one share per trade. The ticker is not spoken in the subtitles.
• Micron: A separate test order is placed; it is pending because the market is closed. The ticker is not spoken in the subtitles.
• Alpaca: Brokerage/API and paper-trading account.
• Yahoo Finance (yfinance): Market-data source.
• Claude Code: Used to create and automate the bot.

Key figures and cautions

• The example Alpaca paper account is funded with $50,000 in simulated money; no real money is used in the demonstration.
• The presenter cautions that moving-average crossovers can perform poorly in choppy, sideways markets, where repeated crossings may lead to multiple small losing trades. He stresses that the strategy is not a guaranteed way to make money.
• The presenter says the bot can execute consistently and without emotion, but provides no returns, win rates, risk limits, stop-loss rules, or other performance or risk-management metrics.
• The demonstrated trade order is pending while the market is closed; the tutorial does not show a completed trade or live-market results.

Disclosure

Ray says he is not a financial advisor and that the video is not financial advice. He frames the walkthrough as an AI use case and recommends testing strategies with paper trading before using real money.

Presenter and sources: Ray Fu (Ray); Claude Code; Alpaca; Yahoo Finance.