Tutorial shows building an AI trading bot on Claude Haiku 5.5 with TradingKit
Overview
A solo founder's tutorial, shared on X, walks through building an AI trading bot with Claude Haiku 5.5, Anthropic's cheapest model, using Claude Code and the TradingKit backtesting engine.
The steps backtest a BTCUSDT 1-hour strategy with fees, keep only results with at least 100 trades, drawdown under 20% and profit factor above 1.1, then verify the best candidate in TradingView and forward test it for three months. The author argues that a cheap model whose strategy passes these filters is preferable to an expensive model that does not. The post reports no performance results for the bot.
Written by AI from the articles below · updated Oct 10, 10:26 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
Davidd Tech@DaviddDotTechXTutorial: building an AI trading bot with Claude Haiku 5.5AIA solo founder's guide shows how to build an AI trading bot with Claude Haiku 5.5, Anthropic's cheapest model, using Claude Code and a backtesting engine from TradingKit. The steps cover backtesting a BTCUSDT 1-hour strategy with fees, filtering results for at least 100 trades, drawdown under 20%, and profit factor above 1.1, then verifying the best one in TradingView and forward testing for three months. The post argues that a cheap model that survives these filters beats an expensive one that does not.

Heat trend
Not enough continuous observations to show a trend yet.