In Episode 7 our strategy failed the 3 year stress test — minus $43. Claude said the 50/200 SMA was too slow for M15. So we tried shorter SMAs — the 20/50 — that react faster to price changes.
The result? 970 trades. 39.6% win rate. Minus $689.
16 times worse than before.
Faster SMAs don’t mean better signals — they just mean more noise. The problem was never the SMA settings. The problem was the timeframe entirely.
This is Episode 8 of The Imperfect Algorithm — building an AI Forex trading bot from scratch using Claude AI. No coding experience. New episode every Sunday.
📺 Watch the full series: https://www.youtube.com/playlist?list=PLvv_2uZI7NLOCSiYgB7lWdOoQWPQDCl9R
📂 All code free on GitHub: https://github.com/theimperfectalgorithm/forex-bot
🔔 Subscribe to follow the build
forex bot python, SMA crossover strategy, AI trading bot 2026, algorithmic trading results, claude AI coding, MT5 python, build in public, EURUSD strategy, backtest results, moving average strategy

#ForexBot #AITrading #ClaudeAI #AlgorithmicTrading #BuildInPublic #ForexTrading #MT5Python #BacktestResults #MovingAverage #BeginnerTrading