Machine learning for trading: two halves of the data agree with a correlation of 0.71. Split them so that one price move cannot land in both halves, and it drops to 0.01. The trap of overlapping signals, on 40,955 real signals.

📚 Free courses with full code and data for subscribers of my Telegram channel Data Lab: https://t.me/lavrinenko_data

🎓 The full course on Udemy, “AI Forex Trading Bot with Machine Learning, Python & MT5”: https://www.udemy.com/course/build-a-machine-learning-trading-bot-in-python/?referralCode=59ABEDBF83EA5E7A1812
📘 My other courses on Udemy:
• Forex Algorithmic Trading: Backtesting Strategies in Python: https://www.udemy.com/course/test-it-before-you-trade-it-backtests-without-curve-fitting/?referralCode=AAFD579BEC179990057A
• Forex Risk Management & Position Sizing for Trading Robots: https://www.udemy.com/course/forex-risk-management-position-sizing-for-trading-robots/?referralCode=90AEC80D7222F92C6E93

Chapters:
0: 00 Intro: 0.71 that drops to 0.01
0: 12 The split-half check
0: 53 A convincing picture
1: 19 Signals come in ladders
1: 48 Each half gets copies of the same moves
2: 19 One signal per move: 0.01
2: 50 The same trap in significance tests
3: 26 Rows are not independent
3: 47 Split by time, not by row
4: 17 What to test next

I’m Vasyl Lavrinenko. For more than 20 years I’ve been building trading systems for real investors, and my Expert Advisors have won awards at international trading competitions. On this channel I test trading ideas on real market data and show what actually works, and what doesn’t.

💬 Which idea should I test next? Tell me in the comments.
🔔 Subscribe and turn on notifications, so you don’t miss the next analysis.

Topics: machine learning trading bot in Python, AI trading bot for forex, algorithmic trading with Python, MT5 expert advisor programming, Python trading strategy, machine learning algo trading.
#tradingbot #machinelearning #forex

Educational content only, not financial advice. Trading involves a risk of loss.