Empowering Financial Technical Analysis using Computer Vision Techniques

Technical analysis has received much debate on its values in the financial industry in the past decades, partially due to its subjective nature of extracting and encoding patterns from the candlestick charts. Although financial professionals manually work with charting patterns during their daily trading activity, technical analysis using computer vision and deep learning models has emerged only recently. We propose a solution that addresses two research objectives of technical analysis: 1) predicting bullish/bearish; 2) predicting technical analysis patterns using computer vision techniques. We show that our approach provides a promising avenue to identify technical analysis patterns accurately, which may shed light on automating technical analysis in the financial industry.