Convolutional neural networks (CNNs), inspired by the human brain's ability to recognize visual patterns, excel in tasks like object detection, facial recognition, and image classification, making them powerful tools for extracting insights from visual data. However, we are traders, so a natural question arises: Can we use that in trading? A recent paper shows that we can actually do it. Utilizing CNNs, Niklas Paluszkiewicz introduces a novel approach to pairs trading by visually analyzing historical price movements while converting traditional time series data into image representations.
The post Trading Arbitrage Portfolios Based on Image Representations appeared first on QuantPedia.