JP Morgan And Major Quant Hedge Funds Are Using Computer Vision To Trade Stocks
Artificial Intelligence & Machine Learning News
Why Aren’t You Using It In Your Strategy Dr. Valeriy Manokhin wonders?
A quiet revolution is taking place inside the biggest quantitative trading floors. JP Morgan, top hedge funds, and advanced proprietary shops are no longer relying only on classic time series tools such as moving averages, lags, or volatility filters. They are now treating financial data as visual information. Computer vision, once reserved for self driving cars and medical imaging, has become a new engine of signal discovery on Wall Street.
One of the most powerful tools behind this shift is the Gramian Angular Field! (Gramian Angular Field Explained Clearly for Time Series Feature Transformation)
The GAF transforms a standard one dimensional time series into a dense visual texture. It does this by encoding the angular cosine similarity between every pair of time points. The result is a heatmap like image that captures structural relationships normally hidden in traditional indicators. A GAF highlights transitions, regime shifts, and recurring patterns with a clarity that standard numerical features rarely match.
These transformed images feed directly into models that excel at pattern recognition. Convolutional networks and vision transformers scan the texture for shapes, gradients, and anomalies. Trend changes appear like contour breaks. Cycles show up as repeating blocks. Volatility clusters take on distinctive visual mass. Sharp disruptions stand out in ways that allow models to classify markets more effectively than when operating on raw prices alone.
This move toward image based analysis reflects a broader evolution in quantitative finance. Firms seek representations that allow machine learning systems to explore the geometry of data. Computer vision provides a way to interpret markets not as narrow sequences, but as multidimensional structures full of information. This method often reveals edges inaccessible to traditional signal design.
The adoption of computer vision by major institutions raises a direct question for investors and researchers. If JP Morgan and leading quant firms now use image based time series analysis to gain an advantage, why would any serious investor ignore it?



