Repository containing software for the detection of double top and head and shoulders patterns. In addition to the code used to generate the trained model capable of recognizing them, for which the Gramian Angular Field method was used to convert the time series of the market to images, then use these images to train a CNN model. We also add the code used to form the labeled data set, so that we have the labeled patterns to be used to train the network. Finally, the raw data of the main markets used over a 10-year period and the patterns that were labeled are included.
