Χρήση τεχνικών μηχανικής μάθησης στις αυτοματοποιημένες συναλλαγές : μια ανασκόπηση
Machine learning techniques in automated trading : a review
This project is an attempt to review some machine learning applications in the field of the stock market. Six such studies were taken under consideration, each of which utilizes a large amount of market data, such as the limit order books of stocks with small time steps, as training data in Machine Learning systems, aiming to predict stock exchange opportunities in the short future. Various machine learning techniques and architectures are being tested including Multi – Layer Perceptrons (MLP), Support Vector Machines (SVM), Recurrent Neural Networks (RNN), Long Short – Term Memory RNNs (LSTM), Convolutional Neural Networks (CNN) or combinations of the above. Based on the studies’ results, the CNN architecture seems to be a more appropriate solution on the given problem, due to the dynamic nature of the stock market environment.