Implement and comparison various configurations of neural networks for analyzing financial markets
With the Machine Learning I did the reaserch of the use of neural networks in the field of forecasting prices and trends in financial markets. The process of developing a neural network began with the building of a simple network, implying architectures that have already been successfully used to solve such problems. It became a good starting point, after which I could try to change all the fixed parameters and extract maximum performance from the network.
In this research I explored a multilayer perceptron and a convolutional neural network. Neural networks were implemented and a comparative analysis of the results was carried out and showed that a convolutional neural network can predict the stock prices of financial markets a little better than a multilayer perceptron.
Based on the results obtained, we can observe the following: there are too many errors, when predicting data further than the time t+1. This means that neural networks can't predict the exact values of stock prices. But it is also clear that the trend lines go in the same direction, and here we can conclude that neural networks can follow the trend.