Artificial Intelligence Approaches Applied To The Financial Forecasting Domain

Nowadays, Financial Markets represent a crucial part of the world economy. Financial Markets have grown exponentially in the last decades and they provide finance for companies that can be used to hire, invest and grow. On the other hand, Financial Markets provide a big opportunity for people to invest their money in shares or equities to build up money for their future. The possibility to increase one's capital through investment has led researchers in the last decade to focus their work on predicting the performance of the market by exploiting novel Machine Learning and Deep Learning tools and techniques. Several approaches have been proposed in the literature, ranging from time-series pattern recognition analysis to more complex approaches based on Machine Learning and Deep Learning. Following this trend, the main contribution of this dissertation is the proposal of three novel approaches to tackle these issues. Firstly, starting from scratch, we propose a fully automated optimized ensemble approach, where an optimized feature selection process has been combined with an automatic ensemble Machine Learning strategy. Secondly, we exploit a Deep Learning approach with an ensemble of CNNs, trained over Gramian Angular Fields (GAF) images, generated from time series related to the Standard \& Poor’s 500 index Future. More precisely, a multi-resolution imaging approach is used to feed each CNN, enabling the analysis of different time intervals for a single observation. Finally, we propose an improvement of the previous approach with a multi-layer and multi-ensemble stock trader. This method starts by pre-processing data with hundreds of Deep Neural Networks and then, a reward-based classifier is used to maximize profit and generate stock signals through different iterations. At the end of this dissertation, accompanying the current Machine Learning Interpretability trend, we propose a Visual Framework for in-depth analysis of the results obtained from Deep Learning approaches, tackling classification tasks within the financial domain and aiming at a better interpretation and explanation of the trained Deep Learning models. The proposed Framework offers a modular view, both general and targeted, of results data, providing several financial-specific metrics.