Trading systems are tools to aid financial analysts in the investment process in companies. This process is highly complex because a big number of variables take part in it. Furthermore, huge sets of data must be taken into account to perform a grounded investment, making the process even more complicated. In this paper we present a real trading system that has been developed using semantic technologies. These cutting-edge technologies are very useful in this context because they enable the definition of schemes that can be used for storing financial information, which, in turn, can be easily accessed and queried. Additionally, the inference capabilities of the existing reasoning engines enable the generation of a set of rules supporting this investment analysis process.

Abstract. Trading and Stock Behavioral Analysis Systems require efficient Artificial Intelligence techniques for analyzing Large Financial Datasets (LFD) and have become in the current economic landscape a significant challenge for multi-disciplinary research. Particularly, Trading-oriented Decision Support Systems based on the Chartist or Technical Analysis Relative Strength Indicator (RSI) have been published and used worldwide. However, its combination with Neural Networks as a branch of computational intelligence which can outperform previous results remain a relevant approach which has not deserved enough attention. In this paper, we present the Chartist Analysis Platform for Trading (CAST, in short) platform, a proof-of-concept architecture and implementation of a Trading Decision Support System based on the RSI and FeedForward Neural Networks (FFNN). CAST provides a set of relatively more accurate financial decisions yielded by the combination of Artificial Intelligence techniques to the RSI calculation and a more precise and improved upshot obtained from feed-forward algorithms application to stock value datasets.

No abstract