Predicting Stock Trends Based on Expert Recommendations Using GRU/LSTM Neural Networks

Predicting the future value of the stock is very difficult task, mostly because of a number of variables that need to be taken into account. This paper tackles problem of stock market predicting feasibility, especially when predictions are based only on a subset of available information, namely: financial experts' recommendations. Analysis was based on data and results from ISMIS 2017 Data Mining Competition. An original method was proposed and evaluated. Participants managed to perform substantially better than random guessing, but no participant outperformed baseline solution.