Stock index prediction has been a challenging problem due to difficult to model complexities of the stock market. More recently deep learning approaches have become an important method in modelling complex relationships in time-series data. In this paper: we propose novel deep learning models that combine multiple pipelines of convolutional neural network and uni-directional or bi-directional gated recurrent units. Proposed models improve prediction performance and execution time upon previously published models on large scale S&P 500 dataset. We present several variations of multiple and single pipeline deep learning models based on different CNN kernel sizes and number of GRU units.

Predicting variations in stock price index has been an important application area of machine learning research. Due to the non-linear and complex nature of the stock market making predictions on stock price index is a challenging and non-trivial task. Deep learning approaches have become an important method in modeling complex relationships in temporal data. In this paper: (i) we propose a novel deep learning model that combines multiple pipelines of convolutional neural network and bi-directional long short term memory units. (ii) Proposed model improves prediction performance by 9% upon single pipeline deep learning model and by over a factor of six upon support vector machine regressor model on S&P 500 grand challenge dataset. (iii) We illustrate the improvement in prediction accuracy while minimizing the effects of overfitting by presenting several variations of multiple and single pipeline deep learning models based on different CNN kernel sizes and number of bi-directional LSTM units.