Stock trend prediction analysis using deep learning

The stock market prediction problem has
historically been an arduous one, primarily due to
the excessive noise, non-linearity, and nonstationarity that exist within financial time series
(FTS) data. The application of deep learning for
stock market analysis has made it more common
to develop models based on data to predict future
stock market trends. Continuing this trend further
is the comprehensive evaluation of how well
modern deep learning techniques outperform
previous standard statistical and machine
learning approaches to predicting future stock
market price movements presented in this paper.
Based on past stock prices, technical indicators,
and potentially news & social media sentiment
information will serve as baseline data for this
project. Advanced approaches to modeling the
data are required in order to achieve the model
prediction performance necessary for decisionmaking. Appropriately designed networks such
as the Long Short-Term Memory (LSTM) neural
network along with other neural network types
(e.g., Gated Recurrent Unit (GRU) and
Convolutional Neural Networks) offer potential
for identifying local/temporal feature patterns
within the stock market.