A Deep Learning Approach for Prediction of Stock Price Based on Neural Network Models: LSTM and GRU

In recent years, there has been a predominant use of Artificial Neural Networks (ANN) to solve real-world problems. One such conundrum is the prediction of stock prices since they are convoluted, sporadic, and dynamic. This study focuses on the use of RNN, the Recurrent Neural Networks to identify and forecast trends that stock prices will follow using based on historical data. The LSTM and GRU models are considered for developing prediction models of stock prices. These models do not have the vanishing gradient problem, unlike traditional RNN. Both models are compared and analyzed to find a model that would predict satisfactory prices with maximum accuracy which can help a trader to make efficient decisions.

I. Introduction

The stock price of a single share of a public traded company is the amount one is willing to pay to own it. It represents the company's value. The stocks are transferred and traded in the stock market. The stock market is a legal medium that accumulates a large amount of capital. It is considered as the reflection of a country's economic development. It is because stock price illuminates the demand and supply relations of the stock market. The share price has dynamic nature, and its changes depend on various factors. The factors that affect stock prices can be positive or negative. Some of them are the law of demand, production and management changes, and public perception about the company. The law of demand is considered as the main force that drives stock price. Apart from these, the economic, political, trader behavior such as individual aspects and other variegated conditions, their interactions over time inflates changes in the price of a stock. Since these stock values are convoluted, sporadic, and dynamic, forecasting stock prices are regarded as a conundrum. Despite the fact, that these stock values emitting such characteristics it attracts superfluous people to study and analyze the trends in the stock market. One of the main reasons is one successful price estimation helps in making efficient decisions regarding buying or selling of stocks, which in turn would make humongous profits or elude severe losses. Sometimes, traders and investors follow their experience and intuition to estimate stock prices, but it creates huge risks. Therefore, a profound analysis is required for the prediction of stock prices with accuracy.