STOCK PREDICTION AND FORECASTING USING STACKED LSTM- RECURRENT NEUTRAL NETWORK MODEL

Abstract

With its high risk and high return, the stock market is drawing more and more people’s attention these days. A stock exchange market portrays savings and in vestments that are beneficial to the national economy’s effectiveness. Future equity returns can be predicted using publicly accessible information from the current and historical stock markets. Stock market trend prediction has also piqued the interest of statisticians and computer scientists, owing to the fact that it poses complex modelling challenges. There are methods or algorithms that can be used to forecast stock valuation with a high degree of accuracy. Because of the noise and uncertainties involved, observ ing and forecasting movements in the stock market price is difficult. A variety of factors, such as a country’s economic shift, commodity value, investor emotions, weather, political events, and so on, can affect the market value in a single day. Artificial intelligence and increased computing power have ushered in a new era in which programmable methods of predicting market prices have proven to be more accurate. A good forecast of a stock’s future price would yield a sizable profit. In this study, RNN and LSTM are combined to forecast the stock market’s movement. We have proposed a deep learning- based model to make prediction more reliable and simpler. The project focuses on the use of Long Short Term Memory algorithm (LSTM), which is an advanced form of Recurrent Neural Network. We checked the accuracy of our model using stacked LSTM and forecasted the future close prices of stock data through a backtesting method by using multi-layer LSTM networks. After performing the experiment, we were able to forecast the forthcoming 10 days closing price of the given stock.