Abstract
The stock market is a place to lay investments in companies to boost their growth. The stock market can play an important role in a nation's future. A good stock market of a country always produces a decent mindset for entrepreneurs in those countries. But the stock market is a very volatile place. The price fluctuates rapidly in a short moment. There is also some common misconception among small shareholders that big companies always have a good price. The stock price can be changed due to the company's profit or loss at that moment, but it is not only bound to that. The weather forecast, festivals, and international relations of countries also play an important role. However, this project is for general purposes, to predict stock in normal situations. Anyone can use the data to grasp the whole situation of a company for predicting near future. By stock prediction, govt. may also find irregular and suspicious stock fluctuation. To sell and buy stocks only help of stock prediction will be a very risky idea. But to find out some trends, prediction can help. Here, we have used time-series data to predict the next values. Normal deep learning models performs very well by learning complex time-shifted correlations between stepwise trends of a large number of noisy time series, using only the preceding time steps’ gradients as inputs. Thus, different models predict different results. Such correlations are present in stock prices, and these models can be used to predict changes in a price’s trend based on other stocks’ trend gradients of the previous time step. In more narrowly defined terms, this applied part is situated at the intersection of computational finance and financial econometrics. Combining and comparing two or more models can give us a good result. And combining it with random values may increase the fixed trends of a specific model. Thus, an average value and randomness can give us a better insight.