Stock prediction and trading using long-short-term memory neural networks

Abstract

In this project, I applied the Long Short Term Memory (LSTM) model to predict the future stock price Zoetis (ZTS). I first developed an LSTM algorithm to predict the stock prices. Subsequently, the original model was then subjected to five variations to determine which variation of the model gives the most accurate result of the future stock price. To simplify the experiment, I have taken into consideration closing price of the Zoetis stock. From the experiment, it was found that the model with 300 epochs, 1 layer and 256 nodes in each layer has the best result , followed by the model with 100 epochs, 1 layer and 256 nodes each layer. The model with 250 epochs, 4 layers, 256 epochs and with dropouts has the worst performance.