A Stock Price Prediction Model Based on MLP

This paper proposes a stock price prediction model based on a multilayer perceptron (MLP) network, using NVDA (NVIDIA Corporation) stock as the research subject. First, the fundamental principles and structure of the MLP network are introduced, along with its advantages in regression tasks, particularly its ability to handle complex nonlinear relationships. Then, historical data of NVDA stock is preprocessed using Python and relevant libraries (such as Pandas, NumPy, Matplotlib, and PyTorch), including data normalization and the division of training and testing sets. This paper provides a descriptive summary of the construction of the MLP model and its parameter setup and optimization. In training, the loss function employed is mean squared error (MSE), and the parameter update rule is the Adam optimizer. Experiments are conducted to evaluate the model's efficiency by analyzing its prediction, and the actual vs. predicted prices are represented visually. The MLP model demonstrates effectiveness in NVDA stock price forecasting tests where it delivers precise price fluctuation pattern recognition. This paper investigates the model prediction accuracy and enhancement potential together with financial market prospects for the MLP model's stock price prediction capabilities.