Stock Price Prediction Based on VMD-MLP: A Research Study

Stock price prediction is one of the core issues in financial market research, as its accuracy is crucial for investment decision-making and risk management. However, due to the highly nonlinear and non-stationary nature of stock prices, traditional time series forecasting methods struggle to effectively model their complex fluctuation patterns. To address this issue, this paper proposes a hybrid forecasting model based on Variational Mode Decomposition (VMD) and Multi-Layer Perceptron (MLP), referred to as VMD-MLP. First, VMD is employed to perform multi-scale decomposition on the original stock price series, breaking it down into multiple modal components with distinct spectral characteristics. This process reduces data non-stationarity and enhances the distinguishability of key features. Subsequently, MLP is used to train and predict the decomposed features, and comparative experiments are conducted with various machine learning models without VMD processing to assess the impact of VMD preprocessing on prediction performance. Experimental results indicate that, compared to traditional methods, VMD preprocessing significantly improves the prediction model's accuracy, reduces noise interference, and enhances generalization capability. The VMD-MLP model outperforms benchmark models in terms of MAE, RMSE, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> metrics on the test set, demonstrating its effectiveness and applicability in stock price prediction tasks. This study provides new insights into the analysis of complex financial data.