Optimizing Stock Price Prediction Based on Triangular Topology Aggregation Optimizer Using Long Short-term Memory Network
Year: 2025
Type: article
Abstract: Stock price prediction has always been a complex and challenging task in the financial field. This article proposes a novel method to optimize long short-term memory networks (LSTM) through a triangular topology aggregation optimizer, in order to improve the accuracy of stock price prediction. This method combines deep learning and advanced optimization techniques, aiming to provide more effective support for investment decisions in the stock market. We first introduced an optimized LSTM model and used a dataset for predicting stock prices. By observing the changes in loss between the training and validation sets, we found that the loss value of the training set gradually decreased from 0.04 to below 0.005 and approached convergence. Meanwhile, the loss value of the validation set remained below 0.005. This indicates that the performance of this model in stock price prediction is quite outstanding during both the training and validation phases. After analyzing the test set, the results showed that the predicted stock prices of this model were very close to the actual values, and could accurately predict market trends both numerically and trendwise. In addition, the MSE (mean square error) results obtained through the evaluation indicators of the model show that the MSE of the training set is 1.938, the MSE of the testing set is 1.944, and the RMSE (root mean square error) of the training set is 1.392, and the RMSE of the validation set is 1.394. These results indicate that the evaluation metrics of the training set and the test set are not significantly different, further proving that the model has strong generalization ability and can continue to demonstrate good predictive performance on new datasets. In summary, this article proposes an effective stock price prediction method by combining a triangular topology aggregation optimizer with an LSTM model, and verifies its efficiency and practicality in the training, validation, and testing stages. This type of research not only provides new ideas for analyzing the stock market, but also provides strong support for investors to make more reasonable decisions in a dynamic market environment.
Author: Yulin Jin
Institution: University of California, Berkeley
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4407827880