Akademik Veri Yönetim Sistemi
Black Sea Journal of Engineering and Science, cilt.9, sa.1, ss.362-368, 2026 (TRDizin)
This study investigates the effectiveness of specific neural network architectures for predicting stock prices using data from the BIST30 index. Recognizing the inadequacy of traditional models in handling the volatile and interconnected nature of financial markets, our study introduces a hybrid deep learning model. We evaluate the performance of a Graph Convolutional Network (GCN) compared to a hybrid model that integrates both GCN and Long Short-Term Memory (LSTM) layers. This hybrid approach uniquely captures the crucial relational dependencies among BIST30 stocks (via GCNs) alongside the sequential temporal patterns in their price movements (via LSTMs). Our methodology involves constructing a stock graph for GCN-based feature extraction, while the LSTM component processes individual stock time series. The outputs from these processing streams are integrated to produce final predictions, incorporating additional external factors like dollar and gold parity. Our findings indicate that the hybrid model, by leveraging the structural insights of GCNs and the temporal memory of LSTMs, exhibits superior performance in capturing the complex dynamics within the stock market data compared to the standalone GCN approach.