Akademik Veri Yönetim Sistemi
Journal of the Faculty of Engineering and Architecture of Gazi University, cilt.41, sa.1, ss.79-91, 2026 (SCI-Expanded, Scopus, TRDizin)
This study is a pioneering research that analyzes the impact of planetary movements, aspects, retrogrades, lunar phases, and eclipses on financial markets based on scientific principles. It tests the hypothesis that financial markets can be influenced not only by economic indicators but also by planetary dynamics using deep learning models. For the first time, planetary positions have been systematically integrated into financial market prediction models to create a large-scale dataset. Planetary parameters obtained from the NASA JPL Horizons database were combined with financial price movements (opening, closing, highest, lowest, and adjusted closing prices), the strongest features were identified using Mutual Information (MI) and Recursive Feature Elimination (RFE), and data dimensionality was optimized using PCA. Advanced deep learning models such as LSTM, GRU, ANN, and RNN were trained to predict financial market movements, the statistical significance of planets on markets was evaluated using the Granger Causality Test, and the most significant features were provided as input to the model. The results demonstrate that planetary movements can enhance market prediction power. LSTM and GRU models achieved accuracy rates of 98.7+ for the NYSE and 96.8+ for Bitcoin and Russell, with the integration of planetary parameters increasing prediction accuracy by 10. Granger testing confirmed statistical relationships between specific planetary alignments and financial market reversals. This research offers an alternative perspective on investment and risk management strategies by demonstrating the sensitivity of financial markets to planetary movements using scientific data.