Prediction of the stock market index with the help of visibility graph

The dynamics of the market and the stochastic behavior of index time series worldwide have made forecasting trends a primary concern for researchers and investors in the capital market. Therefore, various models are used for forecasting. Time series analysis and prediction methods based on networks, including the visibility graph model, have proven effective for time series analysis. In this research, we aimed to predict the TEDPIX index for the years 2013-2024 using three visibility graph-based methods. These methods include the most similar node model, the weighted similar node model, and our proposed method, the cross-visibility graph model. We converted the time series of the index and some auxiliary variables into a visibility graph and predicted the aforementioned time series using these three models. The results showed that the performance of the weighted similar node model and the most similar node model varied depending on different conditions. However, our proposed model, the cross-visibility graph model, and the weighted similar node model generally provided the best results in predicting the index and were more accurate compared to other models.