Wave Net-TSRS Model for Time Series Prediction in Finance

In economies, it can be used as a very important money market instrument in terms of ensuring economic stability with central bank reserves and maintaining adequate liquidity. In recent years, researchers have focused on forecasting reverse fluctuation without influencing reserve factors. Interest, inflation, and exchange rate are used as finance economic variables in the estimation of economy reserves because those are influencing factors of reserves held in central banks that can cause their reserves to rise and fall. In the research, the monthly base data of China, Japan, South Korea, India, Russia, Indonesia, Saudi Arabia, and Turkey, 1.1.2000–11.01.2021, were discussed. In this study, our intention is to propose a simple artificial neural network topology named Wave Net-TSRS, which has potential for time series data. We don’t need to tune a collection of many parameters due to the automatic feature engineering of the proposed topology. Instead of that, each convolutional block of the topology was designed with gated activations, residual connections, and skip connections specifically. When compared to other existing topologies, the proposed algorithm with a designed specific topology has robust and reliable evaluation of reverse fluctuation and prediction of the future.