Neuromorphic devices hold diverse potential applications like brain inspired computers, promising a high-performance arithmetic system with low power usage. Reservoir computing (RC), a type of recurrent neural network, achieves learning by adjusting weights between intermediate and output layers. Time-delay RC introduces a delay, creating virtual nodes to update a node within the middle layer. In this study, the time series prediction performance of the Ag/Ag2S nanoparticles device was tested by defining NARMA2 and memory capacity (MC) tasks using the Time-Delay RC system. The Ag/Ag2S device successfully achieved NARMA2 with an accuracy rate of 89.46%. The performance of the Ag/Ag2S device in NARMA2 was also obtained regarding the delay time between feedback in the time delay system.