Stock price forecasting with deep learning techniques is gaining high importance in recent times. Deep learning techniques are efficient in capturing the intricacies of dynamic and nonlinear stock data. The popular deep learning techniques are applied to forecast stock prices. The performance of different deep learning techniques in stock forecasting is different. Hence, an investigation into deep learning skills in stock forecasting has become essential. Furthermore, exploration of modern research on ensembles as well as hybrid deep learning-based forecasting techniques is important. This paper presents a recent five-year review from 2020 to 2024 on stock market forecasting with deep learning techniques. This review work focuses on different deep learning models along with their performance and limitations on different stock databases. The outcomes of the review process explore the scope of opportunities for further research in deep learning-based stock forecasting.