This paper studies the problem of forecasting electricity prices in continuous short-term electricity markets, specifically focusing on the intraday volume-weighted average price of hourly products in the last three hours of trading. Two state-of-the-art recurrent neural network architectures, namely the Temporal Fusion Transformer and the DeepAR network, are compared against well-established statistical models, such as the Linear Regression-LR, ARX, and SARIMAX models, concerning their forecast accuracy. Historical electricity market and grid data from European Energy Exchanges were used to create a forecasting dataset and train and compare five different model structures stemming from traditional statistical methods or contemporary deep learning-based counterparts.