Machine and deep learning-based models are the emerging approaches in handling time series prediction problems. These methods have been shown to bring out more accurate predictive results than conventional regression-based models. It has been reported that Recurrent Neural Network (RNN) with memory, such as Long Short-Term Memory (LSTM) are superior compared to Autoregressive Integrated Moving Average (ARIMA) for the purpose of memorizing longer sequences of input data. In recent years, a deep neural network, namely one-dimensional convolutional neural network (1D-CNN) has been explored for time series forecasting in several application fields. There have been a few research works which compared the performance of 1D-CNN and LSTM in some specific application areas, such as environmental time series or hydrological time series. The research question of interest is then whether 1D-CNN outperforms LSTM in forecasting financial time series, a very important field of time series analysis. Experimental results on six benchmark financial time series datasets have demonstrated that 1D-CNN is more predictively accurate and computationally inexpensive compared to LSTM-based models.