In this paper, two machine learning algorithms including multiple linear regression and artificial neural network are employed as forecasting models for the banking stock of Bangkok Bank Public Company Limited, Thailand. Five predictors including the SET50 Index, Barrick Gold stock price, exchange rate of US dollar, Down Jones Industrial Average, and crude oil price from West Texas Intermediate are taken into account. The historical time series for response and predictors were collected from 238 days from January 1 to December 31, 2020. The sizes of training and test datasets are 162 and 76 (70%: 30%). At the significance level of 0.01, the correlation between the response and each predictor is significantly found. Then, the datasets are recruited into the multiple linear regression and neural network models. Only three predictors including SET50 Index, Barrick Gold Stock, and Down Jones are not removed from the multiple linear regression model whereas all five predictors can be added up into the neural network model. Measuring the model performance by root mean square error found that the neural network is much better than the multiple linear regression model, in which the root mean square error are 0.03874 and 2.53624, respectively. As a result, this paper claims that the signals from SET50 Index, Barrick Gold Stock, and Down Jones are important for making a decision in trading on the stock of Bangkok Bank.