The problem of soil movements and associated landslides is common in the areas of Himachal Pradesh State in India due to the hilly terrain. Prediction of soil movements ahead of time may help save lives and infrastructure. Prior research has used machine learning models to predict soil movements but a comparison of different models for soil movement predictions is less explored. Here, we compared various machine learning models like persistence, auto-regression (AR), long-short term memory (LSTM), and multi-layer perceptron (MLP) in their ability of forecasting soil movements on a landslide location. We used data of soil movements collected by a low-cost landslide monitoring system installed at Gharpa landslide in Himachal Pradesh, India. Persistence, AR, MLP, and LSTM models were evaluated to predict downward soil movements along the Gharpa Hill. Root mean squared error (RMSE) metric was used for model evaluation on a 70% training and 30% test data split. Results revealed that the AR and persistence models gives best and second-best results followed by the LSTM and MLP models, respectively. We highlight the implications of our results for time-series forecasting of soil movements in the real world.