The scale of import and export of service trade is closely related to the transformation and upgrading of economy, and the accurate prediction of it can realize the effective grasp of the overall situation of service trade. However, how to realize accurate prediction under the condition of `little data and poor information' has become a difficult problem in the research at domestic and abroad. Based on the data of import and export of Chinese service trade from 2005 to 2018, a grey Markov model is used to simulate and to forecast the import and export scale and industry scale of Chinese service trade from 2019 to 2021. The prediction results show that the grey Markov model has good applicability and high prediction accuracy. At the same time, ${\rm GM}(1,1)$ model and exponential smoothing method are used to simulate and to forecast the same set of data. After comparing the prediction results of the three methods, it is found that the grey Markov model has more advantages in the accuracy and applicability of prediction. In order to further improve the prediction accuracy of the grey Markov model, this paper proposes two kinds of improved methods based on data accumulation and data rolling. The data results show that the two kinds of improved methods can further improve the prediction accuracy of the grey Markov model. This paper provides an effective solution for the forecast of import and export scale of service trade under the condition of `little data and poor information'.