Time domain (TD) features are frequently used for surface Electromyography signal (SEMG) pattern classification paradigm because they are computationally simple and easy to execute. In addition, TD features are extracted without the need for transformation directly from the raw data. Numerous TD features have been utilized as reported in the literature in the extraction of SEMG features. Some of these features, however, are not useful and make the feature vector redundant. Therefore, before the classification phase, it is essential to pick a subset of salient features from the initial function set. In this study, the combinations of two and three TD features were evaluated using an essential statistical method. This technique is; the RES index was called the relationship between Euclidean distance (ED) and Standard deviation (SD). While the ED measures the distance between two scatter groups (two joint ankle movements), the Standard (SD) measures the difference between each scatters group. This method was implemented for data from the shank muscles on the SEMG signals measured. The findings showed that the combination of the features of logWL and logSD outperforms the other combinations of the features chosen for TD. © 2020, Institute of Advanced Scientific Research, Inc. All rights reserved.