This study investigates the extent to which microfeatures -- such as basic text features, readability, cohesion, and lexical diversity based on specific word lists -- affect Chinese EFL writing quality. Data analysis was conducted using natural language processing, correlation analysis and stepwise multiple regression analysis on a corpus of 268 Chinese students' SAT writing in response to a single prompt. The results show that word count, the number of words per sentence, the connecting word frequency, the number of commas per sentence, the stop word frequency and the Coleman-Liau readability index (CLI) contribute significantly towards predicting SAT essay scores of Chinese EFL students. The regression model explains 62.6% of the variance in predicting the SAT essay's score where the number of commas per sentence acts as a suppressor. Surprisingly, none of the word lists were found to be a significant predictor. Consequently, EFL teachers can adopt these simple lexical features to formulate writing strategies for beginner level Chinese EFL students to improve their SAT writing quality. Moreover, developers can include these features as a personalized option in automated writing assistance systems targeted toward beginner level Chinese EFL students.