Due to the noisy nature of social media content, and the rapid propagation of falserninformation, the identification, and detection of fake news become a challenging problem.rnIn recent years, several studies propose to use text representation techniques from contentbasedrnapproaches to automatically detect fake news on the social media. However, fakernnews has a distinct writing pattern, and attempting to capture its distinguishing featuresrnmay help us improve detection rather than focusing solely on text representation. In thisrnstudy, we propose to combine the stance-based features (page score, headline to articlernsimilarity, and headline to headline similarities) with lexicon-based features from textrnrepresentation techniques to enhance the detection performance. To build the detectionrnmodel, we used three machine learning algorithms: Logistic regression, Passive Aggressivernand Decision tree. The proposed approach is evaluated using a newly collected Amharicrnfake news dataset from Facebook. Our experiment results show that the hybrid featuresrn(lexicon-stance) are capable of improving the previous lexicon-based detection results byrn4.1% accuracy, 3% precision, 4% recall, and 4% F1-score. In addition the hybrid featurernimproves the area under curve from 0.982 to 0.995 by reducing the false positive rate byrn4% and improved the true positive rate by 4.4%. Furthermore, we found that page score,rnout of the proposed stance features included, has contributed the most to the improvementrnof lexicon-based fake news detection.