Twitter is one of the most popular microblogging social systems, which provides a set of distinctive posting services operating in real time manner. The flexibility in using these services has attracted unethical individuals, so-called ”spammers”, aiming at spreading malicious, phishing, and misleading information over the network. The spamming behavior results non-ignorable problems related to real-time search and user’s privacy. Although of Twitter’s community attempts in breaking up the spam phenomenon, researchers have dived far in fighting spammers through automating the detection process. To do so, they leverage the features concept combined with machine learning methods. However, the existing features are not effective enough to adapt the spammers’ tactics due to ease of manipulation, including the graph features which are not suitable for real-time filtering. In this paper, we introduce the design of novel features suited for real-time filtering. The features are distributed between robust statistical features considering explicitly the time of posting tweets and creation date of user’s account, and behavioral features which catch any potential posting behavior similarity between different instances (e.g. hash-tags) in the user’s tweets. The experimental results show that our new features are able to classify correctly the majority of spammers with an accuracy higher than 93% when using Random Forest learning algorithm, outperforming the accuracy of the state of features by about 6%.