Email phishing is one of the most prolific types of targeted adversarial activities against users. According to the FBI in a 2022 Internet Crime Complaint release, there were 323,972 phishing attempts reported. However, these complaints are eclipsed by the 3.4 billion of reported phishing emails per day, totaling over a trillion a year. Although phishing emails are a familiar tool for cyber adversaries, they have become more sophisticated over the years. Using machine learning algorithms to detect phishing emails has been researched for the last few years; however, feature selection to classify phishing emails resulting in acceptable performance metrics is still an active research area. This paper discusses a novel approach to select features and evaluate their effectiveness on various classification algorithms to detect phishing. The findings from this research provided a nuanced proof of concept for a novel feature selection method that provides efficient performance.