Artificial Intelligence has leveraged the word embedding space extensively in Natural Language Processing (NLP) for effective text representation. This research explores a novel approach to document genre classification by examining the velocity in which a document orbits through the word embedding space. Using Word2Vec embeddings, we follow the movement of documents as they progress through this space and extract peak frequency representations using the Fast Fourier Transform (FFT), which are then used for machine learning classification of genre. Experiments on book summaries spanning ten genres employ three machine learning models: Logistic Regression, Random Forest and Multi-Layer Perceptron (MLP) classifier. Results show that Logistic Regression failed to capture the document genre, while Random Forest and MLP achieved significantly higher accuracy. These findings demonstrate the viability using the word embedding space topology as a new representation of a document.