The composition of a defined set of subunits (nucleotides, amino acids) is one of the key features of biological sequences. Compositional biases are local shifts in amino acid or nucleotide frequencies that can occur as an adaptation of an organism to an extreme ecological niche, or as the signature of a specific function or localization of the corresponding protein. The calculation of probability is a method for annotating compositional bias and providing accurate detection of biased subsequences. Here, we present a Sequence Analysis based on the Ranking of Probabilities (SARP), a novel algorithm for the annotation of compositional biases based on ranking subsequences by their probabilities. SARP provides the same accuracy as the previously published Lower Probability Subsequences (LPS) algorithm but performs at an approximately 230-fold faster rate. It can be recommended for use when working with large datasets to reduce the time and resources required.