Willison’s Amplitude (WAMP) is a quantitative feature used to interpret a wide range of biosignals. The WAMP has a user-defined parameter known as the Tolerance Parameter (TP) whose selection significantly affects the class separability provided by WAMP. Most state-of-the-art recommendations for selecting the TP of WAMP are tailored to the specific requirements of the applications. It is not advisable to optimize the parameters of a feature extraction algorithm using a loss function specific to the classifier. We use Reptile Search Optimization (RSO) to automate the selection of TP. A loss function/fitness function termed as Prediction Error in Percentage (PEP) that quantitatively reflects the class separability provided by the features is also introduced. Experimental results on both Electromyography (EMG) and Electrocardiogram (ECG) datasets show that the RSO offers lesser percentage prediction error than arithmetic optimization, simulated annealing, nutcracker optimization, political optimization, white shark optimization, particle swarm optimization, student psychology optimization, and artificial bee colony optimization. Automating the selection of TP with the help of RSO will help exclude the burden of exhaustive searches and improve the separability offered by the WAMP.