Abstract: Computer-readable media, systems and methods may improve performance efficiency of generating custom position data of multi-leg transactions. For example, a system may receive a request to generate a custom multi-leg transaction involving variable ratios that may be evaluated against historical data derived from electronically traded commodities. Responsive to the request, the system may execute a computational workflow that applies rules to parse encoded futures symbols of the multi-leg transaction, build timeseries for historical years based on the input and the historical data and generate a GUI portion that may be transmitted to the user device via the interface. In some examples, the system may use a classifier to identify time intervals that are similar to one another for purposes of a given multi-leg transaction. The classifier may include a rule-based classifier that applies decisioning rules or a machine-learning (ML) classifier that is trained using training datasets.