At its core, a math proxy is a stand-in. It relies on a fundamental assumption: that Object A, which is easy to measure, shares a strong correlation or structural similarity with Object B, which is difficult to measure.