In many real-world scenarios, decision making is crucial especially under uncertainty. Such environments require appropriate models to evaluate the data effectively. One such task is assessing how closely stakeholder's opinions or judgements align with the ideal or actual values. To address this, we suggest a prototype using soft similarity measures between two soft sets for making decisions. We define various soft similarity measures in relation to a reference soft set (ℱ, P)and analyze properties of these measures like monotonicity, sub additivity and additivity. We introduce concepts such as Pseudo signed soft similarity measures, Pseudo positive soft sets, Pseudo negative soft sets and soft integral. We applied this framework to the problem of selecting a reliable stock market analyst by comparing analyst's predictions with actual data across multiple trails. The result showed that the method effectively identified the most consistent analyst. This approach offers a flexible and functional procedure for making decisions in the contexts of uncertainty where human judgement is the key factor.