Franchise cricket leagues rely heavily on effective player selection at auctions, but human bias too frequently leads to poor performance and missed talent. Using a subjective judgment and a historical pricing model to determine winning teams at auction has led to much greater financial inefficiencies, poor franchise team composition around the globe, and undervalued high-potential emerging players with little to no visibility. This research presents a machine learning based Player Selection Assistant that uses international cricket performance statistics to predict suitability and auction value on a regional basis. The research evaluated four models: XGBoost, Random Forest, Decision Tree and Artificial Neural Networks using regional performance metrics which include batting strike rates, Bowling economy and fielding. The results show that ensemble methods provided superior performance over traditional approaches, with XGBoost (RMSE: 3.458, R²: 0.846) and Random Forest (RMSE: 3.475, R²: 0.870) outperforming other methods. The framework has regional ability suitability scoring/indices as well as a pressure index to provide consistent and transparent data-driven recommendations to reduce human bias and improve player selection performance model for auction decisions in franchise cricket leagues.