Cricket, a sport steeped in history with a well-established governing body and thriving economy, places immense importance on selecting winning team combinations comprising batsmen, bowlers, and all-rounders. The traditional selection process is meticulous but often biased. This study proposes a data-driven approach utilizing historic match data under complex dynamics for optimal team selection in cricket. Secondary data from Sri Lankan One Day International matches since 2009 was feature-engineered, including derived pitch conditions from historic pitch report data. A neural network with 7 dense layers, using 7 input features and classifying into three outputs, achieved a 76% performance under an 80:20 split on training and testing data for 100 epochs. Additionally, three Fuzzy Inference Systems were developed for player rating based on historic performances, achieving accuracies of 75%, 67%, and 62% for Batting Performance FIS, Bowling Performance FIS, and All-rounder Performance FIS, respectively. Despite inheriting a data imbalance problem and unmeasurable attributes such as psychological and physiological aspects, the study's deliverables serve as decision support models for cricket team selection. Future work should focus on empirical methodologies to enhance the performance of neural network models and Fuzzy Logic Inference Systems, ensuring a more adaptable squad selection model in cricket.