Probabilistic Approach To Predict The Bowlers For The First Two And Last Two Overs In T-20 International Cricket Using Machine Learning Algorithms
Year: 2026
Type: article
Abstract: This paper seeks to develop models based on artificial neural networks to forecast the performance of the cricket bowlers during critical moments of the matches, i.e., the initial two and final two overs. The study examined the decisions made by captains and coaches to select the bowlers regarding the metrics such as Bowling Strike Rate, Bowling Economy, and Bowling Average using data obtained by 167 T-20 international men cricketing bowlers in the period 2021-2023. Bowlers were ranked as being either Selected, Recommended or Dropped based on their performance by splitting the data into separate sets of these key overs and applying heuristic rules. The research showed that advanced machine learning techniques are effective and can enhance the process of decision-making and team make-up optimization in cricket and that the prediction of bowler classifications is highly accurate in many repetitions of the artificial neural networks. Moreover, the study revealed the significance of some of the independent variables in predicting bowler performance, which can help to refine the selection strategies. In sum, the paper highlights the possibilities of advanced analytics to optimize performance in cricket.
Institution: University of Peshawar
Language: English
Subfield: Economics and Econometrics
Domain: Social Sciences
Keywords: cricket bowling, artificial neural networks
Open Access status: diamond
OpenAlex ID: W7220002039
