Dynamic cricket match outcome prediction

To propose a model where match outcome is predicted ball by at the start of second inning. Our methodology not only incorporates dynamically updating game context as progresses, but also includes relative strength between two teams playing match. We used 692 matches from all seasons (2008–2018) to train our model, and we 59 current season (2019) test its performance. Here have engineered 11 players 10 bowlers, their metrics are tracked function each over throughout during inning, keeping in consideration changing target score one attributes. Initially, tried Logistic Regression, Naive Bayes, K-Nearest Neighbour (KNN), Support Vector Machine, Decision Tree, Random Forest, Boosting, Bagging, Gradient Boosting with an accuracy 76.47%(+/–3.77%). With deep learning, various flavours LSTM GRU like vanilla, Bidirectional stacked models results found very impressive 76.13%(+/–2.59%). All these flavors were tested using approaches such one-to-one sequencing, one-to-many many-to-one many-to-many which discussed this paper. An accurate prediction how many runs batsman likely wickets bowler take will help team management select best for