Cricket Player Performance Forecasting

Winner · The Sledge Hack, Analytics Vidhya · September 2023

How many runs would each player score, and how many wickets would each player take in the India–Australia Cricket World Cup match on 8 October 2023? That was the forecasting challenge in Analytics Vidhya’s The Sledge Hack.

The competition provided ODI batting and bowling histories for the 30 players selected in the India and Australia squads. I used those historical performances to produce player-level predictions for two separate targets: runs scored and wickets taken.

My approach

I treated runs and wickets as separate time series forecasting problems. I experimented with SARIMA model configurations and combined selected forecasts using an ensemble approach. The final submission contained a predicted run total and wicket count for each player.

Evaluation and result

The final score was based on a weighted average of the root mean squared errors (RMSE) between predicted and actual runs and wickets after the match. My submission scored *14.3608 on the final private leaderboard, ranking **1st among 1,073 registered participants*.

The preliminary public leaderboard used a benchmark before the match was played. My submission ranked *14th* there, with a score of *8.8975*. The final private leaderboard determined the competition result.

Recognition

I won *first prize: ₹25,000 in cash and a GenAI program valued at ₹60,000*.

Explore the project

– [Project overview and source code]
– [Detailed approach]
– [Analysis notebook]