Deep Extra Cover

Telling stories of Limited Overs Cricket

I have been working on this “project” since 2011. The idea is to visually tell the story of a game of cricket, filling in the details that you might miss out when you look at the scorecard. 

The basic feature is a line that shows the likelihood that the team batting first wins. The higher this line, the greater the chance that the team batting first will win the game. This way, you can look at how the game has swung between the teams, and understand the “story of the game”.

The model runs live, using data from the ESPNCricinfo API. The model to evaluate the likelihood of a team winning (based on the score) is similar to models used to price financial derivatives. You can also think of the model being broadly similar to the WASP

Check out the stories of recent T20 games here

And the recent ODIs here

The latest T20 game, as parsed by my algorithm. The higher the graph, the more the game is in favour of the team that batted first. Key periods of the game have been annotated with the boxes. The tables below show the contributions of each player to their team’s cause.

A video explaining the model


An example of match analysis using Deep Extra Cover