Can analytics predict the winner of T20 matches?

Uncertainty of the game
Cricket is known for its uncertainty and this uncertainty increases as the number of overs reduces. This uncertainty and fast pace of game makes T20 more interesting and thrilling. We have seen quite a few upsets in this format. For instance, the opening match of T20 world cup 2009 England vs. Netherlands and what a match that was. Netherlands managed to beat the host country. One day champion Australia was not able to qualify for the super eight stages. This list is full of surprises and will keep growing.

Analytics in cricket
The relationship between analytics and cricket is old. Nearly 150 years ago Englishman and Sussex fast bowler, John Wisden, published his first annual Almanac and turned cricket statistics into a business. Now, Wisden is the Bible of cricket. New Zealand cricket has been using SAS, a statistical analysis tool, for a long time to define real time strategy.
So, can analytics predict the winner of the match in spite of the uncertain nature of the game?
Yes, analytics can predict the winner of the match with a certain degree of error. We have built a statistical model to predict the winner of the match after the first inning and till now it has correctly predicted the winner of match for 12 games out of 13, in the ICC T20 tournament, 2009. We are not considering the New Zealand vs. Scotland match of 7 overs, as it was too short. Now you are eager to see the result of match predictions. The wait is over. Here are the results.

Group Matches














Super Eight Matches

Overall prediction accuracy as on June 14, 2009= 17/19 = 89.5%
*The model can predict the winner after the first inning.

*This is a mathematical model and shouldn’t be used for any purpose.