Artificial Intelligence and Machine Learning in Cricket

5 min readDec 31, 2020

Artificial intelligence, data science, and machine learning are the rapidly growing fields in the Information Technology field. The world is slowly conquered by computers in the form of above three. IoT is everywhere nowadays even in someone’s bedroom. Technology is rapidly growing in that phase, which also reduces the human work in day to day life.

One major field of application of technology is sports. The scope of technology is bigger in sports. In particular, if we take the game cricket, the use of technology is bigger. In the last two decades, we are able to see the advancement of technology in the game. Different camera angles, live streaming, live commentary all came into the game due to the growth of technology.

The Decision Review System is the biggest example to show, how technology is advanced in the game. In such circumstances, machine learning, artificial intelligence, and data science are also not an exception. In last two decades, data analytics have played a major role in the game. If we take the cricket before 2000s, there was not jobs called stat analyst or video analyst. But nowadays they are prime members in the support staff panels of each team.

In such extent, the field of Information Technology has an impact in the game. Let’s have a look at major introductions in the game from the fields, artificial intelligence, machine learning, and data science.

Duckworth-Lewis method

Duckworth-Lewis method is the well-known name for anyone familiar with the game. It is used to find the winners of game, in case of a rain-interruption or other interruptions to the game in the limited overs format. The Duckworth-Lewis method uses two main factors: the overs remaining, and the chasing side’s wickets lost. This type of analytics is used mainly for the team batting second. Earlier, it was calculated manually with lot of man effort. The development of data analytics and machine learning models, made it less complex in terms of calculation.

Lucky Index

The Luck Index is a measure of the magnitude of luck by factoring in a number of events, including toss, umpiring errors, no-ball out and drop catch. The main aspect of the algorithm is to score various lucky events in the match. Data science algorithms perform different scenarios to assess the impact of lucky events. The events like toss, no-balls, run-outs always have huge impact in the game.

Superstars

Superstars utilize ESPNCricinfo’s rich ball-by-ball database and the complex algorithms to combine Luck Index, Forecaster, and Smart stats and provide accurately predict the events of the cricket game. This model is not only about predicting events of the game, but also about identifying the capabilities of players. This is widely used by franchises in leagues like Indian Premier League (IPL). The model can be used to identify the amount a franchise can spent on a particular player based on his records, capabilities, and occurrences.

Power Bat

During the ICC Champions Trophy played in 2017, International Cricket Council (ICC) introduced bat sensors, which are placed in top of bat handles. Intel was the manufacturer of this technology, which was later modified with technology advancements and named as a Power Bat. It can be said as a next stage development in smart wearables. It helps to predict several factors like Quality of batsman shot, Speed of the ball, and Twist of the ball after hitting the bat or its edge. The data gathered is stored in a cloud and made available for prediction tables and graphs during Live Match broadcasts.

Sticker Sensor

The Power Bat is later evolved as Sticker Sensors. It is placed in the back blade of the bat, underneath a regular sponsorship sticker. The sensors can gather details of bat speed, bat rotation due to the impact of the ball, power, and quality of the shot. Its size is not more than a credit card. It was done by a Microsoft Team in India and it is made using Azure AI model. A trusted communication way is underuse that connects bat and edge machines with Azure, which has a Microsoft hardware platform. The data can be received in even mobile applications too.

Runs Potentials

It gives the final score based on the evaluation of several processes that occur during a game to determine the fate. The ML model takes all the factors like historical data of players, teams, and venues, pitch conditions, weather forecasts to predict scores in a game. This technique was developed by two New Zealanders, Scott Brooker, and Seamus Hogan. This technique takes into consideration the venue’s history, weather, pitch, as well as scoring rates to make predictions. The model is supporting bowlers how to swing a ball in a unique way that batsman fails to play.

Smart Stats

As stated above, the famous cricket website ESPNCricinfo posses a larger database of the game with stats from a long period of time. Even they have made it available for any one through ‘STATSGURU’. Recently they introduced a new system called ‘Smart Stats’, which is used for analyzing batting and bowling performances by taking into account the context in which they occurred.

Usually most of the cricket fans would have a sound knowledge about average, strike rate, economy rate of players and often players are evaluated using those numbers. But ESPNCricinfo developed an algorithm slightly different to the usual one. Here the numbers are not only decided by the actual numbers of that day, but by adding context of the game. Context includes pitch conditions, quality of opposition, and match situation — in terms of the pressure on the player. A new set of metrics are introduced namely smart runs, smart wickets, smart strike rate, smart economy rate, and batting/bowling/match impact.

It can be easily seen through an example they given for the list of Purple Cap holders in the history of Indian Premier League. There are some deviations between the actual purple cap holder and the one deserves through this algorithm.