Cricket Predictions and Luck
Lately, all the cricket action has kept me glued to some good reads and analysis by veterans and analysts of the sport. It almost feels as if we are privy to the outcomes of a future that is yet to unfold. Shane Watson in his prediction for the ongoing Asia Cup, picked India as the most probable winner after the teams thrived under coach Rahul Dravid and skipper Rohit Sharma. In an interview, he said, “My predicted winner is India. They’re so strong and depending on what the conditions are, they adapt easily”. But as they say, predictions don’t always work!
I always marvel at how technology has abstractly applied the Experience and Reflective Learning theory by John Dewey, where he believed that through experience man learns about the world and only by the use of his experience man can maintain and improve himself in the world. Artificial Learning and Data Science have increased prediction accuracy in the game of cricket. However, each player’s performance impacts the performance of the team sometimes in unpredictable ways. For example, Michael Hussey single-handedly led the Aussies to victory against Pakistan in the semis of ICC World Twenty 20 in 2010. In yesteryears, a player’s performance depended on a few factors like his form, performance against the opponents, and the advantage of the pitch. The team management, coach and captain analysed the statistics and would select the best XI for a match.
With the advent of AI and ML-based algorithms merged into cricket, I feel it began doing cooler things, almost like fortune telling, but backed with data and analysis. The entire game process and smart strategies, even more precisely with deep insights, the vantage point of a team in a particular venue against a certain opponent increasing the chance of winning the game. Interestingly, Microsoft has lent its technology for detailed information during a match with the help of AI and the internet of things (IoT). Weighted Association Rule Mining (WARM) algorithm was used by the Indian Cricket team for analysing its performance against Sri Lanka and South Africa. [LINK]
Interestingly, in the year 1996, bowler Anil Kumble created a software package for data analysis and became the first to introduce a digital system in strategic planning. Indian Institute of Technology, Madras in collaboration with ESPN Cricinfo leveraged an AI deeper insight algorithm on players’ performance and matches that were unavailable earlier. The stats gave a clear perspective of a player’s performance based on factors like pitch quality, opponents’ current performance and the pressure on the team in the current scenario. Another compelling factor that gives a clearer picture was taking into account spontaneous factors such as the impact of toss, umpiring errors, no-balls, wides, and dropped catches that are often ignored during the assessment of a player’s performance. Sachin Tendulkar’s performance in the semis of 2011 would be inclusive of the fact that not only did he score 85 runs but was dropped by Pakistani fielders four times.
I pondered on how accurate the game predictions could be and smiled at the thought that in yesteryears it would clearly depend on the fact where my mum sat during the match, hilariously every time she went to the kitchen, India would lose a wicket till the time she was forbidden to step out of the living room. However, machine learning largely depends on humans for the data that is fed in.
The overall data depends qualitatively and would predict the exact fate of the match, especially how a batsman faces a particular bowler or a fielder’s natural reflex kicks in during the match that curbs the batman. The algorithm helps in predicting a few standard factors such as the luck index, run potential, and forecaster metrics (interestingly matches the data to the sequence of events with the real-time events to predict a score.)
The Smart Stat Metrics include the pressure indices, sudden wickets, and player quality index. Almost like a report card, it predicts the best batsman or bowler during the course of the game, especially during the power play or death overs.
The impact of AI and ML has been huge on cricket as well, data science plays out various scenarios to evaluate the events in general, giving it a common baseline. However, luck that influences superstition is a factor that still remains beyond the control or analysis of a human or an algorithm, such was the case of the Zimbabwean duo Grant Flower and Mark Dekker, who would go to bat together and one would say, “I hope you get a hit on the head” with the other replying “same to you”.
Every time Sachin Tendulkar went to bat he always put his left pad first, and Shane Waugh always kept a red rag in his pocket while batting after he had scored a century with the rag in his pocket in 1993. We are in a realm of transition where men build machines for relevance, acquisition, and accuracy. Despite the fact that trepidation is what keeps the match exciting, however, technology is gradually seeping into cricket just as in any other sport that will toss cricket for a head every time it is played.
