Making data the 12th Man in Cricket
Cricket – played by 10 full member nations, 96 part-time nations, for 160 years, across all 4 seasons of the year, with 540,290 matches of ball-by-ball information, at 11,960 grounds in the world, and 531,253 detailed player statistics – is surely a game that has all permutations and combinations of data to predict to the niftiest accuracy the exact outcome of the next ball that might be bowled. Or not?
*This post was initially published in ABC Tech and Games on Mar 26, 2015*
Cricket – played by 10 full member nations, 96 part-time nations, for 160 years, across all 4 seasons of the year, with 540,290 matches of ball-by-ball information, at 11,960 grounds in the world, and 531,253 detailed player statistics – is surely a game that has all permutations and combinations of data to predict to the niftiest accuracy the exact outcome of the next ball that might be bowled. Or not?
With so much data to back itself upon and some of the most mathematically inclined people as its largest fan base, cricket still is a novice in the creation and consumption of analytics as a game changer. Of course, all of us have seen various roll-ups and slices/dices of data, along with best-for statistics for different batsmen and bowlers throughout the game – but never a ‘successful adoption’ of analytics to the functioning and decision making of this great game.
There have been anathemas to this very idea of applying analytics to the game from respected sports journalists. Why does analytics seem impractical and sound theoretical in the world of cricket? What are the top 3 things that can change the creation and consumption of analytics for cricket?
1. More, better and totally different data – a new cricketing lens
It’d not be an exaggeration to claim that the 160 years of cricketing scorecards could be easily compressed into one 16 GB pen drive. What the game of cricket and the coaches, trainers and the players themselves need are far more meticulously captured data such as the ball variation for every single ball bowled, the bat deflection and bat-ball combination for every single batting shot, the field set-up for every single over and every wicket – essentially a 10^6 increase in data captured.
Akin to the Pitch-f/x and Hit-f/x from the world of baseball, cricket should also start creating the treasure trove of bowling and batting minutiae from 3D data generated through the high-speed video and Doppler radars. At this level, the data would no longer be about bowler X to bowler Y for 2 runs on the offside, but, bowler X with a spin variation of 23% and a shorter pitch to a batsman with a bat deflection of 12% on the front face on the offside with a 4 fielder setting (2 on the deep, 1 in the middle and 1 on silly) scored 2 runs in the deep – this kind of data is very powerful!
At the end of the day, coaching and playing decisions are not going to be made at just a player level – so even if the same batsman had faced the same bowler in 500 other combinations – what matters the most is the idea that a specific combination of bowling techniques at specific overs of the game for certain batsmen have a “greater” probability of producing a wicket.
2. Cross-pollination of ideas – Adopting problem-solving approaches from other areas
The world of cricket can benefit greatly by adopting the learnings and best practices of other sports groups and even industries – we call this the ‘cross-pollination of ideas.’ Instead of focusing on predicting the results of a specific batsman-bowler matchup, the game of cricket needs to transcend to the “bigger” problems. A case in point is the Netflix approach. Netflix is one of the leading pioneers at tracking customers’ viewing habits and preferences, and based on their taste buds, pair them up with the most suitable recommendation of movies/videos. This is done by breaking down user watching habits into constituent parts that could be analysed and correlated with their movie preferences – a mathematical construct called behavioural modeling.
The choice of such an approach of problem solving might not exactly help how much one batsman would score against one specific bowler, but rather help understand how a specific batsman performs against a certain bowler’s “attributes.” This behavioural deep dive into the batsman and bowler psyche helps understand the weak and strong zones for the batsman and also easily enable in identifying a smart actionable fix to plug the weak spots.
We have already started seeing this application in the current ICC world cup, where a batsman’s strong and weak zones against specific types of bowlers are discussed. The need of the hour is to identify other such generic problems to address – fielder strong and weak zones, bowler’s wicket and run-bleeding zones, to name a few, and make the call for the right kind of data to support these new analytical endeavors.
3. Adopt it and show it some love
Last, but certainly not the least, is the widespread usage and application of these analytical exercises in actual international games and local domestic fixtures. The urban legend is that the betters are some of the earliest adopters of statistics – after de Moivre disserted on the “Doctrine of Chances” to help gamblers analyze and predict their games of chance. There definitely is a huge market for key cricket stakes in the betting world and it would be heartening to see someone design the new “Doctrine of Sixers” to make a game book for cricket betters.
Cricket is a beautiful game and at its core, a by-product of talent, hard work and interminable luck. There is going to be a lot of resistance in the dressing rooms and selection boards over the extreme datafication of an age-old, well-functioning game. But if cricket has to move on to the next level and become a truly global sport, a well-dashed application of the science behind the game is needed – which can be leveraged by one and all, to eventually make the game far more competitive and truly enjoyable. Making data the 12th man in cricket would be an ideal first step.
Conclusion
Analytics is not a crystal-ball, nor is it a magic sorting hat to decide a world-cup winner based on just historical data. What analytics can do for cricket – is help coaches identify more areas for improvement, help betters make smarter betting decisions and help the players make the game more competitive. If analytics can help in saving the world by making cities safer from terrorism, predicting crime rate by zip codes and designing better disease response strategies – it surely can help cricket transform into a truly global sport!