Deconstructing cricket May 22, 2018

How T20 went from being a bit of fun to downright futuristic

Now more than ever in cricket, on-field strategy is dictated by how teams make sense of all the data available to them

In the 2016 World T20 semi-final, Andre Russell was instructed, on the basis of data, to go after slower balls because of his poor record against them © IDI/Getty Images

"We're going to bowl first because I haven't got a clue what's going to happen."

It was a sultry night at The Oval on June 13, 2003, the night of the first ever games of professional T20 cricket. Like everyone else, Surrey's captain Adam Hollioake did not know what to do. "Like many, we took it as a bit of a joke to begin with," he later admitted. His team still won the maiden T20 competition.

Yet it would prove to be a seminal week for cricket and big data. First came the creation of T20. That same summer came the release of Moneyball, Michael Lewis' book on how the Oakland Athletics used data-mining to get an advantage in Major League Baseball.

Fifteen years on, cluelessness about how to approach T20 has been replaced by minute, systematic planning. A cocktail of unprecedented money in the sport, more fixtures, and the growing ascendancy of domestic T20 have elevated the importance of strategy and data.

"T20 is such a fast game that it needs snap thinking on the field. In Test cricket you come off at lunch, you talk to the players," explained Phil Simmons, who was head coach of West Indies when they won the 2016 World T20. "You have to be a lot more precise [in T20] because you have no time to make adjustments. In Test matches you assess by sessions; in one-day cricket you assess by overs; in T20 cricket now you assess by balls. Every ball is an event."

The first professional cricket team to believe that Moneyball's ideas could give them a decisive edge on the field were England during Peter Moores' first stint as head coach, from 2007 to the start of 2009. One of the more influential acts of Moores' first stint was his handing a copy of the book to Andy Flower, who became his successor. Under Flower, captain Andrew Strauss and team analyst Nathan Leamon, a former Cambridge maths graduate, England glimpsed the potency of statistical analysis in cricket.

Numbers helped England become No. 1 in Tests, and even, for a time, one-day internationals, as well as to lift the 2010 World T20, which remains their only global tournament victory in the men's game. Leamon used Monte Carlo simulations to map out probable outcomes in Test matches, playing computerised games with different sets of players and varied tactics to inform England's strategy.

His analytical work also went deeper than anyone else in cricket had ever gone. Leamon broke the pitch down into 20 blocks, of 100cm by 15cm each, and instructed which was the optimal block to target when bowling to each opposing batsman. Using wagon wheels to show where players scored their runs was already common; borrowing from baseball, Leamon took this further, and isolated where batsmen scored runs in particular phases of their innings.

Before England played India, then the No. 1 Test team, at home in 2011, Leamon ran the numbers on Sachin Tendulkar, who had scored eight centuries in his last 15 Tests. Befitting such an accomplished batsman, Tendulkar scored his runs evenly around the wicket. But at the start of his innings, he seldom scored on the off side. Using this information, England packed the off side at the start of Tendulkar's innings, and hung the ball outside the off stump, choking off his favoured scoring shots early in his innings. No team had previously discovered this slight chink in his armour; in the height of the English summer, he averaged only 34.12. England won 4-0 - and went to the top of the world Test rankings.

Nathan Leamon was among the first in cricket to see value in using data to influence team strategy © PA Photos

Leamon's strategies were effective in T20 too. His finding that left-arm seamers were more effective than right-armers informed England's controversial decision to pick Ryan Sidebottom ahead of James Anderson for the 2010 WT20. Sidebottom picked up ten wickets in the tournament, including wickets in his first two overs against Australia in the final; a few hours later, England lifted the trophy.

All of this foreshadowed how data could impact T20. In the early 2000s, as Australia were pushing the boundaries of the possible in cricket, their board enlisted a sports-analytics company to study cricket statistics and see if there was a way these could give a fresh edge. The company referred to the sport as "the Monster", such was the sheer volume of data and variables to process.

In T20, the monster can be tamed. As innings only last 20 overs, it is easier to pre-plan strategy than in the other formats, where scenarios are less predictable. Conditions are also more similar over the span of a T20 match, making comparisons easier. And, most importantly, as matches are played far more frequently than in the longer forms, data has been created faster, which can be analysed to find patterns that can lead to a team making better decisions on the pitch. While analysts say that data - for competitions, players or grounds - is superfluous when it is two years out of date, elite T20 players can easily play 50 or more games in that period.

Data is changing how teams play on the pitch. Even the very best T20 sides, which create the illusion of spontaneous excellence, have learned the value of using data sagaciously to inform team strategy. West Indies' triumphant team in the last WT20 is a particularly instructive example.

Their spectacular boundary-hitting created the impression that their victory was simply a triumph of raw talent and brilliance. The truth is more nuanced. Every day, Simmons would chat with Gaurav Sundararaman - then the team's analyst, now a statistician with ESPNcricinfo. They discussed "match-ups" - the best and worst player v player combinations between the West Indians and the teams they were playing - the ground characteristics, a par score, and second-guessed what the opposition would do, and how best to counteract it.

These conversations were not merely theorising for theorising's sake. Instead, Simmons says, the analytical conversations influenced West Indies' tactics throughout the tournament - like selecting an extra spinner, Sulieman Benn, against South Africa, and using Chris Gayle's offspin for three overs against their left-handers. They also used data to tell their players what to expect: Andre Russell was told to pre-empt slower balls, because of his poor record against them; he launched several out of the Wankhede during the seminal chase against India in the semi-final. In that same game, West Indies' pre-match analysis suggested that 200 was a par score, so they were content to allow India to cruise to 192 for 2 from their 20 overs. And, in the final, using data helped Carlos Brathwaite predict where Ben Stokes would try to bowl to him in the last over - towards the leg side, which was the largest side of the ground.

Rather than follow traditional cricketing thinking about rotating the strike, West Indies decided to embrace boundary-hitting above all else, and were contemptuous of the statistic of "dot-ball percentage"; after all, four dots and two sixes still equals 12 runs an over. Like the three-point revolution in basketball, the six revolution in T20 is influenced by analysts regarding it as the most efficient shot.

Such rigorous thinking links many of the most successful T20 sides - including Kolkata Knight Riders in the IPL; Northamptonshire in the T20 Blast; and Perth Scorchers in the Big Bash.

Not every player embraces data. The role of the coaches and captain is to act as an intelligent filter, to ensure that only a couple of pieces of choice information reach individual players, and their minds are firmly on the game at hand, not the numbers in a spreadsheet. But preparation for T20 matches has evolved away from casual chats in team meetings. For coaches and captains, dossiers of 25 pages are common; these dissect the opponent's strengths and weaknesses, suggest "set-plays", which are the optimum ways that a bowler can set up a batsman, and explain the best parts of the ground to target and which end might suit each bowler best.

Cluelessness about how to approach T20 has been replaced by dissecting the format in minute detail - and the coaches regard the best analysts as worth fighting for almost as much as any player.

When new Australia coach Justin Langer was appointed interim coach for a T20 series against Sri Lanka last year, he took one man with him: Dean Plunkett, the Perth Scorchers team analyst.

"He's an absolute genius," Langer said recently. "He's got a great programme that he's developed himself, a bit like the old Moneyball system. It gives us really clean indicators on our opposition, on our team and who we should select." Plunkett's pre-match dossier contained key performance indicators and optimal match-ups, which captain Adam Voges said "give me a good idea of who I want to bowl at different times".

Plunkett was particularly valued for modelling Perth's squad to identify the overseas players who would provide the greatest additional value, identifying unheralded players like Alfonso Thomas and Yasir Arafat. Rather than complicate thinking, Langer said analytics could clarify it. "In all that pressure and the hectic nature of the game, the more we can simplify things, the better. This might contradict most people's idea of data and analysis."

In this journey, cricket is beginning to embrace more outside voices, who see the game through a clinical analytical lens. Run by former bloggers, Islamabad United have won two of the first three editions of the Pakistan Super League; Mohammad Khan won the Caribbean Premier League in his first year as general manager. A number of analysts have used statistical blogs to enter positions in franchises.

"There is much more potential," Plunkett believes. "The more that people use information, the more efficient they will become with it. I think this is happening in a lot of industries. I'll keep writing code to process the data better and better, and technology will hopefully keep improving the speed at which it runs my algorithms."

If this revolution remains unfinished - professional cricket teams rarely employ more than a single full-time analyst, while it is not uncommon for US sports teams to employ ten each - the contours of change seem clear: strategy and data will become more integral to T20 in the format's next decade.

Here lies the central paradox of T20. The format once derided as a gimmick with no room for strategy is leading cricket to embrace strategy and analysis with greater sophistication than ever before.

Tim Wigmore is a freelance journalist and author of Second XI: Cricket in its Outposts