Wednesday, 4 March 2015

Cricket’s great data debate: art v science

Andy Bull in The Guardian

In July 2007, after a history reckoned to stretch back almost 4,000 years, the game of draughts was finally solved. After two decades of work, a team of computer scientists at the University of Alberta finished sifting through the 500 billion, billion possible positions on the board. Their computer programme, Chinook, was now unbeatable. So long as neither player made a mistake, every game it played was guaranteed to end in a stalemate. Later that same summer, Peter Moores was appointed as head coach of the England cricket team. Moores was one of the new breed of coaches. A numbers man, and disciple of Michael Lewis’s much abused book, Moneyball. He even gave a copy to his batting coach, Andy Flower. Moores was so keen on advanced computer analysis that he used it as the sole basis for some of his decisions – the decision to recall Ryan Sidebottom to the side, for instance.

When Flower took over the team, he hired Nathan Leamon, a qualified coach and a former maths teacher, as the team’s analyst. The players nicknamed Leamon “Numbers”. He was extraordinarily meticulous. He used Hawk-Eye to draw up spreadsheets of every single ball delivered in Test cricket in the preceding five years. He ran match simulations – accurate to within 5% – to help England determine their strategies and their team selections. For the bowlers, he broke the pitch down into 20 blocks, each of them 100cm by 15cm, and told them which ones they should hit to best exploit the weaknesses Hawk-Eye had revealed in the opposing batsmen. Bowlers should aim to hit that particular block at least twice an over. Do that, Leamon told them, and they would “markedly increases the chance of success”.

England, it was said, were making better use of the computer analysis than any other team in the world. And it was working. They won the World T20, the Ashes home and away, and became, for a time, the No1 team in all three formats of the game. Leamon’s work was picked out as one of the reasons why. And yet now they’re losing, that very same approach is being singled out as one of the things they are doing wrong. You can see why. After England’s nine-wicket defeat to Sri Lanka, Eoin Morgan said “Going in at the halfway I think we got 310, probably 25 for both par, and again, stats back that up, par is 275, 280.” It was, Morgan thought, the bowlers who were to blame for the loss. They had delivered too many bad balls. He said he didn’t yet know why. “Over the next couple of days, we will get the Hawk-Eye stuff back and the proof will be in that.”

On Tuesday morning, Kevin Pietersen tweeted that England “are “too interested in stats”. He was echoing Graeme Swann’s comments from last summer. “I’ve sat in these meetings for the last five years,” Swann said. “It was a statistics-based game. There was this crazy stat where if we get 239 – this was before the fielding restrictions changed a bit so it would be more now, I assume – we will win 72% of matches. The whole game was built upon having this many runs after this many overs, this many partnerships, doing this in the middle, working at 4.5 an over.” Swann said he was left shaking his head.

Two respected players, both speaking from fresh first-hand experience, agree that England have become too reliant on computer analysis to tell them what to do. But balance that against the irritation old pros in all sports feel about big data. Just last week the great blowhard of the NBA Charles Barkley unleashed this tirade: “All these guys who run organisations who talk about analytics, they all have one thing in common – they’re a bunch of guys who have never played the game, and they never got the girls in high school, and they just want to get in the game.” Analytics, Barkley added, were “just some crap that some people who were really smart made up to try and get in the game”.

Barkley was shot down in flames. As Bryan Curtis summed it up in his wrap over on Grantland, commentators argued that Barkley’s rant was “unintelligible” and “wholly useless”, that he was a “dinosaur” who “didn’t even realise that the war is over”, and that “the nerds make the decisions”. In England though, where we’ve been slower to adopt analytics, the consensus seems to be that Swann and Pietersen are on to something. England’s over-reliance on the numbers has become a theme in the coverage of the team, particularly among ex-players. You can hear it when they bemoan, among other things, England’s reluctance to bowl yorkers at the stumps. That’s a tactic that has worked for years, one that has been honed by hard experience. But England’s analysis has told them that slow bouncers and full balls sent wide of off-stump are harder to score off.

The thing is, in an age when all teams are using computer analysis, a tactic isn’t good or bad because it looks that way, or because it is different to what has been done before. It is simply good if it works and bad if it doesn’t. The received wisdom is being challenged, and that’s a good thing. At the same time, cricket isn’t checkers. It can’t be solved by computer. It’s not a question of intuition versus analysis, or art v science, as David Hopps put it in a recent piece on Cricinfo. The laptop is just another tool in the box, useless unless the players understand the value of the information it provides, and no more valuable than their own ability to adapt and improvise during a match. If Swann and Pietersen are right, then England are wrong. At the same time, the lessons Leamon taught the team undoubtedly played a valuable part in their earlier success, something the sceptics seem to have forgotten.

No comments:

Post a Comment