How playing the numbers game thrust data gurus into front line
Test teams are using analysts more than ever – and it is reaping rich returns for Ashes rivals
Two years ago, Justin Langer stood in as Australia Twenty20 coach. He took with him a solitary member of his backroom team at Western Australia and Perth Scorchers: analyst Dean Plunkett. For his dossiers, match-ups – who should bowl when to which opponent – and work identifying signings, Langer acclaimed Plunkett “an absolute genius”.
As Australia head coach, Langer has brought this rigour to international cricket – including Tests, where data has traditionally been less prominent than in T20. Dene Hills, a former first-class player and now Australia’s performance analyst, has been used more prominently than under Langer’s predecessor Darren Lehmann, who had little time for data. While Hills’s pre-match team meetings aim to simplify information, there is a strong empirical grounding to decisionmaking. Indeed, earlier this year Australia negotiated a trial to use Cricviz, the cricket analytics company. In the 2015 Ashes, Australia attacked England with their quickest, scariest bowlers.
It did not work: Australia’s quicks leaked 3.8 runs per over. In their pre-series planning, Australia resolved to avoid such mistakes and focus on a relentless line and length. That meant omitting Mitchell Starc until the fourth Test, while selecting the altogether less glamorous Peter Siddle in the first two. Australia’s pace bowlers have yielded 2.9 runs per over.
It is not the only instance of data influencing who takes the field this series. When Moeen Ali’s place was discussed after the first Test, Steve Smith’s Test average of 27 against left-arm spin, compared with 104 against off spin, counted in Jack Leach’s favour. Joe Denly’s impressive weighted average in county cricket – a metric taking into account factors such as the quality of opponents – has been a factor in his Test selection.
England’s analysis uses a grid
system, which breaks down the pitch into 20 blocks, of 100cm by 15cm each, finding the optimal length to bowl both for the ground and the opposition batsman. The numbers suggested that Craig Overton’s natural length – slightly back of a length – was ideally suited to Old Trafford.
Before this Test, Australian national selector Trevor Hohns said that, “mindful” of England’s record bowling to left-handers, they were keen to play more right-handers. Since 2017, Stuart Broad averages seven runs fewer per wicket against left-handers. While Australia did not select Mitchell Marsh, as Hohns floated, they promoted Marnus Labuschagne to three to give Australia an extra right-hander.
Data can turn hunches into facts, thereby encouraging bowlers to double-down on a weakness. Two batsmen who started the series well showcase as much.
Travis Head made 35 and 51 at Edgbaston, but averages 14 in Tests against balls from seamers hitting the stumps. When he played down the wrong line to Broad, it was the fifth time out of six Head had been bowled or lbw this series.
Rory Burns began the series with 133, yet more significant was his second-innings dismissal fending away a short ball. Australia have exploited this scintilla of weakness ever since. Seamers delivered 30 per cent of short balls to Burns at Edgbaston, which rose to 48 per cent at Headingley; short balls have dismissed Burns in four of his past five innings. The barrage he received last night showed how Burns’s prospects may hinge on his fortitude playing the short ball.
In a sense, none of this is anything new. Bowlers have always dissected batsmen and worked out the best way to attack them – and on occasion it does not work, as England’s seamers found as they toiled away against Smith yesterday with no reward. But data has accelerated and clarified this process. Rather than relying on the fallibility of the human brain, now teams can gauge whether there is an empirical basis to their beliefs. Never has it taken less time for batsmen’s weaknesses to be distilled into cold numbers, and for bowlers to hit upon the optimal approach. All of this may well be part of the broader story of average scores trending downwards in Tests. At Nottinghamshire, Broad was shown findings that opposing batsmen were leaving 28 per cent of his deliveries; right-handers even more. The figure was “a lot higher” than the norm for the best bowlers, says Kunal Manek, the Nottinghamshire analyst. During his finest spells, batsmen only left around 13-15 per cent of Broad’s deliveries.
“There was a strong enough correlation to build a case,” Manek recalls. “This stat intrigued him. He obviously thought it was pretty relevant to finding form again ahead of the Ashes.” Broad later acclaimed these findings as “brilliant coaching and analyst work”, and crucial in him getting batsmen to play more.
None of this is to deny how data can confuse more than clarify. If Denly’s dismissal was an endorsement of Australia’s short-ball strategy against him, it took his Test average below 23 from seven Tests. This Old Trafford pitch has less pace than previous Test wickets here, rendering the back-of-a-length style that is Overton’s forte less effective. Data is easy both to oversell and denigrate. Yet it is playing a greater part in Test cricket than ever before. And so navigating both the promise and perils of data will become ever-more important.