print icon
Stats Analysis

What looking at James Anderson's and Ishant Sharma's careers as collections of spells tells us about them as bowlers

There's a fair bit to be learned by looking at these bowlers in terms of the support (or lack of it) at the other end

Kartikeya Date
30-Mar-2021
In the latter stages of his Test career, James Anderson has taken fewer wickets but made up for it with the control he has been able to exert on the opposition's scoring rate  •  Michael Dodge/Getty Images

In the latter stages of his Test career, James Anderson has taken fewer wickets but made up for it with the control he has been able to exert on the opposition's scoring rate  •  Michael Dodge/Getty Images

Batting partnerships are recorded as a matter of course in cricket scorecards. What about the bowling equivalent?
Using ball-by-ball records, which are now available for 887 Test matches, from 1999 to 2021, I have constructed spells records for each bowler, along the lines of batting partnership records. For example, if a bowler bowls a five-over spell starting in over 49 of a Test innings and ending in over 57 (overs 49, 51, 53, 55 and 57), then the overs at the other end during this spell are 50, 52, 54 and 56. The record for each spell includes the runs conceded and wickets taken by the bowler during the spell, and the runs conceded and wickets taken at the other end during the spell.
The overall picture presented by the spells record is as one might expect in Test cricket. When wickets fall at one end, they are also likely to fall at the other. The table below shows the overall record of spells organised by the number of wickets that fell at the other end during the spells. All told, three out of four spells are bowled without a single wicket falling at the other end. Only a little over 6% of spells involve at least two wickets falling at the other end. When wickets are falling at the other end, bowlers do better than when they aren't.
The record is sufficiently large now to include the full careers of players like James Anderson, Dale Steyn and R Ashwin (but not those of players like Glenn McGrath, Shane Warne and Harbhajan Singh). In the rest of this article, the possibilities of the spells data set are illustrated using the example of Anderson. Ishant Sharma's record is used as a short second example.
Unlike limited-overs cricket, which is a contest of efficiency, Test cricket is a game of control. The absence of a predetermined limit on the length of an innings over which the ten available wickets are to be taken means that batsmen seek to accumulate runs as safely as possible. Bowlers are the masters of the contest. If a bowler does not bowl a bad ball, and bowls according to the set field, the batsman is unlikely to score freely. Test cricket is a contest of control because it is the bowler's accuracy that shapes possibilities.
Consider the spells record for Anderson. This is presented as a rolling record of 150 chronologically consecutive spells. The first graph below gives a comparison of Anderson's bowling average over each 150-spell period compared to the bowling average achieved at the other end during these 150 spells. The labels on the horizontal axis give the end dates of each of the 150-spell periods.
The record suggests that Anderson's career can be organised into four periods. In his early years, he was not the best bowler in the England side. His wickets were initially more expensive than those taken at the other end, and this meant that he did not hold a regular spot in England's XI.
He returned in 2006 and won his spot in the strong English attack that would win the Ashes in Australia in 2010-11, and win in India in 2012-13, reaching the top of the world rankings.
A third phase of his career - perhaps the finest - began after Graeme Swann retired and that world No. 1 team broke up. Anderson carried the English attack, along with the mercurial Stuart Broad. His wickets came cheap and when he was bowling, England were at their attacking best. The support at the other end was sporadic, though, and this meant that other than in England, the team's results were poor. They were hammered in Australia in 2013-14 and 2017-18 and in India in 2016-17.
A fourth phase appears to have commenced in about the second half of 2018. The English attack has greater depth now, especially on the fast-bowling side of things, and support for Anderson has improved.
A similar comparison of economy rates is given below. The economy rate in a Test match is an important aspect of control. This record adds texture to the four phases above, especially after the first phase, when Anderson was evidently either not sufficiently accurate or bowled the attacking length too often and went for runs.
In the second phase of his career, bowling in a strong all-round bowling attack that had both significant seam-bowling depth and quality spinners in Swann and Monty Panesar, Anderson could afford to be extremely attacking. He conceded about three runs per over. When that team broke up, the record suggests that Anderson changed his approach and decided to become more defensive and restrictive. In the fourth (and currently ongoing) phase of his great career, Anderson has mastered this restrictive style. He concedes less than 2.5 runs per over, while England concede runs at three an over at the other end.
Anderson's career trajectory is illustrated well by his record on his four Ashes tours so far. On his first, in 2006-07, he conceded 4.4 runs per over and took only five wickets at 82.6 apiece. On his second tour, in 2010-11, he conceded 2.9 runs per over and took 24 wickets at 26 apiece without taking a single five-wicket haul. This last suggests (much as it does with Pat Cummins' 29 wickets in the 2019 Ashes in England without a single five-wicket haul), that Anderson bowled in a strong all-round attack. In 2013-14, he conceded 3.2 runs per over and took 14 wickets at 43.9 apiece. This was the tour on which England's world No. 1 team broke apart. In 2017-18, Anderson conceded 2.1 runs per over, and took 17 wickets at 28.
This restrictiveness alongside slightly lesser wicket-taking potency has become a feature of his bowling, especially away from home in this latest phase of his career. In Australia, India, West Indies and Sri Lanka - where the conditions are not traditionally conducive to Anderson's brand of medium-fast seam and swing bowling - his last 14 Tests have brought him 42 wickets at 23.7 apiece. That's only three wickets per Test, so while the wickets have been cheap, he has not been a significant wicket-taking threat. But he has been extremely difficult to score off - 2.16 runs per over. That kind of control is a captain's dream.
The spells record provides further insight into these four phases of Anderson's career. The table below shows these four roughly defined phases as more or less equal numbers of matches, deliveries and wickets. The spells in each phase are classified into two groups. The first includes all spells by Anderson where no wickets fell from the other end. The second includes all spells by Anderson where at least one wicket fell from the other end.
As one would expect, Anderson has typically had better returns when wickets have been falling at the other end (the 2016-21 period is marginally an exception). This is, as the first table in this article shows, generally true for the average bowler in Test cricket. One can imagine why this is - the fact that wickets are falling from the other end suggests that the conditions are probably more bowler-friendly, or that the lower order is in, or both.
Apart from that second phase, Anderson has bowled about two-thirds of his spells when no wickets have fallen at the other end. His returns from these spells show just how far he has come as a Test match bowler. He began as a highly gifted seam and swing bowler with a natural outswinger and a superb, simple, repeatable cartwheel action. When there was little help from the wicket (as evident from the fact that no wickets were falling at the other end), he was unable to exert control in that first phase. In the second phase of his career, Anderson benefited greatly from bowling in a strong attack. It is the only phase in which the majority of his wickets came in spells where at least one wicket also fell at the other end.
Today, Anderson is a true maestro. He is able to control the scoring regardless of the conditions, and regardless of what's happening at the other end. In part, this could be because opponents have decided to see him off. But it is far more likely that opponents would concede wickets to Anderson if they tried to take liberties against him. In other words, it is far more likely that batsmen are compelled to see him off.
Reviewing Ishant Sharma's career in this way offers revealing insights. If we organise his 101 Tests into three roughly equal phases, then his slump in the middle phase is one of more intriguing such troughs in modern Test history. If he had that type of slump after 30-odd Tests today, he would almost certainly lose his spot in the Test team, given the fast-bowling options available to India today. The third phase shows the colossal extent to which his returns have improved.
Why might this be? In part, this is probably because the Indian attack Sharma bowls in today is better than it used to be. In part, this is probably also because there are probably fewer featherbeds today than there used to be.
The BCCI publishes scorecards on its website, and wherever available, they publish ball-tracking data under the Hawk-Eye tab. A review of this data (which is not available for Sharma's full career, but is available for about 7800 deliveries of it, since 2011) shows that in the period until 2016, his average length was 7.4m from the batsman's stumps. Since 2016, his average length is 7m. The same record, which is also not exhaustive for Anderson, shows Anderson's average length (over roughly 13,800 deliveries for which the record is available) during this period to be 7.1m. Bowling a fuller length on average seems to mean that Sharma can attack the stumps more often than he used to, and is consequently more lethal.
Like Anderson, Sharma is able to make his own weather in the Test match arena. His record is not a barometer for how helpful the conditions might be. He commands the batsman's attention, and as is the case with Anderson, even that is often not enough for the batsman to survive.
The spells data set produces a picture of the Test match game from the bowler's point of view. This view is much neglected in the game, and the spells dataset should become a regular feature of the scorecard, in the same way that batting partnerships are. It provides a texture in the landscape of the Test match game that is otherwise difficult to observe.

Kartikeya Date writes at A Cricketing View. @cricketingview