S Rajesh is stats editor of ESPNcricinfo. @rajeshstats
If all goes to plan in these extremely uncertain times, the third Test between Australia and India later this season will be Ishant Sharma's 100th; he will become only the second India seamer, and the 11th overall (including the seam-bowling allrounders but excluding Jacques Kallis, who was more batsman than bowler) to reach the landmark. His average, which is currently 32.39, is the worst among these 11 bowlers - the others all average below 30 - but the fact that he has got this far, with a respectable bowling average (Zaheer Khan finished with a career average of 32.94) is in itself cause for celebration.
A look at Ishant's career graph will explain why. After 70 Tests, his average was a mediocre 37.05; at the time he played his 70th - in July 2016 - only two players had played at least as many Tests and bowled at least 1750 overs with a poorer average, and neither was a specialist bowler: Carl Hooper averaged 49.42 from 102 matches, and Ravi Shastri 40.96 in 80 Tests. Both averaged over 35 with the bat, compared to Ishant's 8.37.
It is extremely rare for a bowler to find so much success after long periods of poor performances. In most cases, the selectors lose patience and faith in the bowler and look for other options, so credit to the various selection panels too for keeping faith in him. In the second half of his career, Ishant has amply repaid that belief.
If we split his career into two halves, his averages read thus: first half (48 Tests) - 38.44; second half (49 Tests) - 27.21. That is a 29% improvement in the second half. Among all bowlers who have played at least 60 Tests and taken 200-plus wickets - there are 60 bowlers who make the cut - no bowler has a higher improvement percentage.
The next two in the list are James Anderson and Richard Hadlee, both outstanding exponents of their craft, for whom the experience gained from playing more matches far outweighed the slight loss of pace in their older years. Anderson's average has improved from 30.74 to 23.28, while Hadlee's improvement is even more creditable: from an already top-notch average of 25.54 in the first half of his career, he improved further to 19.62 in his last 43 Tests. Also in the top ten are two other high-class allrounders: Imran Khan (from 24.74 to 20.55) and Chris Cairns (32.34 to 27.05).
While these are shining examples of bowlers who worked at their craft and improved as they went along, there are others who had an extremely strong first half, but then fell away as their careers went along. Take, for example, another star allrounder of the 1980s, Ian Botham. His overall career bowling average of 28.4 was still splendid, but in his last 51 Tests it ballooned to 36.51, and he only managed 152 wickets, which works out to three per Test. In his first 51, he took 231 wickets at 23.06. The increase in average in the second half was nearly 60% compared to his first. As the graph below shows, that is quite the opposite of Ishant's career chart. Like Botham, Kapil Dev's numbers too dipped in his second half, though not to the same extent: the average only changed from 28.07 to 31.80, though his wickets tally decreased significantly, from 251 to 183.
And then there are the others whose averages in the two halves are almost identical. Check out the nine bowlers in the list below: the percentage change in averages between the first and second half of their careers is less than 2.5%, which means there was hardly any difference between the two halves (though they may have had peaks and troughs within each half, given that some of these bowlers have had pretty long careers). The two bowlers with the least difference are both spinners: Shane Warne and Bishan Singh Bedi. Warne was marginally better in the first half, and Bedi in the second, but in percentage terms, the difference was less than 0.5%.
Coming back to Ishant, while he has made great strides in the second half of his career, those strides have been even longer in the last third of his career. In his most recent 33 Tests, going back to August 2015, he has averaged 22.90. (In 18 Tests since the start of 2018, he has done even better, averaging 19.14.)
If we do a similar comparison between the first two-thirds and the last third of a bowler's career, then Ishant's improvement goes up to a whopping 39%, which is easily the best among all bowlers who have played at least 60 Tests and taken 200 wickets. Anderson remains in second place, with an improvement of 29%: in his last 51 Tests, his average has dropped to an incredible 21.18, compared to 29.77 in the first two-thirds of his career. Mitchell Johnson, Morne Morkel and Tim Southee are the others who have significantly better numbers in the last third of their careers, improving from early-30s averages to sub-25 numbers.
At the other end of the spectrum is Ray Lindwall, the Australian fast bowler, though in his case the fall was significant only when compared to his own extremely lofty standards. Lindwall averaged 20.07 in his first 40 Tests; in his last 21, his average dropped to a still respectable 31.32, but the drop was 56%, due to the high standards he had set in his first 40 matches. Botham averaged 35.74 in his last 34 Tests, after averaging 25.94 in his first 68, while Shaun Pollock's numbers fell away too, relatively speaking, after he had averaged an outstanding 20.7 in his first 72 Tests.
Ishant's bowling graph has been soaring skyward over the last few years, just as his bowling average is steadily moving towards sub-30. At the end of 2017, his career average was 36.55; in a matter of a couple of years and 18 Tests, it has improved by more than four points, to 32.39. That's an extremely steep rate of improvement, which is even more creditable given that it's tougher to make such rapid progress relatively late in a career. However, it needs to be remembered that Ishant's is still a career in progress, and these numbers will change - for better or for worse - by the time he retires. That bowling graph could yet reverse its direction towards the end of Ishant's career - the last few Tests for most bowlers fetch poorer returns than their peak years - but Ishant has already done enough to justify the tag of most improved bowler.