June 11, 2016

The bowlers who hit the zone and stayed there

A look at the best streaks for bowlers in Tests using the bowling average and wickets-per-Test measures

Muttiah Muralitharan had a streak of 565 wickets in his Test career over which he averaged an astounding 18.98 © AFP

As I said in my previous article, this piece is a follow-up to the multi-part series of articles on batsmen who reached amazing heights, such as averages of 60-plus or runs-per-Test values of 95-plus, in their careers. That article looked at golden streaks achieved at any time during a batsman's career.

In Tests, bowlers have every chance to come into their own, as opposed to limited-overs matches. Hence it is essential that I recognise their equal status in Test cricket. This also gives me the chance to do a mid-career streak-based analysis for bowlers and not work with a fixed period starting from the beginning of the career. The same process can then be applied to analysing batsmen later.

Unlike the batting average, the bowling average is a measure without any fundamental weakness. It is simple in its conception. The bowler has captured x wickets while conceding y runs, and what could be simpler than y/x? The two other measures, bowling strike rate and bowling accuracy, are two components of the bowling average. These are performance-related measures. In addition, there is the productivity related measure, Wickets per Test (WpT). Thus there is an element of simplicity and elegance as far as the bowling measures are concerned.

This analysis is in two parts. The first part looks at the bowling average as the key measure. The second part looks at WpT as the key measure.

Bowling average
The bowling average measure favours the pace bowlers. This is mainly because their strike rates are better. This more than offsets the slightly poorer accuracy figures. However, let me state this irrefutable fact. What I have said applies to the recent five to six decades. Taking all 14 decades into consideration, seven spinners find their place among the top 20 bowlers on the bowling average table.

But I am not going to resort to any, possibly unsound, tweaks to adjust the bowling average (BowAvge) figures based on bowler type. These analyses are derived from base numbers and I am going to call the numbers as I see them.

The idea is simple to describe and tough to execute. Start with each Test in the bowler's career. Go through each succeeding Test through to the last Test and determine the cumulative BowAvge. Determine the streak at which the BowAvge is the lowest.

There is one problem, however. John Lever made his debut for England in 1976 against India. He finished with 10 for 70 in that match. The BowAvge at the end of the Test was 7.00. Without any doubt, this was the best average in John Lever's career.

Unfortunately, it is a one-Test streak. Think about the BowAvge figures at the end of the first Test for Bob Massie and Narendra Hirwani. So there is a need for a cut-off. I have used the key measure of wickets. I have set up 100 wickets as the minimum to determine the best BowAvge. This is a significant cut-off, requiring between 15 and 25 Tests.

So, at each iteration, I start my comparisons only after 100 wickets are reached. Now we come to another problem. It is obvious that capturing 200 wickets at 20.00 is a more effective performance than capturing 100 wickets at 20.00. No problems there. Similarly capturing 100 wickets at 15.00 is better than snagging 100 wickets at 20.00. These are straightforward comparisons. However, let us look at other realistic comparisons. Say, a player has taken 100 wickets at a BowAvge of 15.00: an excellent performance indeed. But he continues to perform, doing better, and reaches 200 wickets at a BowAvge of 20.00. Which is better? There is no pat answer. Subjective statements such as "I think 100 wickets at 15.00 is better because BowAvge is more important", or "I think 200 wickets at 20.00 is better since wickets are more important" are worthless. These statements do not provide satisfactory answers because these are not objective.

What is needed is an index combining the BowAvge and number of wickets. The number of Tests is not relevant since a pace bowler can take 200 wickets in 40 Tests while a spinner may manage only 160 wickets in 40 Tests. A lower BowAvge is better and a higher number of wickets is better. So I have adopted an index using these two contrasting measures. The exact formula used is not relevant here, since it will detract from the main issue. It is essential to remember that, as a performance measure, the BowAvge is more difficult to improve quickly. One could bowl in twice the number of Tests and capture double the number of wickets. But the BowAvge cannot be improved from 20 to 15 just like that.

Let the readers not forget that this is only to distinguish between two streaks. Both belong to the same bowler and there are no external comparisons. This is to identify, say, a performance level of 100% over one year against a performance level of 90% across two years and 85% across three years. Readers will understand more as they come to the tables.

The other qualification criterion is that the BowAvge should be less than 20. This is essential since 700 wickets at 25 will surely outscore 200 at 15. And there is no doubt that we are looking at outliers. An average of 20 is a tough cut-off, somewhat equivalent to a batting average of 60. There are a few bowlers like Muttiah Muralitharan, Glenn McGrath, Malcolm Marshall, Waqar Younis, Imran Khan, et al, who have a number of entries. I have used the index as a guide to select what I perceive as being the best streak for these bowlers. For most bowlers there are a few entries.

1. The top 11 bowlers who excelled for long (Avge)
BowlerTestsWicketsBow AvgeStart TestEnd TestTestsWicketsBow AvgeIndex
M Muralitharan 133 800 22.73 1394(1998)1839(2007) 79 565 18.98 125.2
GA Lohmann 18 112 10.76 24(1886) 48(1896) 14 100 9.79 102.1
GD McGrath 124 563 21.64 1296(1995)1558(2001) 64 324 19.97 90.2
MD Marshall 81 376 20.95 884(1980)1100(1988) 50 268 18.90 86.6
Imran Khan 88 362 22.81 869(1980)1058(1986) 35 184 15.93 85.2
SF Barnes 27 189 16.43 65(1901) 133(1914) 27 189 16.43 83.7
RJ Hadlee 86 431 22.30 817(1978)1031(1985) 43 238 19.21 80.3
CEL Ambrose 98 405 20.99 1100(1988)1257(1994) 41 200 19.15 73.9
SM Pollock 108 421 23.12 1318(1995)1528(2001) 48 204 19.44 73.5
Waqar Younis 87 373 23.56 1137(1990)1248(1994) 23 145 16.46 73.1
AA Donald 72 330 22.25 1291(1995)1471(1999) 39 200 19.47 72.7

Muralitharan defies description. His streak spanned the middle of his career, and lasted 79 Tests. His haul of 565 wickets at a BowAvge of 18.98 is undoubtedly the best ever streak by a bowler. Only one other streak could match this: Bradman's 6996 runs in 52 Tests at 99.94. During this long streak, he averaged what many top pace bowlers do not achieve in their careers.

During this particular 565-wicket streak, Muralitharan had three independent qualifying streaks in which he averaged below 20, and the fourth, just above 20. A single streak like this is something many top bowlers achieve once in their careers. These are outlined below.

1. 24 Tests, 167 wickets at 20.35.
2. 15 Tests, 110 wickets at 18.48.
3. 22 Tests, 151 wickets at 18.80.
4. 18 Tests, 137 wickets at 17.92.

A friend of mine once told me that Muralitharan got many wickets because he played in a weak team. I asked him two questions. He could not answer either.

1. If Muralitharan played in a weak team, his high WpT figure is understandable. But how did he manage to secure an extraordinary average of 22.73, which is better than the averages secured by Imran Khan, Keith Miller, Ray Lindwall, Wasim Akram, Michael Holding, Dennis Lillee et al?
2. How is it that Shane Warne, playing for a much stronger team, averaged 25.42? Anil Kumble averaged 29.65, and Mitchell Johnson averaged 28.41. Both played for strong teams.

Normally I am wary of including pre-1912 bowlers in any discussion. Conditions were radically different then. However, George Lohmann was special. He is the only bowler to have a stretch of 100 wickets with a BowAvge of under 10.00, 9.79 to be precise. Granted, the wickets were uncovered and the South African teams were novices. But this is some streak. Only one pre-1912 bowler (Johnny Briggs) and one post-1950 bowler (Jim Laker) have achieved sub-15.00 streaks, let alone sub-10.00 streaks.

McGrath had to share the spoils with at least two other world-quality bowlers, including Warne. However, he was an accurate and very effective bowler, whose career BowAvge is a magnificent 21.64, achieved across 124 Tests and 563 wickets. This long career had a 64-Test sequence during which McGrath claimed 324 wickets at a terrific average of 19.97. Unlike Muralitharan, this streak started as early as McGrath's 14th Test.

Malcolm Marshall was arguably the best West Indian pace bowler. In 50 Tests during the 1980s, he took 268 wickets at 18.90. Look at the average. And let us not forget that there were only eight teams, including a reasonably strong Sri Lanka, playing during these years. And that there were other world-class fast bowlers bowling at the other end. Like McGrath, this was right at the beginning of Marshall's career: from his seventh Test onwards. Embedded in this long career segment were two independent streaks of 148 wickets at 19.74 and 120 wickets at 17.87 respectively.

As a bowler Imran Khan was outstanding in the first three-fourths of his career. Like with Muralitharan, his streak starts in his 26th Test and lasts 35 Tests. During this period he took 184 wickets at an amazing 15.93. This sort of average was not even achieved by Sydney Barnes. But for the injury sustained by Imran during the early '90s, he might very well have crossed 400 wickets at 22 in his career.

Best streaks of 100-plus wickets at sub-20 average by bowlers in Tests. © Anantha Narayanan

The graph is self-explanatory. I have not shown Lohmann, since he had a freak sub-10 average. Instead I have added two important bowlers, Ian Botham and Akram from the second list. The difficulty in maintaining low averages across hundreds of wickets is shown clearly by how the widths of individual bowler's graphs drop. Look at how far left of Muralitharan McGrath and Marshall are. Incidentally Muralitharan has a 100-wicket run at 15.21 and 230-wicket streak at 17.51.

2. The 13 bowlers who excelled briefly (Avge)
BowlerTestsWicketsBow AvgeStart TestEnd TestTestsWicketsBow AvgeIndex
JC Laker 46 193 21.25 386(1954) 456(1958) 20 103 14.81 68.5
Wasim Akram 104 414 23.62 1132(1989)1268(1994) 28 156 19.29 64.8
IT Botham 102 383 28.40 807(1977) 874(1980) 24 134 18.22 63.5
AV Bedser 51 236 24.90 327(1950) 373(1953) 18 104 16.32 62.5
AK Davidson 44 186 20.53 430(1956) 502(1961) 23 120 17.55 62.4
RR Lindwall 61 228 23.03 275(1946) 375(1953) 37 154 19.89 62.4
J Garner 58 259 20.98 802(1977) 989(1984) 35 153 19.99 61.9
DL Underwood 86 297 25.84 622(1967) 701(1972) 24 113 17.58 60.5
GAR Lock 49 174 25.58 412(1955) 513(1961) 24 100 17.39 57.5
VD Philander 32 126 22.09 2016(2011)2108(2013) 19 102 17.57 57.5
DW Steyn 82 406 22.54 1827(2007)1893(2008) 17 103 17.78 57.1
SK Warne 145 708 25.42 1236(1993)1314(1995) 23 129 19.32 58.8
A Kumble 132 619 29.65 1201(1992)1310(1995) 20 100 23.11 43.3

These bowlers reached the sub-20.00 BowAvge mark in fewer Tests and hence have been placed in a separate table.

Jim Laker was almost unplayable between 1954 and 1958. It is appropriate that 1956, when he single-handedly routed the Australians, was right in the middle. His 103 wickets at 14.81 remain the only instance of a sub-15 bowling average after 1900.

Botham's 134 wickets at 18.22 dated from the beginning of his career and ended after the famous Ashes tour by Australians in 1981. When we now compare someone like Ben Stokes to Botham, we should not forget this level of bowling.

Akram took 156 wickets at 19.29 during a period that spanned the debut of Waqar Younis. Waqar brought in the required attacking edge and the two masters ruled the roost.

Warne had an excellent, albeit short, streak of 129 wickets at 19.32 early in his career. Once he perfected his variation and control, the mystery behind his top-quality legspin contributed to this. He was quick to recover from the disastrous 1 for 335 that started his career and was able to get to a sub-30 bowling average within 12 Tests.

Kumble never reached the tough mark of 20 in any bowling streak in his career. His best streak was 100 wickets at 23.11 during the early days. It must not be forgotten that during the first half of his career, Kumble was often used as a stock bowler, and that his overseas bowling became strong only during the mid-2000s.

There are three pre-1900 bowlers who achieved this mark. Briggs took 100 wickets at 13.38, Charlie Turner 101 at 16.53, and Bobby Peel 101 at 16.98.

Imran Khan at his peak took nearly seven wickets per Test © PA Photos

Wickets per Test (WpT)
WpT is a more straightforward compilation. It is also a productivity-based measure. As such, the spinners get equal opportunities. Spinners are the workhorses and put in long spells. So they have as many chances as the pace bowlers to get wickets. In all Test matches, pace bowlers have taken 15.8 wickets per Test and spinners eight wickets per Test. However, the number of spinners has only been around half the number of pace bowlers (59 against 106 among those who took 100-plus Test wickets) so their chances in the WpT stakes are quite good.

The methodology is simple. It is similar to what was done in the best BowAvge streak determination. For each Test for the bowler, I run a forward scan until the end of his career and determine the best streak as far as WpT is concerned. However, the same problem remains. In one golden stretch Muralitharan took 101 wickets in 12 Tests leading to an unbelievable WpT value of 8.42. But he also had a stretch of 565 wickets at 7.15. It is obvious that the latter stretch is a far more important and momentous one. How do I automate this process? Again, I have employed an index methodology. This time the index is created using the wickets and WpT figures. The algorithm does not need to be explained now, but the results do matter and we will see that now.

3. The top nine bowlers who excelled for long (WpT)
BowlerStart TestEnd TestTestsWicketsWpTIndex
M Muralitharan 1394-1998 1839-2007 79 565 7.15 170.0
SF Barnes 65-1901 133-1914 27 189 7.00 96.2
SK Warne 1615-2002 1763-2005 27 173 6.41 84.3
Waqar Younis 1135-1990 1268-1994 29 181 6.24 84.0
DK Lillee 768-1976 910-1981 32 193 6.03 83.8
RJ Hadlee 924-1982 1050-1986 29 177 6.10 81.2
CV Grimmett 194-1930 251-1936 28 169 6.04 78.5
HMRKB Herath 2026-2011 2136-2014 24 150 6.25 76.5
GA Lohmann 24-1886 49-1896 15 108 7.20 74.8

What more can be said of Muralitharan? Look at his career split:

Phase 1: 135 wickets in 34 Tests at an average of 31.16 and WpT of 3.9.
Phase 1: 565 wickets in 79 Tests at an average of 18.98 and WpT of 7.2.
Phase 1: 100 wickets in 20 Tests at an average of 32.53 and WpT of 5.0.

Muralitharan's first phase was ordinary. He more than made up with an astounding second phase lasting nine years. I repeat: the 565 wickets at a WpT value of 7.18 can only be bettered by Bradman's career streak. Muralitharan's last 20 Tests were ordinary only by his own standards. He bought his wickets dearly but still managed to take five wickets per Test. It is of interest to note that there is another streak of Muralitharan's that ran for 90 Tests: 619 wickets at a WpT value of 6.88.

Muralitharan has strong claim to being the best Test bowler ever. I am willing to concede if someone says let us include him in a list of three bowlers and then talk. But nothing less. Those who do not give him this recognition can only be speaking from a biased and myopic point of view.

Barnes is the only bowler whose best streak coincided with his career. There were times when his WpT value exceeded 7.0 but that was over fewer wickets/Tests. His best streak, based on the Index value, coincides with his career. He is the only bowler to have a career WpT value of 7.

Warne comes into his own in the WpT stakes. Playing in a strong team he often had to compete for wickets with McGrath and company. As such, his run of 173 wickets at an excellent rate of 6.41 wickets per Test is outstanding. The fact that Australia, blessed with a strong team, often took 20 opposition wickets was an advantage. Still, it is remarkable that Warne took well over 35% of the wickets.

Waqar's streak starts at the beginning of his career. The start to his career was one of the best any bowler ever had. Waqar, helped by Wasim, Aaqib Javed and Mushtaq, made Pakistan the most potent attack in the word during the early 1990s.

I am glad that the doyen of fast bowlers, Dennis Lillee, has been placed in the top table. Despite the presence of other quality bowlers in Australia, Lillee dismissed 193 batsmen in 32 Tests. This included the period when Lillee was absent from the Australian team because of the World Series.

4. The nine bowlers who excelled briefly (WpT)
BowlerStart TestEnd TestTestsWicketsWpTIndex
Imran Khan 892-1981 946-1983 15 102 6.80 68.7
MD Marshall 971-1983 1039-1986 21 127 6.05 68.2
A Kumble 1616-2002 1741-2005 21 127 6.05 68.2
R Benaud 431-1956 482-1959 17 110 6.47 67.9
Mushtaq Ahmed 1273-1994 1384-1997 15 101 6.73 67.7
AK Davidson 483-1959 537-1963 16 100 6.25 62.5
Saeed Ajmal 1992-2011 2073-2013 16 100 6.25 62.5
DW Steyn 1843-2007 1902-2008 18 108 6.00 62.4
EAS Prasanna 620-1967 669-1969 17 102 6.00 60.6

Imran at his peak took nearly seven wickets per Test. Marshall's WpT of 6.02 should be looked at in perspective - that he had a set of attacking and top-quality pace bowlers operating with him. Kumble's run was during the later stages of his career, when he was far more effective overseas. He was also ready to buy his wickets. Dale Steyn is the only modern bowler to find a place here. Vernon Philander follows closely.

Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systems

Comments have now been closed for this article

  • GV on July 8, 2016, 4:09 GMT

    Cant agree more with Ariz Khan on Botham. In fact, he had quite stunning figures of 284 wickets after 66 tests and 13 centuries, including some impossible ones. No one has come close. His captaincy impacted his career, but even after being relieved, he made a fantastic come back. But he started stiffening up soon after that, and in 1982-83 in an exciting Ashes series in Aus, he was not very effective. But Imran registered some stunning performances at home and away against West Indies, which was to some extent matched by Kapil and Hadlee but not Botham. So not an outlier.

  •   Ariz Khan on July 7, 2016, 3:23 GMT

    After reading Dr. Talha's comment n'th time I finally managed to understand it (100 in first inning), its equivalent to some one suggesting, ~12 years back, that Tendulkar is better than Bradman as TOTAL of his averages in ODI & tests was more than Bradman's!! My mistake of ever bothering to answer him. BTW Ananth, Botham had the greatest start as an allrounder with an average of 40 (6 centuries) and 139 wickets at 18 in his first 25 tests. He was made captain and played against Windies simultaneously and failed. We would never know how he would have performed against them at that point, if was not made captain. He continued to fail, till Brearly took captaincy against touring Australia in 1981. He was man of the match in next 3 tests, starting with THAT 1981 Leeds test, and took 10 wickets in the last test of the series. Before captaincy, there was BOMBAY test and away series against full-fledged Australia where he was highest run maker and wicket taker for England.
    Two contrasting actions by the English selectors, one stupid and the other terrific-just in time, threatened to derail Botham's career and then got it back on rails. Brearley deserves a lot of credit. Botham's start was somewhet equivalent to Mike Hussey's.

  •   Ariz Khan on July 7, 2016, 2:37 GMT

    I am waiting for your next article. There are lot of bowlers who have not gotten their due, Marshall has started getting his due in last decade or so. Its surprising to see how much accolade Wasim gets and how little Imran gets. Its true that Akram was far more versatile, could do everything a fast bowler can do with the ball and more. Imran on the other hand did not have those many repertoire. Mainly In-swing/seam bowler, occasional yorkers and reverse swing (rarely moved the ball away, if ever). But he was far better new ball bowler, something that will become apparent with your weighted-average analysis. He was the best allrounder of the 80s, but some go the extent of comparing his period to Don Bradman, which I think is imbecile. During the said period of Dr. Talha, he played 54 tests, scored ~2600 runs and took 227 wickets, Botham in his first 54 tests had about 400 more runs and 22 more wickets! So IF Imran was equivalent to Don in those 54 tests, I would put Botham above Don!!
    I think only an all-rounder who delivers an average of 50 over 4000 runs and 25 over 200 wickets, say in 50 Tests, can be compared to Bradman.

  • pushkar on July 6, 2016, 20:42 GMT

    Hi Ananth, I am a big fan of yours and this is terrific analysis as always. I know pure numbers paint a story of a player but it removes the context of the game. Be that as it may, I am a little surprised that Kumble scores so low. He is afterall the 3rd highest wicket taker in the world and has won his country many matches. Also, in his time he really did not have too many good bowlers to partner with (much like Murali and very unlike Warne). So how is it that he is scoring so low? Is bowling average the right measure? Shouldn't there be a measure where each bowler's average is compared to other bowlers of the same test (relative average). Or perhaps shouldnt top order wickets count for more ? I am not looking for various cuts of data just to make Kumble look good, but trying to analyze any bowler in a fair manner. I think bowling average skews it a little
    The next article covers a lot more areas. It will be published on Saturday.
    The problem with Kumble's career (maybe the problem with many bowlers) is that at one point in mid-career, Kumble's away average was around 40. It is great that he had a much better second half career away and he finished with 35.86. Harbhajan has a still higher average of 38.9. Contrast these with Murali's 27.8 and Warnes's 24.6.

  • shreenath on June 24, 2016, 17:31 GMT

    Hi Ananth Nice numbers. Similar comparisons for batsmen would be nice. I hope I haven't missed that piece ?
    The last article was on Batsmen.