Do strong bowling pairs have it better than lone warriors?

"Did Shane Warne lose something because he played alongside an equally great bowler, Glenn McGrath?" "Did Muttiah Muralitharan gain substantially because he was a lone warrior?" Questions of this sort are posed to me often. They might not have clear answers, but I decided that I would embark on a voyage to seek them.

To do this, I have selected three great bowling pairs and two magnificent lone warriors. I will look at every aspect of their bowling in terms of wickets taken, performances in matches won, and the bowling support they received.

Warne and McGrath walk in as automatic selections. Arguably they form the greatest bowling arrowhead any team has ever fielded. I considered many other pairs before picking Wasim Akram and Waqar Younis, and Courtney Walsh and Curtly Ambrose as the other two pairs. The remarkable achievements and outstanding skills of these pairs stand out. I did not consider James Anderson and Stuart Broad because they are still active.

For the lone warriors, I have selected Murali and Richard Hadlee. No other bowlers did more for their country single-handedly than these two. For the pairs, I will only consider the Tests in which both bowlers played. This will allow me to perceive the bowling pair as a single entity. Here, the set of Tests will be their entire career.

Career-to-date-location average

For each bowler, this figure represents the home/away bowling average at the start of a Test. These values are not updated at the end of the first innings. The bowling average is calculated normally. The away figures represent the composite of all away locations, including neutral ones. Some tweaking is necessary to take care of the first few Tests.

Until a bowler crosses 20 wickets in the appropriate location, special tweaks are in place. A regular bowler, who has captured in excess of 50 wickets in his career, will have as his career-to-date-location average his career average or 25.0, whichever is higher. If he is a casual bowler with fewer career wickets, the figure will be his career average or 40.0, whichever is higher. This ensures that bowlers like Narendra Hirwani, who took 16 wickets on debut, do not have outrageously low figures and bowlers like Shane Warne, with nightmare starts, have some protection. Once the bowler crosses 20 wickets, the actual average kicks in. In general, the values stabilise after the bowler has played around ten Tests.

In this analysis I am going to draw almost all my inferences from the graphs. This way the readers can share the voyage of discovery with me. Other than the look into the bowling support received, there is going to be an analysis of wickets taken per Test. Ultimately this single measure does the most towards achieving wins for a team.

Taking ten wickets in a match is approximately equivalent to scoring a double-hundred, in terms of frequency, as explained below. However, these two monumental feats differ significantly when we dig deeper.

There have been 435 instances of bowlers capturing ten wickets in a match. Out of these, 321 Tests have ended in wins for the team, 40 in draws and 74 in losses. This shows that if a bowler captures ten wickets in a match, there is 74% chance that his team will win. There have been 366 instances of batsmen scoring double-hundreds in an innings. Out of these, 185 have ended in wins for the team, 163 in draws and 18 in losses. So a batsman scoring a 200 gives his team only a 50% chance of a win. It is possibly a loss-avoidance effort. This is something to keep in mind as we begin the analysis.

Shane Warne and Glenn McGrath
Let me look first at the three featured pairs. Warne made his debut in 1992 and McGrath almost two years later. Both ended their careers in the same Test, in 2007. So one could say they had almost totally overlapping careers. The total span was 177 Tests out of which both played in 104 and neither in 15.

Some general information about the graphic presentation. The Y axis represents 20 wickets - ten wickets on either side, going from bottom to top and vice versa. Normally such graphs have the bars drawn up or down from a central axis. However, I have drawn these as inward-facing graphs so that the wickets captured are very clearly shown.

Each wicket is shown as a rectangle. The wickets captured by the featured pair/bowler are shown in red and the ones taken by the other bowlers in grey. The white spaces indicate the wickets that did not fall.

The X axis represents the Tests played by the bowlers. In the graphs covering all matches, such as the one above, each black blob at various points along and beneath the X axis represents a win for the team. In the graphs covering won matches, the black blob means that the bowler has taken the most wickets or the pair has together captured 11 or more wickets.

The most striking factor in the first graph above, comparing Warne and McGrath with the other bowlers in the team, is that the average wickets-per-match values for the two bowlers and the rest are very close; there is only a 10% differential. This is a clear indication that the other two or three bowlers contributed a lot and Warne and McGrath were never a truly dominant pair for Australia. A more dominant pair would probably have had WpT (wickets per Test) values of 10.5 against 7.5. Therefore, we are already beginning to see the effect of the overall bowling quality. This resulted in 71 wins for Australia but also meant the pair (and each of the two bowlers constituting it) didn't take a disproportionately large number of wickets.

For the leading pairs, I have defined a dominating streak as a minimum of three Tests in which they took 12 or more wickets in each match, or four-plus Tests with 11 wickets per match. It is not surprising to note that there is only a single streak of five Tests in which Warne-McGrath were thus dominant. All such streaks are marked on the graphs in darker red.

Now let us look at the graph of Tests won by Australia when Warne and McGrath bowled together.

Australia won 71 out of the 104 Tests in which Warne and McGrath played together. That is a very impressive 68.2%. However, a perusal of the graph above will indicate that even in these Tests, Warne and McGrath took 10.6 WpT and the other bowlers 9.3. The differential is still 10%, which makes the contribution of the other bowlers very significant.

In 33 of these matches, Warne and McGrath made a more significant contribution than the other bowlers - in other words, took 11 or more wickets. That means in 38 of these matches the other bowlers - Jason Gillespie, Brett Lee, Stuart MacGill et al - equalled or outperformed the famed pair.

Warne and McGrath had four streaks in which they contributed significantly to Australian wins. In the first and the fourth of those, they took 12 or more wickets in each of three Tests. In the middle streaks, they took 11 or more wickets in four or more Tests.

Towards the end of the careers of Warne and McGrath, Australia won Tests despite a lack of serious contributions from the duo. A look at the last nine Tests will make this clear. The two instances where we see white spaces are from 2006: the SCG Test in which South Africa declared and the Adelaide Test in which England declared.

Let us see whether the next chart reinforces the insights we have gained.

The bowling-average graph has a middle line signifying an average of 40. A horizontal side-by-side plot is not possible because of the large number of Tests to be plotted. In the chart above, I have plotted the career-to-date-location average of Warne and McGrath vis-a-vis the two other best bowlers in the side.

Across the 104 Tests, Warne and McGrath averaged 23.9 (taken together). Barring a few occasions, they never went above 25. That is terrific. However, we see that the support bowlers were not far behind. They averaged 27.4 across those Tests, and had better figures than Warne and McGrath in some Tests. Remember that these are not two bowlers but two-bowler combinations from a collection of bowlers ranging from Brett Lee to Colin Miller. This value is only 15% higher than Warne-McGrath's.

Courtney Walsh and Curtly Ambrose
Walsh made his debut in 1984 and Ambrose four years later. Walsh called an end to his career in 2001 and Ambrose a year before that. Therefore, their careers did not overlap in the same way as Warne and McGrath's. The total span was 132 Tests, out of which neither played in seven. However, I am only interested in the 95 Tests in which both played.

In those 95 matches, Walsh and Ambrose averaged only 8.0 WpT and the other bowlers, 8.4 WpT. The difference between them and Warne-McGrath can be clearly seen.

Walsh and Ambrose never dominated and were often outbowled by team-mates. This is more pronounced in the initial stages of their careers, when Malcolm Marshall and Patrick Patterson were on the scene. There is only one streak - and that one does not properly fulfil the criteria drawn up. Towards the end, in a golden sequence of five Tests, Walsh and Ambrose captured 56 wickets. There is also a seven-Test sequence just before that, in which they took 75 wickets. Just around their 65th Test, there was a period of nine matches in which Walsh and Ambrose were off colour and West Indies won just one Test.

Now let us look at the graph of Tests in which Walsh-Ambrose took West Indies to wins.

The graph above is cheese to Warne-McGrath's chalk. Walsh and Ambrose were part of 42 wins out of the 95 matches in which they played together -44.2%, which is an indication of the gradual decline of West Indies. The differences with Warne-McGrath are stark. The average WpT stands well separated at 9.7 and 10.3, with the featured duo taking a back seat, in contrast to Warne-McGrath.

In 1994, Walsh and Ambrose had a golden run in four wins, taking 12, 16, 15 and 15 wickets. The pair was its peak around the midpoint of their careers together.

The early part of their career as a pair is intriguing. In the first 14 Tests they played together in West Indies wins, Walsh and Ambrose were the leading bowlers in only three, possibly because the other giants were still in play. This changed slowly and they became the dominant pair in the majority of the remainder of their Tests.

Out of the 42 Tests they won together, they were the dominant bowlers only 19 times, which is less than 50%.

One has to marvel at the consistency of Walsh and Ambrose. From the beginning, barring a few Tests, their bowling average almost never went above 25 as a pair; they were consistently around the 24 mark, with very little variation whether they were playing home or away.

The average for the support bowlers for these two is 28.7, and consists of three distinct historical phases: around 25 at the beginning, 30 around the middle, and nearing 35 at the end.

Wasim Akram and Waqar Younis
Akram made his debut in 1985, and Younis almost five years later. Akram retired in 2002 and Younis a year later. Their total span was 130 Tests, out of which neither played in 13. However, I am only interested in the 61 Tests in which both played.

In the 61 Tests they played together, Akram-Younis averaged 9.2 WpT and the other bowlers 7.3 wickets, a clear 20% behind - in contrast to Walsh-Ambrose. The domination of Akram and Younis is also more pronounced. There are three streaks of five, three and six Tests fulfilling the criteria drawn up.

Akram and Younis were part of 28 wins in the 61 Tests they played together. This works to 45.9%, comparable to the West Indian pair. However, when we look at the graphs, the differences are stark. The average WpT for Akram-Younis in matches Pakistan won stands at 11.6, ahead by over 25% to the average of the other bowlers, 8.4.

In the first 13 Tests they played together in Pakistani wins, Akram and Younis dominated as no pair has ever done and took 11 or more wickets in each match. In fact, their average was 14.5 wickets per Test during this golden period.

Then the wheels came off. Just twice in the next 15 Tests did Akram and Younis exert this type of dominance. Out of the 28 Tests they won together, they were dominant 15 times, which works out to 53.5%. The architects of wins in the remaining 13 Tests were the other bowlers - mainly Saqlain Mushtaq and Shoaib Akhtar.

The chalk-and-cheese situation continues. Akram-Younis across their careers together had better figures than even Warne-McGrath, with a combined average of 22.7. However, the support was quite average: an average of 32.1, over 40% higher. In the initial stages, the presence of Imran Khan and Abdul Qadir meant that the support was decent - averaging around the 26-27 mark. But this ballooned to well over 30 and even touched 40. Towards the end, with Saqlain and Akhtar establishing themselves, the support got better.

Did the bowling pairs lose out?
It is evident that Warne and McGrath had not just each other to contend with when it came to taking wickets, they also had the other bowlers. This overall excellence helped Australia as a team and took them to a win percentage of nearly 70%. However, Warne-McGrath as a pair took a clear hit in their personal numbers because of the quality of the bowlers they bowled with and finished with career WpT figures of 4.88 and 4.54. In wins, they were dominant only 46% of the time.

This is not the case with Walsh and Ambrose. Especially in the second half of their careers, they had less support, which gave them greater chances to contribute more effectively. Therefore, we can only say that they might have deprived each other of some share of the spoils but they certainly did not lose out as a unit.

In victories, Walsh and Ambrose were dominant only 45% of the time. Their career WpT figures are 3.93 and 4.13 respectively. These are not the figures of a dominating pair. It is true that they played second fiddle to start with. However, during the latter part of their careers, there were many windows of opportunity to do more.

In wins achieved, Akram and Younis were dominant 54% of the time. Their career WpT figures are 3.98 and 4.29 respectively. Especially in the second half of their careers, they had less support and there was an opportunity to do more. The difference between the first 13 Test wins, in which they averaged 14.6 WpT, and the next 15 wins, in which they averaged 9.1 WpT, stands out. Therefore, we have to conclude that they shared the wickets available but they certainly did not miss out.

Muttiah Muralitharan
Here the comparison is between one master bowler and the other bowlers. We will consider the 133 Tests Murali played in his career. First the wickets comparison with the other bowlers.

Muralit averaged 6.0 WpT. All the other bowlers together averaged 9.6 WpT, which is only 60% more than Murali. Considering that three bowlers shared these 9.6 wickets, Murali's average WpT is nearly twice that of any of the other bowler.

Now for the streaks. For individual bowlers, I have considered capturing ten wickets in a match as a minimum consideration. After all, that is quite rare and almost always wins matches. Murali had two streaks of four matches in which he took ten wickets in each Test. I suggest that the reader rewind and read this statement a couple of times. Four consecutive ten-wicket hauls is like a batsman scoring four consecutive double-hundreds - never ever achieved, although it must be said that Kumar Sangakkara accumulated 200 runs in each of four consecutive Tests. Moreover, Murali achieved this monster streak twice in his career.

It took him 35 matches to get his first ten-wicket haul. Afterwards he achieved the feat 22 times in 98 matches - approximately one every 4.5 Tests. By way of comparison, Anderson and Walsh have each taken five wickets in an innings once every 5.5-6 Tests.

In 98 of his 133 Tests, Murali was the leading wicket-taker. In five matches, he did not capture a wicket. However, only in two of these did he get a decent bowling spell going.

Sri Lanka won 54 of the 133 Tests Murali played in. This works out to 40.9%, and might seem low. However, it must be understood that the bowling support was average and the batting, until the great duo Mahela Jayawardene and Sangakkara were established, patchy. In the 54 wins, Murali took 8.1 WpT. The other three bowlers captured 11.9 WpT, and this works out, as an average per bowler, to less than 50% of Murali's capture rate. A clear indication of how much his presence meant.

The first of the two amazing ten-wicket streaks was against India, Bangladesh and West Indies (twice). This was the 2001 West Indies - a good team. The second streak was against England (twice), South Africa and New Zealand.

In 43 of the 54 Tests won, Murali was the sole or shared leading wicket-taker for Sri Lanka. That works out to an amazing 80%. Out of the 22 times he took ten wickets in a match, 18 ended in wins.

In the chart above, I have plotted the career-to-date-location average of Murali against the average of the three best of the other Sri Lankan bowlers. We begin to see the true situation now. Through his career Murali averaged 25.2: a few runs fewer at home, and a few runs more away. This explains the jagged nature of the Murali half of the graph. There are clear groups of Tests on either side of around 24.

This chart also depicts how he struggled until around the 40th Test of his career. He was consistently averaging around 30. Then the transformation took place.

However, look at the support he received. The average for the supporting bowlers across all the Tests is 33.6, with the value often nearing 40. There was Chaminda Vaas almost right through Murali's career, Lasith Malinga and Rangana Herath on and off, and Ajantha Mendis during the last two years or so. However, there was no sustained support, especially away, and this is clearly shown by the fact that in most of the Tests the support average varied between 30 and 40.

Richard Hadlee
In terms of WpT, Hadlee dominated less than Murali did. His WpT stands at 5.0, while the other bowlers took nearly 10. These figures are not as good as the corresponding ones for Murali.

Allowing for a bit of leeway when determining the streaks, there is only one streak of note. Around two-thirds of the way into his career, Hadlee took 33 wickets in three Tests against Australia away. That included the greatest Test bowling performance of all time in my book - the 9 for 52 in Brisbane in 1985-86. Hadlee reached the ten-wicket mark fairly early, in his ninth Test, and got to this mark eight more times. He didn't ever take ten wickets in a match in each of two consecutive Tests. And in two matches, he failed to take a wicket at all.

Of the 86 Tests Hadlee played in, New Zealand won 22, just over 25%. This is a reflection both of the absence of quality bowling support and of an effective batting line-up.

Hadlee's average WpT in wins moves up significantly, to 7.9. His bowling partners chip in with 12.1. In this regard, Hadlee almost matches Murali's figures (8.1 and 11.9). Hadlee was the dominant bowler in 16 of those 22 winning Tests. That is a fair share, maybe not as high as Murali's but quite good.

There was a three-Test streak in which Hadlee was the dominant bowler, and that was during the early part of his career.

The figures above are similar to those of Murali. Hadlee's average of 25.4 and the other bowlers' average of 33.5 are almost identical to the 25.2 and 33.6 in Murali's case. For quite some time, Hadlee's away figures were in excess of 35. However, this changed later in his career. Murali had widely varying home and away figures, while towards the end of his career Hadlee seems to have had similar home and away figures. Note the almost straight-line graph towards the end of his career.

Did the lone warriors gain?
The biggest difference between Murali and Hadlee is in their contribution towards their team winning. Murali was the dominant bowler in over 80% of Tests won by Sri Lanka. Therefore, the claim that Murali, as an individual, gained by playing in a weaker team might be true but he deserved it because he was not just a single bowler forming 25% of the attack. Often he was half the attack. He was able to create a number of additional wins for Sri Lanka.

Hadlee was handicapped by the absence of bowling and batting support. However, his contributions to New Zealand's wins were very substantial. Overall, his 5.0 WpT did not indicate that he gained anything extra as a lone warrior. If anything, the lack of penetration at the other end probably cost him a little. It is clear that among these two, Murali had the better support overall, and this helped him and the team.

Murali v Warne
Now to compare one lone warrior with one bowler from a pair. The graphs have changed quite a lot for this analysis. Until now, a vertical column represented a single Test, within which different measures for different bowling combinations were presented. However, the graph below is a comparison of the careers of bowlers from different teams and different time spans. The graphs are shown as two distinct ones. The plotting of wickets captured goes upwards for both bowlers. The sample is the complete career of the concerned bowler.

Murali took ten wickets in a match 22 times, which works out to one every six Tests, while Warne achieved this ten times - a good tally indeed, but one that pales before Muralitharan's number.

I was able to find no less than five streaks of very good performances for Murali and had to lower the bar to locate three streaks for Warne. If we take a very effective contribution by a bowler as eight or more wickets in a Test, Warne achieved that mark 31 times, around 21% of his career, and Murali, 46 times, around 35%.

If we take a sub-standard contribution by a bowler as three wickets or fewer in a Test, Warne had those 53 times, more than a third of his career. Murali's tally is as low as 33, around 25% of the Tests he played in.

There is a huge difference in the WpT values for the two bowlers: 6.0 for Murali against 4.9 for Warne.

The only measure on which Warne scores higher is in the number of wins: 92 out of 145, which works to 63.4%. Murali led Sri Lanka to 54 wins, or 40.9%.

Hadlee v McGrath
Hadlee took ten wickets in a match nine times, which works out to once every ten Tests, while McGrath achieved this only three times - poor by comparison.

I was able to find only a single streak of good performance each for both Hadlee and McGrath.

McGrath took eight or more wickets in a match only 14 times, a mere 11% of his matches, and Hadlee 19 times, around 22%.

McGrath has failed - if we describe taking four or fewer wickets a Test as failing - no fewer than 47 times, which is more than a third of his career, while Hadlee's tally is 29 Tests, around 34% of the Tests he played in.

There is a slight difference in the WpT values for the two bowlers: 5.0 for Hadlee against 4.5 for McGrath.

Like Warne, the only measure in which McGrath scores high is in the number of wins: 84 wins out of 124, which works to 67.8%. Hadlee led New Zealand to 22 wins, which works to 25.6%.

These comparisons only substantiate the conclusions already drawn. Warne and McGrath clearly lost out to the other quality support bowlers in their team. The other two pairs had less support and this is shown in the results. However, they themselves could have done more. Among the lone warriors, Murali, through his own efforts and better support, substantially improved the performance of his team.

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

Stats analysis

Do strong bowling pairs have it better than lone warriors?

RELATED ARTICLES