Bowling February 5, 2010

The best bowler, across years and formats

A month back, I had done an analysis on the the best batsmen, across years and formats; here is a corresponding analysis for bowlers

Muttiah Muralitharan: the leader in Tests and ODIs © AFP
Finally the analysis many of you have asked and been waiting for patiently. This has been on the drawing board for the past six months and I have had quite a few exchanges with many readers to fine-tune the analysis. A lot of care has been taken care to equalise performances by the players across years and across formats. This follows the best batsman analysis.

First, the "Twelve Commandments" followed in doing the analysis.

1. Equal weight for Tests and ODIs. T20-Intls not included since many top players have not played any T20-I matches and anyhow very few matches have been played. Let the number of T20-I matches cross 1000 before we consider it worthy of inclusion in this type of analysis.
2. Recognise longevity measures, such as wickets captured, but make sure that the total weight does not exceed 20%.
3. Especially for ODIs, recognise and incorporate the important fact that during the early 20 years very few ODI matches were played.

4. Unlike Batting where there is no necessity to distinguish between Right and Left handers, there is a clear need to distinguish in the Bowling analysis between Pace bowlers and Spinners because of the significant difference in the three main bowling measures, viz., primarily the Bowling average and secondarily, the Strike rate and Bowling accuracy. Since this is much more pronounced in Test matches than ODI matches, the handling is different.
5. Recognise how the bowler has performed in comparison to his peers, in the Strike rate and Bowling accuracy measures, for both Tests and ODIs. Mura;litharan and Garner are outstanding in these comparisons.
6. Recognise the fact that wickets of top order batsmen should carry additional weight as compared to wickets of late order batsmen. Pathan and Anderson are the best in this regard.
7. Use only career level figures. Match performances, while very relevant, would make it difficult to be equitable to Tests and ODIs. The peer comparisons cover this to some extent.
8. Give weight for share of team wickets. This is quite relevant in Test cricket, but is probably more relevant in ODIs. A bowler is allowed to bowl only 20% of his team overs (barring rain-hit and abandoned matches). Overall a bowler might have shared around 18% of his team's bowling. As such a bowler who captures, say 22/23% of the team wickets has performed admirably and so on. Let me say that even the spinners have done quite well in this measure. Brett Lee leads in this measure.
9. My idea initially was not to distinguish between home and away performances in Tests. However the following table of the the top-5 bowlers changed my mind. Let me assure readers that enough away wickets have been captured by all these bowlers.

Overall   Home    Away    Ratio

Muralitharan 22.71 20.09 27.02 0.74 Warne 25.42 25.55 25.27 1.01 Kumble 29.65 24.90 37.36 0.67 McGrath 21.64 21.97 21.23 1.03 Walsh 24.44 23.15 25.66 0.90

It is clear that bowlers like Warne, even though he is a spinner, and McGrath performed better away than home. Muralitharan was less effective away, but still had a good away average. However Kumble was very average away. This fact has to be recognised and this has prompted to give some additional weight for away average. This is a change of heart and follows the batting ideas.
10. Since this analysis is limited to bowlers who played between 1970 and 2010, work out the algorithms based on these years. In other words, keep out of the equation Lohmann's outrageous figures. An average of 20.00 is the pinnacle, not halfway down the pole. This has helped to rationalise the analysis quite well.
11. Since this is a pure bowler based analysis, exclude the non-bowling factors such as Captaincy, Results, World Cup wins etc. McGrath, Garner and Warne might have won more matches and World Cups than Muralitharan, Ambrose and Hadlee but that should not be used to decide who is ahead in this bowling analysis.
12. I also decided that I would sum the points at rounded-integer level and would tie bowlers who have similar points. I would not use decimal points to separate any groups.

As usual there has to be a minimum criteria. I have decided on 200 combined international wickets AND minimum of 50 wickets in each format. This has allowed me to include Bond, Roberts, Jayasuriya and keep out Bedi, Chandrasekhar and Gibbs. Keeping a single qualification, a la Batting, 100 wickets in each format unfortunately gave me only 49 bowlers, which is quite a low sample.

The table below is quite illuminating and the ratios have been used in making the adjustments between Pace and Spin. For instance the base average for Test Pacers is 27.5 and for Test Spinners is 31.3. To that extent the Spinner would benefit. The base RpO for Test Pacers is 2.85 and for Spinners is 2.58. To that extent the Pace bowler would benefit. And so on... I have taken these numbers only for bowlers who have captured 100 wickets to ensure that the bar is set higher.

Tests played from 1970 - 2010 (only for bowlers who have taken 100 wkts)
Ratios to All
All   Pace   Spin              Pace       Spin
Avge: 28.6   27.5   31.3 - 13.8%      0.96       1.09
S/R:  62.4   57.8   72.8 - 25.9%      0.93       1.17
RpO:  2.75   2.85   2.58 - 10.4%      1.04       0.94

All ODIs (only for bowlers who have taken 100 wkts)

All Pace Spin Pace Spin Avge: 28.9 27.8 31.9 - 14.8% 0.96 1.10 S/R: 39.6 38.3 43.3 - 13.0% 0.97 1.09 RpO: 4.37 4.35 4.42 - 1.6% 0.99 1.01

The following are the points allotted for different measures.
Tests:  Wickets captured- 100
Adjusted wkts   -  50 (adjusted for matches played during career)
Bowling average - 100
Away bow average-  50
Peer comparison -  50 (bowling strike rate comparison)
Peer comparison -  50 (bowling accuracy comparison)
% of Team wkts -   50
% of Top wkts -    50

ODIs: Wickets captured- 100 Adjusted wkts - 50 (adjusted for matches played during career) Bowling average - 100 Peer comparison - 50 (bowling strike rate comparison) Peer comparison - 50 (bowling accuracy comparison) % of Team wkts - 50 % of Top wkts - 50 Wicket quality - 50

The "Adjusted wkts" measure requires an explanation, especially for ODIs. This is best explained with an example. Take the case of Dennis Lillee. He had a career span of 12 years. That is fine and represents a long career. However the problem is that he played only 63 ODIs during this period. Compare this with Brett Lee who, in a shorter 10-year career, has played 186 matches, over 3 times more. An adjustment is needed and this is explained below.

The average number of ODIs per year played by Australia during 39 years is 18.7. The average number of ODIs played by Australia during Lillee's career is 7.75. The wickets captured by Lillee are multiplied by a factor 2.41(18.7/7.75) and points allotted for this measure. For Brett Lee, his career span number for Australia is 27.9 and the multiplying factor is 0.67 (18.7/27.9). Thus this redresses the wide imbalance which exists in the number of matches, especially ODIs, played over the years.

Note that the country figures rather than individual player figures are used since the player might not play due to injuries or non-selection or in Lillee's case, Packer matches. Note also that the base country is used as the base for doing this calculation for the player. Since the number of matches played by various countries varies by a factor of 2.5 to 1, comparisons with a single across-countries base would go haywire.

This is also done for Tests although the variations are far less for Tests.

For both Tests and ODIs, the overall Bowling average, adjusted for the decade values, carries 10 points. Only for Tests, the Away Bowling average, again adjusted for decade values, carries a weight of 5 points.

Independent peer comparisons are done on both Bowling strike rate and Bowling accuracy. The comparisons are only with similar bowlers.

For ODIs, the wicket quality is used by summing the "batting average" of the batsmen dismissed and dividing by the number of wickets. Also to determine the % of top wickets only batsmen whose average exceeds 30.0 (top batsman by all measures) is considered. Harmison is amongst the best in this measure.

Now let me unveil the tables. These tables are current upto Test # 1950 (second Bng-Ind test) and ODI # 2948 (fifth Aus-Pak ODI).

The best bowlers across formats - across years

Rating                     Spin Test   ODI   Test    ODI
Points                     Flag Wkts  Wkts    Pts    Pts
1000                                         500    500

1 752 Muralitharan M Slk * 792 505 380.6 371.7 2 697 McGrath G.D Aus 563 380 348.6 348.8 3 656 Wasim Akram Pak 414 502 301.2 354.4 4 655 Warne S.K Aus * 708 293 346.5 308.8 5 637 Donald A.A Saf 330 272 309.7 327.4 6 635 Pollock S.M Saf 421 393 302.2 332.5 7 631 Waqar Younis Pak 373 416 295.3 335.2 8 624 Garner J Win 259 146 308.2 316.0 9 623 Hadlee R.J Nzl 431 158 328.1 294.9 10 621 Ambrose C.E.L Win 405 225 328.6 292.3 11 600 Lillee D.K Aus 355 103 305.8 293.8 12 587 Marshall M.D Win 376 157 327.0 259.8 13 580 Holding M.A Win 249 142 287.5 292.4 14 573 Imran Khan Pak 362 182 307.4 265.5 15 570 Kumble A Ind * 619 337 286.9 283.3 16 568 Bond S.E Nzl 87 126 265.0 303.3 17 565 Lee B Aus 310 324 232.5 332.9 18 564 Walsh C.A Win 519 227 311.7 252.5 19 561 Saqlain Mushtaq Pak * 208 288 236.1 324.8 20 556 Roberts A.M.E Win 202 87 262.6 293.4 21 553 Ntini M Saf 390 265 243.4 310.0 22 549 Vaas WPUJC Slk 355 400 241.2 308.0 23 545 Shoaib Akhtar Pak 178 223 249.9 295.0 24 538 Kapil Dev N Ind 434 253 258.7 279.5 25 529 Gough D Eng 229 235 246.6 282.2

Muralitharan heads both Test and ODI tables and is ahead by a comfortable margin. Only the churlish and the narrow-minded would deny this great bowler his place at the top. It is easy to say that he played in a weaker team so he had more opportunities to pick up more wickets. What about the batting and fielding. A strong team would have provided these cushions to their bowlers.

McGrath is second in Tests and third in ODIs and again fully deserves his high position. Has there ever been a better fast bowler? He is ahead of his long time compatriot Warne and Wasim Akram quite comfortably.

Wasim Akram is the other way around. Had a great ODI career (he is second) but had a slightly below-par Test career, of course compared to the Test giants. Overall a phenomenal fast bowler, worthy of his third position.

Warne is third in Tests. His achievements are legendary and do not need further words. He is a hair-breadth behind Wasim Akram. Would we ever see a twosome like McGrath and Warne bowling together?

The outstanding South African fast bowler, Donald is in fifth position. Consistency across the formats is his forte.

The top-10 is completed by Pollock, Waqar Younis, Garner, Hadlee and Ambrose. I am quite happy that there three great pairs in this group. It should be noted that the somewhat low number of wickets of Hadlee and Garner has not prevented them from coming to the top. Lillee misses the cut mainly because of the average ODI placements, not his fault, though. Marshall could also not find his place in the top-10 because of the ODI points.

The top-10 has 2 Australians, 2 Pakistanis, 2 South Africans, 2 West Indians, one Srilankan and one New Zealander. A fair distribution, one would say, with 6 countries represented. For the record, Kumble, Gough, Streak and Mashrafe Mortaza are the other country's best bowlers.

There also two spinners in the top-10 and four in the top-20. This is in line with the overall proportion since only 19 spinners qualified, representing 25% of the total. For the record, Chris Gayle props up the rankings.

Overall this is a fast bowler-dominated period. Only 25% of the bowlers are spinners. Also if one takes great spinners during this period, it would be difficult to look beyond Murali, Warne, Kumble, Abdul Qadir, Saqlain and Harbhajan (probably not there yet). However if I have to select the best pece bowlers, a limit of 15 would leave me unhappy.

To download the complete all-time list, please right-click here and save the file.

Because of the length of the article I am not dwelling on the individual tables in depth. Suffice to say that Muralitharan, McGrath, Warne, Ambrose, Hadlee, Marshall, Walsh, Donald, Garner and Imran Khan could not be bettered as a Test top-10 of the past forty years. This list is dominated by West Indies, the most effective bowling team over the past 40 years.

To download the complete Test list, please right-click here and save the file.

And the ODI-10 of Muralitharan, Wasim Akram, McGrath, Waqar Younis, Pollock, Lee, Donald, Saqlain Mushtaq, Garner and Ntini represented the cream amongst ODI bowlers. I am happy that Saqlain, with an incredible ODI average of 21.7 and strike rate of 30.5 finds a place in the top-10. This group is dominated by Pakistan, rightly so.

To download the complete ODI list, please right-click here and save the file.

A request to readers. You have every right to comment negatively. Every right to fault this analysis. Every right to be upset. Every right to disagree. What you do not have is the right to be abusive, personal or otherwise, to me or to the other readers or to the great players themselves or to other countries.

After a reasonable break I will do a "Who is the best player - across years and formats" article.

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