THE CORDON HOME

BLOGS ARCHIVES
SELECT BLOG
March 5, 2010

Tests - bowling

Bowler consistency analysis - a new take

Anantha Narayanan

Muttiah Muralitharan has taken ten or more wickets in a Test in four consecutive matches © AFPs
First I wanted to do an analysis of the ODI high innings in view of the momentous tryst with destiny of one colossus at Gwalior. Then I decided not to do so since the results may not be exactly what is wanted by the myriad of Tendulkar fans and I am not ready to read and answer hundreds of comments.

The innings, one of the greatest ever, need not be and is not the best ODI innings ever. The numbers 189/189/194/175/183/149/140/158 et al are floating around. By Tendulkar's own high standards, the 175/138/143/134/98 innings lay claim to being his best. But not to take away from the greatness and perfection of the innings. There might be greater innings but certainly no greater batsman during the past six decades. The gap is widening and soon would be insurmountable.

This is similar to the 400 which, despite being the highest Test innings, is nowhere near innings associated with the numbers 270/153/154/281/149/213/293 et al.

Instead I have come out with an analysis based on the excellent suggestion made by Alex Tierno. This is to determine the successful bowler sequences from a minimum of 3 wkts per Test to 10 wickets per Test. The more I did the work the more I felt that this is an excellent method of determining bowler consistency.

I needed to create a completely new Database of Player-Match records. However this will be very useful since I can do many new analysis without resorting to individual programs. That will be a great bonus.

Let us see the tables starting with 3 or more wickets. I have shown the first three in each classification. Where there are multiple bowlers with same number of Tests, the one with the highest WpT figure is shown.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 3 wkts Warne S.K 39 1576 244 6.3 >= 3 wkts Muralitharan M 34 1555 251 7.4 >= 3 wkts McGrath G.D 17 1718 88 5.2

There is a nice surprise in the bread-and-butter classification of 3+ wickets. Shane Warne has taken 3 or more wickets in 39 consecutive Tests, averaging 6.3 WpT (wickets per Test). Muralitharan has achieved this in 34 consecutive Tests averaging 7.4 WpT. This is a true measure of the consistency which these two great spinners employed throughout their careers.

At this point it is worth explaining that there could be 3+ wkts streaks of more than 17 Tests from either Warne or Murali. I have deliberately shown the top three bowlers, rather than the top three bowling streaks, to broaden the scope of the anaysis.

As expected Muralitharan dominates the other classifications, leading in the 10+, 9+, 8+(shared), 7+, 6+, 5+ and 4+ wickets categories.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 4 wkts Muralitharan M 19 1626 135 7.1 >= 4 wkts Bedi B.S 13 785 78 6.0 >= 4 wkts Waqar Younis 12 1192 86 7.2

Muralitharan has captured 4 or more wickets in 19 consecutive Tests, averaging 7.1 WpT. The Indian classicist, Bedi is next with 13 consecutive Tests at an average WpT of 6.0.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 5 wkts Muralitharan M 14 1670 109 7.8 >= 5 wkts Lee B 11 1824 70 6.4 >= 5 wkts Donald A.A 9 1403 61 6.8

Muralitharan has captured 5 or more wickets in 14 consecutive Tests, averaging 7.8 WpT. The unlucky Brett Lee, is next with 11 consecutive Tests at an average WpT of 6.5.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 6 wkts Muralitharan M 8 1670 68 8.5 >= 6 wkts Warne S.K 7 1582 53 7.6 >= 6 wkts Lee B 7 1824 47 6.7

Muralitharan has captured 6 or more wickets in 8 consecutive Tests, averaging 8.5 WpT. Shane Warne is next with 7 consecutive Tests at an average WpT of 7.6.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 7 wkts Muralitharan M 7 1670 62 8.9 >= 7 wkts Barnes S.F 5 117 41 8.2 >= 7 wkts Turner C.T.B 4 25 39 9.8

Muralitharan has captured 7 or more wickets in 7 consecutive Tests, averaging 9.9 WpT. The great Sydney Barnes, is next with 5 consecutive Tests at an average WpT of 82.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 8 wkts Barnes S.F 4 130 49 12.2 >= 8 wkts Muralitharan M 4 1559 42 10.5 >= 8 wkts Turner C.T.B 4 25 39 9.8

Sydney Barnes has captured 8 or more wickets in 4 consecutive Tests, averaging an amazing 12.2 WpT. CTB Turner ties this with 4 consecutive Tests at an average WpT of 9.8. Muralitharan's streak completes the table.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>= 9 wkts Muralitharan M 4 1559 42 10.5 >= 9 wkts Turner C.T.B 3 26 31 10.3 >= 9 wkts Richardson T 3 46 33 11.0 >= 9 wkts Grimmett C.V 3 249 33 11.0

Muralitharan has captured 9 or more wickets in 4 consecutive Tests, averaging 10.5 WpT. CTB Turner is next with 3 consecutive Tests at an average WpT of 10.3.

Wkts/test Bowler            # of  First  # of  Avge
tests  MtNo   Wkts   WpT

>=10 wkts Muralitharan M 4 1559 42 10.5 >=10 wkts Grimmett C.V 3 249 33 11.0 >=10 wkts Lohmann G.A 2 47 27 13.5

Finally the grand-daddy of all sequences. Muralitharan has captured 10 or more wickets in 4 consecutive Tests, averaging 10.5 WpT. The great Australian leg-spinner, Grimmett, is next with 3 consecutive Tests at an average WpT of 11.0.

Murali's four consecutive 10-wkt hauls are shown below. The concerned year was 2001. This is one record which might, like Laker's 19 wickets in a single test, never be bettered.

1559 vs India       34.1  9  87 8   46.5 17 109 3
1561 vs Bangladesh   9.4  4  13 5   35.3  6  98 5
1567 vs West Indies 53.4 11 126 6   31.3 10  44 5
1570 vs West Indies 23.4  5  54 4   35.5 16  81 6
A final note. Only comments on the subject covered in the article will be published.

An interesting exchange of mails

Posted by: Alex at March 11, 2010 10:06 AM
Ananth - Pl see if you can do analysis to answer the following question: Suppose we restrict ourselves to test matches that have produced a result (including the "tie" tests). What fraction of these featured a winning team bowler taking X wkts/match (where X=7,8,9, ...)? If we split this data into decades (or venue countries), is there any pattern? You could do similar analysis for SR and averages. I feel most result-oriented matches feature a winning team with a bowler who takes at least 7 wkts in that match ... very rare to win with a bunch of bowlers contributing 2-5 wkts each.

[[ Alex That is a lovely idea. It will clearly show whether there is a discernible change in the winning methods of Test teams. Will have to do a special program but will be worth it. Thanks Ananth: ]]

Posted by: Jeff at March 12, 2010 9:37 AM
@ Alex
These are the figures that I have:
1269 tests have produced a winning result
On winning teams, the number of times players took & or more wickets are:

7 = 390 times (0.31 per match)
8 = 311 (0.25)
9 = 187 (0.15)
10+=269 (0.21)
For players on losing teams, the numbers are:
7 = 156 (0.12)
8 = 137 (0.11)
9 =  46 (0.04)
10+= 65 (0.05)
Not surprisingly, it's more twice as likely for a player on a winning team to 7 or 8 wkts in a match and up to 4 times more likely for them to take 9+ wkts
There have been 682 draws (therefore 1364 drawing teams), and the numbers are:
7 = 197 (0.14)
8 =  81 (0.06)
9 =  49 (0.04)
10+= 65 (0.05)
These are very similar to the figures for the losing teams, particularly for 9+ wickets.
Obviously this is only the overall figures but I found it interesting that the results for drawn matches were so low. This is the decade by decade split for winning/tied teams.
Note that I have the number of times 7,8,9,10+ wkts have been taken. Some of these will have happened in the same match, so the following figures will slightly over estimate the % of matches they occur in, but the figures should give a good picture of the fact that the % of winning teams with players taking 7 wkts has been increasing over time but the % of winning teams with players taking 10+ wkts has been decreasing.
Era      %7wkts %8wkts %9wkts %10+wkts

PreWW1 25% 31% 15% 26% Inter War 31% 24% 11% 32% 40s/50s 26% 28% 18% 23% 60s 20% 20% 19% 15% 70s 31% 17% 18% 18% 80s 31% 22% 15% 26% 90s 33% 32% 15% 19% 2000s 36% 21% 12% 16%

Jeff
[[That is wonderful. I am tied up with so many things that I could have done justice to Alex's excellent suggestion only after a few days. I will immediately publish your response. Within an hour I will extract the table and put it up on the blog itself. Many thanks and the non-existent hat is off in admiration.
Ananth:]]

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

RSS Feeds: Anantha Narayanan

© ESPN Sports Media Ltd.

Posted by William BISHOP on (April 11, 2010, 1:52 GMT)

murali could never be the greatest spinner of all time,take out his wickets from zimbawe and bangladesh,also tell me how is his record in australia. also note how many overs he bowled in each test, by the law of averages hewould obviously get wickets.There were no other top class bowlers in his side to get agreat amount of wickets.Please dont mention Vaas,check his record ,it is ordinary,so muralli had no team compitition like warne or ouadir.plus i am not so sure he bowl like the other spinners,his stylewas funny.

Posted by Alex on (March 15, 2010, 15:55 GMT)

Jeff --- thanks. I wonder if the slow scoring rate reflects the batting standard of the 60's. It is possible that the 60's featured relatively few truly great cricketers ... maybe a few great ones retired by mid/early-60's while a few great ones just about started their careers in the 60's --- that would be a clear parallel with the 00's.

10+ wkt hauls are quite difficult: one match wonders are more likely candidates than great bowlers in great teams ... Massey, Hirwani, Siva, the list goes on.

Posted by Jeff on (March 15, 2010, 10:45 GMT)

One other thing I noticed:

Murali accounts for 16 of the 57 10 wkt hauls made in winning efforts in the 2000s – a staggering 28% of them.

This is by far the biggest proportion that one man has contributed to any period - his nearest challengers are Grimmett in the 20s/30s ( 6 out of 28 or 21%) and Hadlee in the 70s (4 out of 21 or 19%)

So, if you take Murali out of the analysis, then the 2000s look even worse for 10 wkt hauls, only around 1 in 10 victories features a bowler taking 10 wickets – the figure was around 1 in 3 matches before WWII and about 1 in 5 for the second half of the 20th century

Posted by Jeff on (March 15, 2010, 10:34 GMT)

@ Alex

I think the 60s was a more benign period for cricket in general.

The stats suggest is was a time of great conservatism.

Bowling strike rates were at an all-time high – an average of a wicket every 81 balls (compared to about every 66 balls in the 2000s)

This would suggest that things were much easier for batsmen back then.

But, if you also look at bowling economy rate, you will see it was at 2.38 runs per over – slightly higher than in the 50s, but lower than every other period (eg. economy rate is over 3 in the 2000s.)

So this suggests it was harder for batsmen to score runs in the 60s.

And looking at the combined impact (ie batting average) then you’ll see it wasn’t much different to any other post WW1 decade at 32.10 (for comparison, batting average was 31.90 in the 70s, 32.09 in the 80s & 32.15 in the 30s.)

Posted by Alex on (March 13, 2010, 6:35 GMT)

Jeff --- thanks very much. So, it appears that the 60's either featured relatively few outstanding bowlers or featured a bunch of them operating together & sharing spoils: in 26% of its results, the winner had no bowler who took >6 wkts/match. I think former is the former case since its number for X=8,9,10+ are comparable to those of the 70's (combined: 54% to 53% ... the 00's are even poorer at 49%).

The percentages for X>7 are 67% interwar (probably too much dependence on strike bowlers), 69% in 40/50's (maybe many really great bowlers), and about 64% in 80/90's (many really great bowlers). Even though 00's is poor on it at 49%, it scores 31% for X=7 while 60's score only 20% on X=7. So, it is possible that the 60's was a more benign period for batsmen than the 00's. So much for Pollock's 63 and Sobers' 58!

Posted by Ramesh Kumar on (March 12, 2010, 11:03 GMT)

Excellent idea from Alex. I feel that match performance in tests get real meaning mainly with bowlers and not with batsmen. Alex's idea can be looked at from a specific bowler's perspective and seperately from a team's perspective. The first one can reflect some champion performances and the latter one will give an idea about the composition of great winning teams. Once this template is designed, it can be used to compare with lost and drawn matches.We might stumble upon an effective criteria for assessing match performance of individual bowlers as well.

Ramesh Kumar [[ Ramesh Jeff has done the analysis and this has been posted. Ananth: ]]

Posted by Jeff on (March 12, 2010, 10:18 GMT)

Following on from my previous post, this is the decade by decade split for winning/tied teams.

Note that I have the number of times 7,8,9,10+ wkts have been taken. Some of these will have happened in the same match, so the following figures will slightly over estimate the % of matches they occur in, but the figures should give a good picture of the fact that the % of winning teams with players taking 7 wkts has been increasing over time but the % of winning teams with players taking 10+ wkts has been decreasing.

Era %7wkts %8wkts %9wkts %10+wkts PreWW1 25% 31% 15% 26% Inter War 31% 24% 11% 32% 40s/50s 26% 28% 18% 23% 60s 20% 20% 19% 15% 70s 31% 17% 18% 18% 80s 31% 22% 15% 26% 90s 33% 32% 15% 19% 2000s 36% 21% 12% 16% [[ Jeff That is wonderful. I am tied up with so many things that I could have done justice to Alex's excellent suggestion only after a few days. I will immediately publish your response. Within an hour I will extract the table and put it up on the blog itself. Many thanks and the non-existent hat is off in admiration. Ananth: ]]

Posted by Jeff on (March 12, 2010, 9:37 GMT)

@ Alex

These are the figures that I have:

1269 tests have produced a winning result

On winning teams, the number of times players took & or more wickets are:

7 = 390 times (0.31 per match) 8 = 311 (0.25) 9 = 187 (0.15) 10+ = 269 (0.21)

For players on losing teams, the numbers are:

7 = 156 (0.12) 8 = 137 (0.11) 9 = 46 (0.04) 10+ = 65 (0.05)

Not surprisingly, it's more twice as likely for a player on a winning team to 7 or 8 wkts in a match and up to 4 times more likely for them to take 9+ wkts

There have been 682 draws (therefore 1364 drawing teams), and the numbers are:

7 = 197 (0.14) 8 = 81 (0.06) 9 = 49 (0.04) 10+ = 65 (0.05)

These are very similar to the figures for the losing teams, particularly for 9+ wickets.

Obviously this is only the overall figures but I found it interesting that the results for drawn matches were so low.

Posted by Alex on (March 11, 2010, 10:06 GMT)

Ananth - Pl see if you can do analysis to answer the following question: Suppose we restrict ourselves to test matches that have produced a result (including the "tie" tests). What fraction of these featured a winning team bowler taking X wkts/match (where X=7,8,9, ...)? If we split this data into decades (or venue countries), is there any pattern? You could do similar analysis for SR and averages. I feel most result-oriented matches feature a winning team with a bowler who takes at least 7 wkts in that match ... very rare to win with a bunch of bowlers contributing 2-5 wkts each. [[ Alex That is a lovely idea. It will clearly show whether there is a discernible change in the winning methods of Test teams. Will have to do a special program but will be worth it. Thanks Ananth: ]]

Posted by Behram on (March 10, 2010, 10:55 GMT)

Too much has been read into my comment ,which was essentially the proverbial molehill. My comment was addressed to “Ashwath”’s statement: “This almost proves that murali is statistically the best bowler of all time”. As a standalone analysis, and an object of mild curiosity, this type of analysis is fine. I was merely referring to the gross inaccuracy in interpreting it the way “Ashwath” has done. [[ Behram Sincere apologies. I have always emphasized that the analysis normally cover one aspect and these should not be telescoped upwards to something beyond the scope. There have been comprehensive player analyses covering almost all aspects. Ananth: ]]

Comments have now been closed for this article

ABOUT THE AUTHOR

Anantha Narayanan
Anantha spent the first half of his four-decade working career with corporates like IBM, Shaw Wallace, NCR, Sime Darby and the Spinneys group in IT-related positions. In the second half, he has worked on cricket simulation, ratings, data mining, analysis and writing, amongst other things. He was the creator of the Wisden 100 lists, released in 2001. He has written for ESPNcricinfo and CastrolCricket, and worked extensively with Maruti Motors, Idea Cellular and Castrol on their performance ratings-related systems. He is an armchair connoisseur of most sports. His other passion is tennis, and he thinks Roger Federer is the greatest sportsman to have walked on earth.

All articles by this writer