ODI batsmen against bowler groups: across ages
Measuring ODI batting performances by the quality of the bowler groups faced

Viv Richards: the best average against the top bowling group • AllSport UK Ltd
A few months back I had come out with an article on Test batsmen by bowling quality, in groups. This was one of the best received of all my articles since the analysis took Test batting into hitherto unchartered seas. Many new insights were drawn from the analysis. I think it is time I do a similar analysis for ODI batsmen also since the bowling quality varies considerably across teams and years. The average runs scored by batsmen in their careers is also quite high and an analysis like this will let us look at the batsmen with a new perspective.
I have summarized below all relevant facts related to this analysis. First let me emphasize that this is not a ODI innings Ratings analysis. There are many other relevant factors which would have to be considered in such an analysis. I have not done so in this analysis which is centred on Bowler quality. I would appreciate if the readers do not keep on repeating again and again that other relevant factors such as Pitch type, Innings status at entry, Result, Match importance, Bowler recent form, Innings target et al, have not been included. That would be counter-productive.
1. The Bowling quality index (BQI) is based on Career-to-date values. This is the most dependable and accurate of the bowling measures. There is no situation where the Ctd figure is not the appropriate one. Coupled with the fine-tuned handling of established bowlers described later, this works very well. This takes into account the way a bowler's career shaped up.
2. The BQI is based on the Bowling average. In Test matches the bowling strike rate has greater relevance. However in ODIs, both strike rate and bowling accuracy (RpO) have equal importance and the Bowling Average is a perfect representation of this. Very good averages of say, 25.0, can be reached by a combination of 60 and 0.41 or 50 and 0.5 or 40 and 0.62. All these, patently different, bowlers are considered similar in this analysis. Individual match circumstances might require bowlers with varying attacking and accuracy-related skills, but, in general the average takes care of all conditions.
3. The BQI is based on the actual bowlers who bowled in the particular innings. This is very important. If Imran Khan played as a batsman, to that extent, the bowling attack would be less strong.
4. The BQI is determined using the modified reciprocal method suggested by Arjun Hemnani which irons out the imbalance created by weak fifth bowlers.
5. I have taken care of top bowlers during their initial Initial figures for bowlers with career haul of 100+ wickets. Whatever be the Ctd figures for these qualifying bowlers, their Ctd bowling average will be fixed at their career bowling average levels. This takes care of both situations: Walsh capturing 10 wickets at 50+, nearly 20 more than his career average and Mendis, at one point capturing 25 wickets at 9.83. Of course once any bowler crosses 50 wickets, their Ctd figures will apply.
For the bowlers who have not captured 100 career wickets, their Ctd bowling averages below 50 wickets is pegged at a minimum of 40.0. Makes eminent sense.
6. The computed BQI values will be used only for innings of 10 overs or more. For shorter innings the minimum BQI value is pegged at a minimum of 30.0. This is to prevent situations like Wasim Akram and Waqar Younis bowling 6 overs between them. The BQI would be a very low number.
7. The BQI is reduced by 5% for Home games and increased by 5% for away games. Reader should remember that the lower the BQI, the more potent the attack is. 5% either way is ample and provides some compensation for batsmen playing away. In general this concept is fine and works well in most cases.
It is possible that the visiting team has the right bowlers and can exploit the "away" bowling conditions. However there is no denying that, in most cases, the home bowlers would have the advantage of familiarity with and knowledge of local conditions. Great examples are the recent whitewashes in England and India and the way West Indies are struggling in Bangladesh.
8. No period-based adjustment is done. Such adjustment is relevant only for determining team strength values. If the period was a great one for the bowlers, as the 1971-84 was, it was a tough one for the batsmen and this is taken care of by leaving the relatively lower BQI values as they are. It is obvious that the runs scored during 1971-1984 were more valuable than the runs scored in more batting-friendly conditions later.
Finally the bowling attacks are classified into 5 groups, as described below. The fifth group was necessary to separate the really weak bowling attacks.
There have been 6302 qualifying innings until the fifth ODI between India and England which was played on October 25. The underlying idea is that the middle group should have about a third and the other groups symmetrically lower. In view of the profusion of weak bowling attacks, the first and the last would not necessarily have similar % shares. There may be a subjective element in this part of the exercise but that cannot be avoided. Around 28% for the first two groups means that at any time there are 2-3 really good bowling attacks makes eminent sense. The other cut-offs follow logically. The group cut-off details are given below.
Group B Q I # of Inns %
Pakistan has a few bowling attacks around the 23 mark, as also West Indies of the 1980s and Australia of the 2000s.
The average BQI for this huge sample is 34.4 and the median is at 33.6. This indicates a fairly balanced distribution of values. The Standard Deviation is 5.87. I have explained the whole concept of determining the BQI with the following examples.
First is Match 1833 between Pakistan and New Zealand, played at Karachi during 2002. In the New Zealand innings, Wasim Akram (Ctd 456 @ 23.86) bowled 7.0 overs, Waqar Younis (Ctd 372 @ 23.54) bowled 6.0 overs, Abdul Razzaq (Ctd 136 @ 24.68 (but career 31.84!)) bowled 4.0 overs, Shoaib Akhtar (Ctd 99 @ 20.68) bowled 9.0 overs and Saqlain Mushtaq (Ctd 270 @ 20.90) bowled 4.0 overs. Through the reciprocal method, the the weighted BQI starts life at 22.44. This is multiplied by 0.95 (this being a home game for Pakistan). The final BQI value is 21.32 which places this attack as the best ever one. Any runs scored in this particular innings will get into the highest classification. Astle's 25 (out of 122) might not figure in anyone's list of the best ODI innings. However it was made against the best ever bowling attack which took the field.
The second is Match 132 between West Indies and Pakistan, played at Sydney. Holding (Ctd 41 @ 18.44, taken as career 21.37), Roberts (Ctd 55 @ 18.96), Marshall (14 @ 24.14, taken as career 26.96), Garner (Ctd 35 @ 25.31, taken as career 18.85) and Richards (Ctd 21 @ 37.57, taken as career 35.83) all bowled 10 overs each. The base BQI is 22.98. This is multiplied by 1.00 (this being a neutral ODI). The final BQI value is 22.98 which puts this attack into the top drawer. Any runs scored in this particular innings, say Imran's 62 will get into the top classification.
I have got into details here so as to give the readers a clear idea of the calculations. I have selected two of the best ever bowling combinations put on the field. I have also selected one in which all five bowlers had crossed 50 wickets and their Ctd values were impeccable and another attack in which four bowlers (three greats amongst them) had just started their careers. This will show that the great bowlers have always been given their due credit.
There is so much data available that even the organization of the article is getting into trouble. I can only present in the article a certain amount of data. The serious reader should download the complete files and read the same. I have given below what I would be presenting within the article.
1. Top 20 batsmen for group 1, the top one. Ordered by batting average. 2. Top 20 batsmen for group 2, the second best one. Ordered by batting average. 3. Top 20 batsmen for groups 1/2, the groups which matter. Ordered by batting average. 4. Top 20 batsmen for group 3, the middle and most-populated. Ordered by batting average. 5. Top 10 batsmen for groups 4. Ordered by batting average. 6. Top 10 batsmen for groups 5, the weakest one. Ordered by % of career runs scored. 7. For selected batsman, their group-wise distribution of runs scored and % of career.
Let us look at the tables. First the Group tables based on Batting average. The batsman should have scored a minimum of 750 for Group 1, 1000 for Group 2, 2000 for Group 3, 1000 for Group 4 and 1000 runs for Group 5 to be considered. I cannot use the same cut-offs across bowler groups since the population sizes vary considerably. For instance, taking 1000 as cut-off for the group 1 will let us have only 13 entries. It should also be noted that Runs scored should not be a criteria for ordering since that is a measure of longevity.
This analysis covers all matches upto ODI # 3210, the fifth ODI between India and England. While a few days have passed since the third ODI between Saf-Aus was played, it was too much of an effort to re-do all tables and article.
Batsman Team BG CRuns Inns Nos Runs % Avge
Richards suffers a little bit since the best bowling attacks during his time were from his part of the woods. He still has done very well and averaged 54.38 against the top group. The runs are low but that is an indication of the number of matches played. However it should be seen that he has scored 12.9% of his runs against the top group. Steve Waugh and Gary Kirsten have averaged over 50 and have also scored more than a sixth of their career runs against the top group. It helped that the other respective bowling attacks were very good.
Pietersen is a revelation. Nearly a quarter of his runs have been against the top attacks at an average of 52.11. This single fact is enough ammunition to show the futility of using Batting average as an omnipotent analysis factor. Pietersen has a batting average barely reaching 50 but his runs seem to have a much higher value. Ponting has a lower % but a near-50 average.
Imran Khan's 23.8% of his runs against the top group is nearly as much as that of Pietersen and that too at an average of 44.1. This deserves a special mention especially as he was not the leading batsman of Pakistan.
Dhoni has not scored many runs but he has scored 12.3% of his runs at a high average of 47 against the top bowlers. He is no doubt helped by a slew of not outs. Dravid clocks in with a very respectable 18.3% and average of 42.38. Ganguly has a similar average but lower %. The surprise is that Tendulkar has just about crossed the datum % of 11.25% but a reasonably low average of 35.16. This is possibly because of his opening the batting. However it must be remembered that Ganguly was also in a similar position.
Batsman Team BG CRuns Inns Nos Runs % Avge
Symonds has scored 23.3% of his runs at a very high average, a late-order batting benefit, of 51.65. Sangakkara has done very well, scoring over 2000 runs, 21.0% of his runs, at a very creditable 50+ average. Dhoni also has a near-50 average, slightly below his career average. as does Bevan. Tendulkar has asserted his class against this strong bowling group, scoring nearly 4000 runs, 21.9% of his career runs at an average of 46.83, better than his career average.
Batsman Team BG CRuns Inns Nos Runs % Avge
Now for a special table, the elite group table. In this I have considered the top two bowling groups and selected players who have crossed 2000 runs against the two groups together. This table is ordered by the batting average. As such it represents a table of quality batsmen against quality bowlers.
Bevan and Dhoni are in the top two positions. But they have been helped by a high number of not outs. Hence we should take Ponting as the real top batsman. He has scored near;y 4000 runs, which is 29% of his career runs at an average of 43.47. Truly outstanding batting. Gary Kirsten has averaged 41.56 and scored nearly a third of his career runs against this double group. Tendulkar makes up for his group 1 under-performance and clocks in with a creditable 41.41, while scoring over 6000 runs and just above a third of his career runs. This indicates that both Ponting and Tendulkar have done very creditably against top quality bowling. Haynes is the only other batsman to cross 40. Readers may wonder where Richards, who topped Group 1 is. The fact is that he does not meet the higher cut-off point of 2000 runs for Groups 1 & 2 combined.
Batsman Team BG CRuns Inns Nos Runs % Avge
Hussey and Bevan, no doubt aided by a high number of not outs, are in the top three positions in this staple group. Richards averages 55+. Michael Clarke is the only other batsmen with a 50+ average. Note the very high % of career runs for all these players. Tendulkar's group 3 performance is almost identical to his groups 1/2 performances, at a higher average.
Batsman Team BG CRuns Inns Nos Runs % Avge
Now we get into the weaker bowling groups. Note the number of 50+ averages. Many modern batsmen have feasted on these below-average bowling attacks.
Batsman Team BG CRuns Inns Nos Runs % Avge
The last group is the buffet-lunch group. I have ordered this in a different sequence, the % of career runs. This figure is essential to see how much the batsmen got against the really weak bowling attacks.
As could be expected the top of the table is dominated by players from weaker countries who almost always play against weaker attacks. The top three players have got more than 50% of their runs against very weak attacks. The real surprise is Zaheer Abbas, whose high batting average is now on shaky ground, he having scored 42% of his runs against the lowest group. Same with Srikkanth, whose bubble is blown a little, with over a third of his runs against the buffet-lunch bowlers. And Sidhu and Martin Crowe and Ijaz and Dean Jones.
At the other end, raise your hat for Dravid who has scored only 14% of his runs in this group. The three Sri Lankan stalwarts have got sub-14%. But let us all raise the hat and toast Kallis whose % here is the lowest amongst all established batsmen, a mere 10%. This should put to bed all theories on his scoring against minnows.
In terms of averages, Dean Jones has really feasted with an average of 80+. The average table is led by three Australians of the previous generation. Ganguly has not done his averages any damage by clocking in 60+ here. Tendulkar, with an average of 47 does not seem to have benefited much against these weaker bowling attacks. Lara does not even appear in the top-20 of the averages table.
Now for the group-wise runs and % of career runs for selected 25+ batsmen. The complete file is available for downloading.
Batsman Team CRuns G1-Runs-% G2-Runs-% G3-Runs-% G4-Runs-% G5-Runs-%
Tendulkar seems to have mirrored the overall % pattern, as has Ponting. Note Inzamam's figures. Possibly because the best bowling attack in the world was his team's, he has a lop-sided bottom-heavy distribution. Kallis has scored lower % both against the best and worst attacks and he has a centre-heavy distribution. Dravid has scored fair bit against top attacks while Lara's follows Tendulkar's pattern. Surprisingly, as has Richards. Intriguingly, Hussey's figures against top bowling attacks has been quite below-average. Flower and Shakib-Al-Hasan have low numbers against top attacks since they, Shakib especially, play quite often against weak sides.
This is not an analysis from which the analyst could make finite conclusions. The readers should read and understand the methodology and tables and then come with their views. To view/down-load the complete Team Strength related tables, please click on links given below.
Group tables - by Batting average: please click/right-click here.
Group tables - by Runs scored: please click/right-click here.
Batsman table - by Group (for all 2000+ batsmen): please click/right-click here.
BQI table - ordered by Group/BQI (for all 6302 innings): please click/right-click here.
Batsman-Bqi average across career: as required by Arjun (for all 2000+ batsmen): please click/right-click here. These are in fact the Test tables.
This time the ODI tables. Batsman-Bqi average across career: as required by Arjun/Mahendran (for all 2500+ batsmen): please click/right-click here.
Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systems