Bowling September 17, 2010

ODI Bowlers: a totally new look through BCG charts

This article is a completely different graphical look at the ODI bowlers and is a continuation of a similar article on ODI batsmen
48

Glenn McGrath: in the right BCG quadrant almost all the time © Getty Images

This article is a completely different graphical look at the ODI bowlers and is a continuation of the similar article on ODI batsmen.

Just to recap, Bruce Henderson of BCG (Boston Consulting Group) had created these charts during 1968 to study the Growth-Share aspects of products/business units. This is an excellent way to study two related variables together. These are plotted on a graph which is split into four equal (or unequal) size quadrants. The placement of a particular player, gives excellent insight into the bowler's position in the galaxy of bowlers. However please do not forget that this is clearly a two-dimensional graph between two related variables. Also these are all career figures.

Bowling is a far more cleaner and crisper playing aspect which lends itself to excellent analysis. There are only two independent variables, bowling strike rate and bowling accuracy, in the form of Rpb. These two together can be used to generate the bowling average, which is a single measure incorporating the two constituent parameters very clearly, unlike the batting average which has the dicey not outs concept embedded within. The bowling strike rate is represented in X-axis and the bowling accuracy (Rpo or Rpb, it does not matter what we take) in the Y-axis. The only special requirement is that, in bowling both variables have reverse-effectiveness in that the lower these are the better the bowler is. hence I have laid down the axis from the highest to lowest values.


© Ananth Narayanan


The above represents a typical BCG chart. The bowlers in the top-right quadrant, the red one, are the "Top bowlers". They are to the right of the Bowling strike rate line and above the Rpo line. The ones in the bottom right quadrant, the green one, are the "Attackers". They capture wickets quite frequently but go for plenty of runs. Certainly an asset, but could do better. Similarly, the top left quadrant, the blue one, contains the "Defenders". They take more balls to capture a wicket but are miserly. They are equally valuable as the Attackers. The bottom left quadrant, the orange one, represents the "Also rans". They fall behind in both areas and lag behind the others.

As I explained in the batsmen article, the two central dividing lines can be drawn in two ways. One is to draw the same right in the middle. However this does not take into account the distribution of values. The alternative method is to draw the lines around the median value so that we get around half the bowlers on top of the mid line of the Rpo and around half the bowlers to the right of the mid line of the Bowling strike rate line. This leads to unequal quadrants but would make analysis of the bowlers far more meaningful. Let me add that the drawing of the asymmetrical central lines is my own idea and most of the BCG charts have only centrally located divider lines. However my idea of asymmetrical dividing lines ensures a fairer distribution of players across quadrants.

Finally the chart is drawn on two criteria. The top wicket-takers and the top bowlers based on bowling averages, the minimum wickets requirement for the later selection being 100 wickets.

I have also changed the graph presentation method. I have used Gaurav's suggestion and drawn fixed diameter circles supported by numbers. There is a legend at the right hand side to link the numbers to the bowler names. It has come out very well and the graph is now uncluttered.

The first chart is drawn with the wickets captured as the criteria. 200 wickets are the cut-off for selection. Anything fewer will clutter up the graph. Already I feel we are over-populated. The median Strike rate is 38 and the median Rpo is just short of 4.3. The dividing lines are drawn around these figures. These lines split the distribution approximately equally on either side. Let us now look at the chart.

Graph of runs scored
© Ananth Narayanan


The "Top performers" are led by Glenn McGrath and closely followed by Muttiah Muralitharan and Wasim Akram. Donald and Saqlain have excellent strike rates and quite good Rpo figures. Hence they are also in this quadrant. Warne and McDermott have taken more balls to capture a wicket but are also comfortably in the top bowlers quadrant. It is difficult to question the credentials of any of these ODI greats.

The "Attackers" group is led by Lee, Waqar Younis and Shoaib Akhtar. Incidentally these two Pakistani greats are almost like Siamese twins with almost identical figures (30.5/4.69 and 30.7/4.69). Agarkar has excellent strike rate but is a millionaire when it comes to conceding runs. Ntini has similar strike rate but is far more economical. Srinath, Gough and Zaheer just about make it to this quadrant.

The "Defenders" group is led by Ambrose and Pollock. they are followed by Kapil Dev and Walsh. A few other modern spinners, Harris, Vettori, Kumble and Harbhajan are at the border-line.

The "Also rans", has Afridi and Jayasuriya as prominent members. These are followed by Streak, Razzaq, Cairns and Kallis.

Lee is the outlier as far as the strike rate is concerned with a sub-30 Bpw figure. The three Pakistani greats have just over 30. Ambrose is the outlier as far as Rpo is concerned with a sub-3.5 figure. Pollock and Kapil Dev follow next.

The second chart is drawn with the Bowling average as the criteria. 26.00 is the cut-off with a minimum of 100 wickets. The median Strike rate is around 34 and the median Rpo is around 4.2. The dividing lines are drawn around these figures. These lines split the distribution approximately equally on either side. Let us now look at the chart. The selected bowlers are distributed around the whole graph quite well.

Graph of runs scored
© Ananth Narayanan


This selection is far more stringent because of the dual cut-offs. There are only two bowlers in the "Top performers" quadrant. Only McGrath and Donald make the cut. Their positioning is also intriguingly close to the middle lines. McGrath is more economical but takes a few extra deliveries per wicket. Donald is the other way around. Two of the greatest of ODI bowlers ever.

Lee leads in the "Attackers" group and is closely followed by the Siamese-twins, Waqar Younis and Shoaib Akhtar. Saqlain is also very close to the top performer quadrant. Bond is another exciting new entry who is in a similar position. Maharoof and Flintoff are also in this group.

Lillee and Muralitharan are the leaders in the "Defenders" group and these two are quite close to the top performers. Joel Garner is a well-deserved new entrant here and has an outstanding economy rate. Wasim Akram is well-placed here. This group also has other wonderful ODI bowlers like Hadlee, Pollock, Ambrose and Holding.

Warne is in the fourth quadrant but is very close to the Rpo dividing line. Fleming is close to the Strike rate dividing line. Because we have taken only bowlers of average 26 and below, there are very few poor performers.

Lee is the outlier as far as the strike rate is concerned with a sub-30 Bpw figure. Bond follows him. Garner is the outlier as far as Rpo is concerned with an amazing 3.1. Hadlee and Holding follow him.

There is a clear trend here. Where the bowlers are clustered together, the concentration is on the middle performance quadrants. There are very few bowlers in the high performance and low performance quadrants. This trend is different to the wickets based graph where the bowlers are scattered all over the graph. Hence there are more players are present in the extreme performance quadrants.

I have also drawn the charts for the top players by Strike rate. Unlike the corresponding batting chart, this has got excellent distribution all over the graph. This is mainly because the Bowling strike rate values are closely bunched together. The top group is led by Donald, Saqlain and Bond. The chart can be seen below.

Graph of runs scored
© Ananth Narayanan


The last one is the chart of the top bowlers by Rpo. Here also there is good distribution. The leading bowlers are Lillee, Hadlee, Holding, McGrath, Wasim Akram and Muralitharan. What a collection of greats. The chart can be seen below.

Graph of runs scored
© Ananth Narayanan


I will attempt a similar analysis on Test Batsmen/Bowlers. I think I have got the two variables for batting identified. That will be the Batting average and Average career weighted bowling quality faced. The results seem to be coming out very well. Readers are welcome to give their suggestions.

An important announcement to the readers. In one of my comments I had mentioned that I would create an open mail id to which readers could send their suggestions. To start with I would appreciate if readers can send in their suggestions on batting and bowling performances in the third innings or the Test batting BCG charts. I will complete my work and depending on the reader responses will incorporate a few popular performances amongst these. Please note that this is a one-to-one communication and the contents will not be published. Please continue to use the blog posting method for the comments you want to be published. This is not my mail id and has been created only for this purpose. To separate the spam, it will be a nice idea if all readers can follow a simple idea of making their title as "It Figures Blog: ..............".

The mail id is ananth.itfigures@gmail.com

Since the readers would have to use a mail route I give the readers my assurance that the mail id is safe and will never be used by me for anything other than communicating with the reader specifically. This will not be part of any group mail nor will mails be cc'd.

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

  • Migara on October 5, 2010, 16:37 GMT

    Nice work! But I'd think you should have standardised Averages, SRs and ERs for the comparison. For an example during 80s the ER of most bowlers were 4.4 (Avg total 220) compared to that of 00s where ER os close to 5.5 (Avg total 275). You can compare each parameter of a player to his peers during the period he played. That will allow the distortions taken by the trends in the games to neutralise. It's not very difficult exercise as well because it involves a simple function in a spread sheet. [[ I have already done very extensive work on Peer comparisons. Pl see last year's article. Ananth: ]]

  • Amit on September 24, 2010, 20:50 GMT

    Hi Ananth, That last 'Anonymous' was me. Didn't realize it would let me post on my laptop when my name is not entered (different from how it works when I post from my blackberry). If you have control over it, please feel free to change the last Anonymous (addressed to Alex) to my name. And, keep the wonderful work going. Thanks. -Amit.

  • Anonymous on September 24, 2010, 17:27 GMT

    Alex, thanks for the correction and nice observation on the yorkers. The reverse is also true - if you try bowling yorkers and dont get it right, all your overpitch balls and full-tosses with get you miserable economy rate. Case in point - Agarkar.

  • Alex on September 24, 2010, 9:30 GMT

    Amit - Nice observation. A minor correction: Garner began as the strike bowler (position 1 or 2) at age 31, starting WC '83. He remained so till his retirement in '87 barring a couple of ODIs.

    The era of treating the first 15 overs as the slog overs really started with Greatbatch of NZ in the '92 WC although a few openers such as Srikanth had always treated this phase (or any) as slog. So, for the early strike bowlers, outstanding SR says more than outstanding RPO ... to me Garner's RPO=3.1 is more like RPO=3.5 in 90's and his SR=36 is more like SR=33 in 90's.

    Across all eras, I think the bowlers who excel in the death overs possess great yorkers: Garner, Thommo, Malinga, Holding, Waqar.

  • Amit on September 24, 2010, 4:09 GMT

    Hi Ananth, I wonder if you have the info about when these bowlers bowled in the course of the innings. Most good fast bowlers have the luxury of bowling the first few over (when economy rate was typically lower, alleast until Gilchrist, Hayden and Sehwag came along), but is offsetted having to bowl at death overs. If my menory serves me right, Garner bowled at first change (except WC83) and then 'always' at the death overs. I remember him him bowling maidens in final overs to win tight games. It would be good to know which of these bowlers bowled with a disadvantage and still excelled. -Amit. [[ Amit This sort of information is available only during the past 10 years or so. It is also not available in the public domain. Ananth: ]]

  • Alex on September 22, 2010, 13:56 GMT

    Ananth - re all-rounders, you seem to apply the traditional definition. I like to think of great wkt keepers, great captains, and great fielders as all-rounders as well. [[ Alex For this type of two-dimensional graph based analysis it is better to stick to the traditional definition. For a different type of analysis such as multi-disciplinary contribution to team cause I would look at the wider definition. Ananth ]]

    A quote of I Chappell who loves Mark Waugh: In a late 90's ODI distinguished by Steve Waugh's brilliance, Tony Greig commented on Steve Waugh "this guy has to be the best all-rounder in the world" to which Ian Chappell replied "Tony, he is not even the best all-rounder in his own family!"

  • craigmnz on September 22, 2010, 7:25 GMT

    Anath

    I have to admit that I enjoyed this article (and its batting counterpart) simply because of the way you're using different (and to me, new) techniques. I might have been a first-class scorer for 13 seasons (3 with Wellington, 10 with CD - when I moved to the Wairarapa) but that doesn't make me a stats geek.

    I'm interested in a similar analysis of all-rounders, mainly because I have this nagging doubt that Richard Hadlee doesn't quite belong in the same category as Sobers, Botham or Imran. Undoubtedly one of the great bowlers (seeing his 7 for 23 remains one of my most cherised memories - I wagged school to see it) but I doubt that the NZ selectors ever seriously considered playing him as a specialist batsman in a big match.

    The obvious axes for all-rounders would be batting and bowling average but what about comparative strike rates? The explosive all-rounder as opposed to Mr Reliable? Just a thought... [[ Craig I myself have pushed Test all-rounders ahead of the Test batsmen because of the intricacies it offers. After another very interesting post late this week I will do the All-rounders' article next week. Totally dispassionately speaking, which all-rounders could have been 100% certain of playing in either capacity. Kallis: probably not as a bowler. Kapil: as a batsman, only because of the weakened Indian batting. This argument is applicable to Hadlee as well. If we take the next step, both could not have played for Australia as batsmen. Botham: almost the same as Kapil and Hadlee. And what about Vettori. His batting average is better than Greatbatch, McClean, AD Nourse, Mahanama, Srikkanth, Grant Flower et al. But he probably would not make to many teams as no.6. That leaves us with two players who probably could have played in any team in either capacity. Sobers (poor bowling average but an excellent double-pronged skill) and Imran. And I would venture to add Miller.. However let us wait for the analysis. Ananth: ]]

  • Ravi on September 20, 2010, 18:18 GMT

    Another good article Ananth. Here are my 2p for observations from the charts: 1. Batsmen in the first ODI decade were more orthodox (not necessarily slow scoring)in their approach- build, consolidate-slog for 20 overs each. Also true for 50-over games in the first decade. Possibly therefore seamers had lower RPO than modern ones. Old gen spinners went for more RPO than seamers. GGilmour WC75 sf, Garner WC 79final, W Davies7-51 as examples among many such excellent ones. [[ Ravi Price, Murali, Maninder and Harper are the only bowlers who have bowled 500+ overs and have an Rpo value below 4.0. Only Maninder belongs to the previous generation. Ananth: ]] 2. Fully fit bowlers or those playing in 1 or 2 formats (not all 3) have done well on the SR. To be fair T20 isn't applicable to any except the very recent bowlers. 3. One can't failto notice the WI domination in the firt decade- Holding & Garner's excellent economy rate but slightly lower SR than the very best is a reflection of the slightly more relaxed nature of batting. [[ Also the reduction of risks taken. Ananth: ]]

    4.Wonder if we see any evidence of the effect of the no. of minnow teams? [[ We have minnows such as East Africa even in the 1970s. Ananth: ]]

    5. Just a question: Were there ever 8-ball overs in ODIs? [[ Yes. About 10 or so matches at the beginning, including the first one. However that does not matter since I store the balls information and all my Rpo values are for 6-balls. Ananth: ]]

  • Anand on September 20, 2010, 15:16 GMT

    Ananth:

    Sriram's suggestion on having dividing lines non-parallel to the axes may be an excellent idea for the BCG graphs on all rounders. I am sure with your knowledge and understanding you can some up with some very interesting analysis of all rounders. One thing that has made me wonder is that neither the West Indies team of the 80's nor the Aussies of the 2000's had the top all rounder of their times. I mean in the 2000a Australia had Symonds, Harvey, Watson and in 80s Marshall and Holding were no push overs with the bat but as "all rounders" I dont think Symonds/Harvey compare with Kallis or Marshall/Holding compare with the fab 4 all rounders. But still they dominated the game. It is well known that all rounders make the team stronger but to what extent compared to specialist top class batsmen and bowlers? Would love to see your analysis on that. [[ Anand I am beginning to see two excellent sets of variables for the all-rounder analysis. We will get different perspectives. Ananth: ]]

  • Raj on September 20, 2010, 10:00 GMT

    Hi Ananth,

    Thanks for the great work. May I suggest that a "SUMMARY" graph of ODI bowling performances based on - minimum 100 matches, - SR of < 35.0, - RPO of < 5.5, a - average of < 32, and - mimimum 150 wkts will give you revealing & eye popping but OVERALL PERFORMANCE. That will tell you who the real GREAT PERFORMERS are in ODI bowling. [[ Raj Once you take the SR and Rpo, the average is superfluous since that is the product of SR x Rpo / 6. And the wkts have been already incorporated in selection. So whatever you need is already incorporated. Ananth: ]]

  • Migara on October 5, 2010, 16:37 GMT

    Nice work! But I'd think you should have standardised Averages, SRs and ERs for the comparison. For an example during 80s the ER of most bowlers were 4.4 (Avg total 220) compared to that of 00s where ER os close to 5.5 (Avg total 275). You can compare each parameter of a player to his peers during the period he played. That will allow the distortions taken by the trends in the games to neutralise. It's not very difficult exercise as well because it involves a simple function in a spread sheet. [[ I have already done very extensive work on Peer comparisons. Pl see last year's article. Ananth: ]]

  • Amit on September 24, 2010, 20:50 GMT

    Hi Ananth, That last 'Anonymous' was me. Didn't realize it would let me post on my laptop when my name is not entered (different from how it works when I post from my blackberry). If you have control over it, please feel free to change the last Anonymous (addressed to Alex) to my name. And, keep the wonderful work going. Thanks. -Amit.

  • Anonymous on September 24, 2010, 17:27 GMT

    Alex, thanks for the correction and nice observation on the yorkers. The reverse is also true - if you try bowling yorkers and dont get it right, all your overpitch balls and full-tosses with get you miserable economy rate. Case in point - Agarkar.

  • Alex on September 24, 2010, 9:30 GMT

    Amit - Nice observation. A minor correction: Garner began as the strike bowler (position 1 or 2) at age 31, starting WC '83. He remained so till his retirement in '87 barring a couple of ODIs.

    The era of treating the first 15 overs as the slog overs really started with Greatbatch of NZ in the '92 WC although a few openers such as Srikanth had always treated this phase (or any) as slog. So, for the early strike bowlers, outstanding SR says more than outstanding RPO ... to me Garner's RPO=3.1 is more like RPO=3.5 in 90's and his SR=36 is more like SR=33 in 90's.

    Across all eras, I think the bowlers who excel in the death overs possess great yorkers: Garner, Thommo, Malinga, Holding, Waqar.

  • Amit on September 24, 2010, 4:09 GMT

    Hi Ananth, I wonder if you have the info about when these bowlers bowled in the course of the innings. Most good fast bowlers have the luxury of bowling the first few over (when economy rate was typically lower, alleast until Gilchrist, Hayden and Sehwag came along), but is offsetted having to bowl at death overs. If my menory serves me right, Garner bowled at first change (except WC83) and then 'always' at the death overs. I remember him him bowling maidens in final overs to win tight games. It would be good to know which of these bowlers bowled with a disadvantage and still excelled. -Amit. [[ Amit This sort of information is available only during the past 10 years or so. It is also not available in the public domain. Ananth: ]]

  • Alex on September 22, 2010, 13:56 GMT

    Ananth - re all-rounders, you seem to apply the traditional definition. I like to think of great wkt keepers, great captains, and great fielders as all-rounders as well. [[ Alex For this type of two-dimensional graph based analysis it is better to stick to the traditional definition. For a different type of analysis such as multi-disciplinary contribution to team cause I would look at the wider definition. Ananth ]]

    A quote of I Chappell who loves Mark Waugh: In a late 90's ODI distinguished by Steve Waugh's brilliance, Tony Greig commented on Steve Waugh "this guy has to be the best all-rounder in the world" to which Ian Chappell replied "Tony, he is not even the best all-rounder in his own family!"

  • craigmnz on September 22, 2010, 7:25 GMT

    Anath

    I have to admit that I enjoyed this article (and its batting counterpart) simply because of the way you're using different (and to me, new) techniques. I might have been a first-class scorer for 13 seasons (3 with Wellington, 10 with CD - when I moved to the Wairarapa) but that doesn't make me a stats geek.

    I'm interested in a similar analysis of all-rounders, mainly because I have this nagging doubt that Richard Hadlee doesn't quite belong in the same category as Sobers, Botham or Imran. Undoubtedly one of the great bowlers (seeing his 7 for 23 remains one of my most cherised memories - I wagged school to see it) but I doubt that the NZ selectors ever seriously considered playing him as a specialist batsman in a big match.

    The obvious axes for all-rounders would be batting and bowling average but what about comparative strike rates? The explosive all-rounder as opposed to Mr Reliable? Just a thought... [[ Craig I myself have pushed Test all-rounders ahead of the Test batsmen because of the intricacies it offers. After another very interesting post late this week I will do the All-rounders' article next week. Totally dispassionately speaking, which all-rounders could have been 100% certain of playing in either capacity. Kallis: probably not as a bowler. Kapil: as a batsman, only because of the weakened Indian batting. This argument is applicable to Hadlee as well. If we take the next step, both could not have played for Australia as batsmen. Botham: almost the same as Kapil and Hadlee. And what about Vettori. His batting average is better than Greatbatch, McClean, AD Nourse, Mahanama, Srikkanth, Grant Flower et al. But he probably would not make to many teams as no.6. That leaves us with two players who probably could have played in any team in either capacity. Sobers (poor bowling average but an excellent double-pronged skill) and Imran. And I would venture to add Miller.. However let us wait for the analysis. Ananth: ]]

  • Ravi on September 20, 2010, 18:18 GMT

    Another good article Ananth. Here are my 2p for observations from the charts: 1. Batsmen in the first ODI decade were more orthodox (not necessarily slow scoring)in their approach- build, consolidate-slog for 20 overs each. Also true for 50-over games in the first decade. Possibly therefore seamers had lower RPO than modern ones. Old gen spinners went for more RPO than seamers. GGilmour WC75 sf, Garner WC 79final, W Davies7-51 as examples among many such excellent ones. [[ Ravi Price, Murali, Maninder and Harper are the only bowlers who have bowled 500+ overs and have an Rpo value below 4.0. Only Maninder belongs to the previous generation. Ananth: ]] 2. Fully fit bowlers or those playing in 1 or 2 formats (not all 3) have done well on the SR. To be fair T20 isn't applicable to any except the very recent bowlers. 3. One can't failto notice the WI domination in the firt decade- Holding & Garner's excellent economy rate but slightly lower SR than the very best is a reflection of the slightly more relaxed nature of batting. [[ Also the reduction of risks taken. Ananth: ]]

    4.Wonder if we see any evidence of the effect of the no. of minnow teams? [[ We have minnows such as East Africa even in the 1970s. Ananth: ]]

    5. Just a question: Were there ever 8-ball overs in ODIs? [[ Yes. About 10 or so matches at the beginning, including the first one. However that does not matter since I store the balls information and all my Rpo values are for 6-balls. Ananth: ]]

  • Anand on September 20, 2010, 15:16 GMT

    Ananth:

    Sriram's suggestion on having dividing lines non-parallel to the axes may be an excellent idea for the BCG graphs on all rounders. I am sure with your knowledge and understanding you can some up with some very interesting analysis of all rounders. One thing that has made me wonder is that neither the West Indies team of the 80's nor the Aussies of the 2000's had the top all rounder of their times. I mean in the 2000a Australia had Symonds, Harvey, Watson and in 80s Marshall and Holding were no push overs with the bat but as "all rounders" I dont think Symonds/Harvey compare with Kallis or Marshall/Holding compare with the fab 4 all rounders. But still they dominated the game. It is well known that all rounders make the team stronger but to what extent compared to specialist top class batsmen and bowlers? Would love to see your analysis on that. [[ Anand I am beginning to see two excellent sets of variables for the all-rounder analysis. We will get different perspectives. Ananth: ]]

  • Raj on September 20, 2010, 10:00 GMT

    Hi Ananth,

    Thanks for the great work. May I suggest that a "SUMMARY" graph of ODI bowling performances based on - minimum 100 matches, - SR of < 35.0, - RPO of < 5.5, a - average of < 32, and - mimimum 150 wkts will give you revealing & eye popping but OVERALL PERFORMANCE. That will tell you who the real GREAT PERFORMERS are in ODI bowling. [[ Raj Once you take the SR and Rpo, the average is superfluous since that is the product of SR x Rpo / 6. And the wkts have been already incorporated in selection. So whatever you need is already incorporated. Ananth: ]]

  • Sriram on September 20, 2010, 5:08 GMT

    Hi Ananth. I am a regular reader of your articles and I am big fan of your work. This is another very nice article, but like a lot of people have mentioned earlier, it seems to be difficult to compare players of different eras, unless you incorporate peer ratios.

    I have an alternate suggestion though I am not sure whether it is acceptable. Should the lines dividing the graph into four quadrants necessarily be horizontal or vertical? For example, if the line parallel to the y-axis is replaced by a line with a negative slope passing through the median it will make a difference. Of course, it becomes subjective in the sense that the analysis would be more rewarding to bowlers with low economy rates than the more attacking ones. (or vice versa if you replaced that line with another having a positive slope) [[ Sriram Your suggestion is an out-of-the-box one and has intrinsic merits. However as you yourself have observed it would be difficult to determine the slope. Readers might even start commenting that I have done a subjective determination of the slope to include or exclude certain players. I think we have to be contented that while the dividing lines are asymmetric, at least approximately equal number of players would be placed on either side. Ananth: ]]

  • Murali on September 20, 2010, 3:04 GMT

    I am just 3 weeks old into my MBA, just been introduced to the BCG and McKinsey models. To see you use it to compare bowlers and batsmen is amazing.

    I get the practicality of it all reading this than looking into some generalized example on a article. Good Work.

    Murali. [[ Murali You have made my day. The fact that you can understand the value of BCG charts by relating to Cricket related numbers will certainly help you in your understanding the same in complex business models. Nice to see someone look for an unconventional benefit. Ananth: ]]

  • SJ on September 20, 2010, 2:35 GMT

    Dear Anand, Congratulations on this fantatsic piece of work. It is very interesting to go through this analysis and am sure it is a delight for all those self proclaimed cricketasticians like myself. As an avid reader of your features, i would be very keen to see you coming up with the Best 15 ODI team and Test team (once your analysis on the test batsman and bowlers are done), based on your data analysis.

    Thanks once again......

  • A Friend Indeed on September 19, 2010, 18:27 GMT

    like the idea of using BCG charts but with caveats, otherwise it is an exercise in futility. I mean who did not know that Donald and McGrath are great bowlers. But are they really better than Wasim or Garner? Answer will not come from this kind of charting. Then what purpose this analysis serve except Marking Rana Navedul Hasan as attacker in chief (which is laughable) perhaps idea is to make information easy to see , you will say, and that's where it really breaks down. There are so many charts with different criteria that it is confusing to even look at them. One thinks visualization should make it easy to comprehend info presented. Plus, since all the top performers seem to be from the same era- seem to suggest that there are some basic flaw in the design and approach. May be a decade wise will break-up will provide some clarity. Don't also forget that ODIs were earlier a 60 over game.

    I sincerely hope you guys don't mind my frank opinion and take it as constructive criticism. [[ A friend indeed (or Chi-square until 2 days back ???) Kindly tell me which analysis shows something entirely new. All analysis, including the pure statistical analysis which you love, use existing facts, derive additional insights and present in a new form. Nothing more. Even "Dilscoop" is not original. I saw Marillier do that to defeat India during 2001. However you will go to any length to find imaginary problems. Of course, any one and their neighbour's dog know that McGrath and Donald were great ODI players. But this graph presents their relative positioning and in which of the two key areas were they good. McGrath's greatness is influenced by his lower Rpo figure and Donald's by his better Strt figure. This sort of insight is what I am trying to bring to light. An unbiased reader would appreciate this. But you do not want to. Also I only said that they were in the top right quadrant with good figures in both key areas. I did not compare any of the bowlers. Re Rana Naved, it is your comment. It is laughable that you create your own comment and ascribe it to me to make the analysis look poor. Did I say one word about Rana in my article. He has a very good strike rate but an awful Rpo figure which places him so far down in the graph that he would indeed be a liability. He is a better quality Agarkar. Your desperation to get back to me has made you use the words "attacker-in-chief" for Rana Naved when 8 bowlers have a better strike rate. And finally do not forget that he has captured 110 wickets at 31.5 bpw. I request you to contribute positively or stay away and look for other wonderful articles which meet your requirements. You are not adding anything positive to this blog. I love constructive criticism but dogmatic criticism for the sake of criticising, "thanks, but no thanks". Ananth: ]]

  • absha1 on September 19, 2010, 17:11 GMT

    There was a comment earlier in this thread regarding all rounders and their poorer showings. Please note that while batting all rounders (Kallis, Sobers) can continue for a longer time without majorly affecting their strong suit (it is easier to bat longer than continue a fast bowling career), fast bowling all rounders (particularly Imran, Botham, Dev) can decline spectacularly late on as they rely on their batting.

    Hadlee retired in time. Imran, however, played at least three years beyond his shelf life as a bowling spearhead (I mean, he retired at 39 - Pollock retired at 34 or 35). Same thing goes for the other all rounders - you can't fairly compare them against other bowlers over a career as, well, they often play as batsmen. A better analysis should look at sustained peaks over 10 years.

    Apart from that, this is a fine analysis. I think it complements an earlier analysis on the fifteen great bowlers, although that one too downgraded all rounders who played matches as batsmen

  • khizer on September 19, 2010, 15:21 GMT

    This is obviously a new type of analysis. I just called it similar to Karnaph maps in the language of mathematics.!! is placed at [[ Khizer Karnaugh maps are for simplification of Boolean expressions. The result of an expression is represented in the junction of the concerned variables. Here the player is represented at the junction point of the two variable values. However the importance here is the relative position of this point in the quadrants. It is the quadrant concept which sets this graph apart. Ananth: ]]

  • Anand on September 19, 2010, 13:28 GMT

    Ananth: Thanks for your response(s). I love the readers' comments too and most of the time they are as interesting as your analysis itself. What I am against is, people just criticising without suggesting any improvement or changes. Even in our respective jobs we dont want our supervisors or clients just telling us that what we did is unsatisfactory. We expect them to back it up with suggestions for improvement. There are many readers who come up with amazing suggestions, which you have also implemented in some cases. Continuing on the thoughts about great players being "first among equals", the way I see it is, good players can be found by mere stats. But a deeper analysis of stats tell us who are the players who took the game to a different level. These would be the great payers and legends. The batsmen and bowlers in the top right quadrant of your BCG charts are filled with players who have taken the game to a different level.

  • Alex on September 19, 2010, 12:52 GMT

    Ananth - apologies for a detour on a minor matter but Roberts was not a "potential all-time great" as you state. That label is best reserved for G Pollock, B Richards, Bond, and Bishop etc. who got hit by ban/injuries. Roberts loses out on wkt tally because WI played less than 80 ODI's in his 10 years, all through which he actually led such greats as Garner & Holding. He was arguably a Top 5 ODI bowler in his era. I am sure you have seen and appreciate him. [[ Alex You are right. His case is similar to that of Zaheer Abbas. His average of 47 is not just because of not outs but excellent Balls and StRt. You might remember the article I did on analysis by years highlighting the matches played, Ananth: ]]

  • Deepak Chandran on September 19, 2010, 11:27 GMT

    Hi Ananth

    Wonderful analysis on both the batsmen and bowlers. I just wanted to check something with you. Is it possible to include a match impact variable in the analysis- essentially trying to measure if the performance was in a winning or a losing cause.What I have in mind is to calculate in mind the averages(Runs Per Over,Strike rates etc) by using win loss ratios as a weight.Just to illustrate if a player has played 100 matches out of which 60 matches the team won and 40 matches the team lost- then the average should be average in winning matches *.60 + average in losing matches 0.40. Please do let me know your thoughts on the same

    Regards Deepak [[ Deepak I am not very comfortable since this will benefit the players in stronger teams. Ananth: ]]

  • Jitendra on September 19, 2010, 11:21 GMT

    It is how many wkts taken in a match.ie 255/150=1.7 wkt per match

  • Alex on September 19, 2010, 11:03 GMT

    Ananth - yet again the discussion has taken the turn "Is A greater than B" type, e.g., "Garner could not have been as economical". I think such raw number based charts are best used to identify trends only. It is a bit surprising spinners are poor on RPO _and_ SR!

    I do feel that Ambrose's RPO & Lee's SR are the most remarkable due the reasons cited earlier. - Even Ponting & Warne have said that Ambrose's frugal RPO was possibly his weakness: he didn't mind you bat so long as you didn't score off him and turned truly lethal only when he got hit. [[ A very interesting observation. Any modern bowler who fits the bill. It may be difficult today because of T20 matches. Sacrilege it may be. By any chance do Murali or Kumble fit the bill. Ananth: ]]

  • Krishnadeep on September 19, 2010, 6:43 GMT

    Though I am not a statistician and this is purely my thoughts penned down...can we not introduce a relative fraction for RPO, where the fraction is the runs per over conceded by the bowler(average) to average runs per over of all teams (batting) during those eras in ODIs. If for example Garner had an RPO of 3.11 during his playing days from A to B years and if during the same time, average RPO of all teams is 4 then fraction would be 3.11/4..Similarly for strike rate such a figure can be introduced [[ Krishna Wjat you say is the Peer comparisons which was dealt with exclusively last year. It is not even an era-based. I have done it individually for each player, in other words in the sub-era in which he played, from first to last match. I will probably do a Peer adjustment for teh Test analysis. Ananth: ]]

  • Anand on September 19, 2010, 5:42 GMT

    A general comment to the readers particularly those who think stats are lies and more lies. The title of this blog is "It figures" and the philosophy is statistical analysis and representations. So if stats are not what appeal to you, then this section may not be for you. Also, cricket is a subjective game and numbers CAN NEVER represent the exact picture. To my knowledge, Ananth has never made any qualitative conclusion based on his quantitative analysis. In this blog too, although he only mentions Mc Grath leads a set of blowers etc., I DONT SEE ANY PLAE WHERE HE HAS MENTIONED THE Mc GRATH IS THE BEST ODI BOWLER or any such thing. We are entitled to our opinions on players and are not expected to change those based on the numbers seen here. But the numbers give us some interesting details. Its our choice whether we take interest in them or ignore them. No point in making blanket statements about the futility of the analysis, etc. For non cricket lovers, cricket itself is pointless!! [[ Anand I myself have been guilty of this mistake. I have labelled the word "best" to individuals. Recently I have used the term "best" to a collection of players. Anyhow it is true that most great players are "first amongst equals" amongst an elite group. Only one person transcends this axiom. My editor is our son who is doing his Ph.D in US. I have told him to strictly monitor such transgressions. I would also like readers to do their own interpretations and come out with their observations. This is a sharing forum and I will quit it the day the readers do not come in. I write for another blog where the responders can be counted in one hand. I do that to sustain this hyper-involved time consuming activity. Ananth: ]]

  • Anand on September 19, 2010, 5:29 GMT

    (Contd.) Could this also be a reason that Kallis, Akram are not considered in the same league as the 4 fabulous all rounders of the 80's? I am not under-rating Kallis or Wasim as all rounders. They are phenomenal. But their one department is much stronger than the other. However, for Imran, Kapil, Botham and Hadlee, their contributions with the bat and ball were almost of the same scale (may not be so true for Hadlee but surely for the other three) which made them special all rounders. Today one can expect Kallis to win more games with the bat than with the ball and Wasim to win more with the ball than with the bat. But with Kapil, Botham and Imran, you never know what magic they have in store with the bat as well as with the ball. That, to me is the true sense of the word "all rounder". Again, this is my opinion and differences are most welcome. Ananth, do you think my explanation makes sense? [[ Anand In general I agree with you. We go back to the age-old definition of a great all-rounder. He should be able to walk into the team in either capacity or be capable of winning matches on a regular basis in either capacity. Only the true great all-rounders like Sobers, H, I, K, B make it to this list. Again I do not want to say I have made a blanket statement. In fact this can be resolved through the BCG chart sfor all-rounders with the two averages as the two axes. Ananth: ]]

  • Anand on September 19, 2010, 5:15 GMT

    Ananth: Thanks for this chart. In your chart on batsmen I had commented about missing the names of the 4 big all rounders (Imran, Botham, Kapil and Hadlee) and you had provided the right reason that they were all rounders and did not meet the qualifying criteria. Its also interesting to see that even as pure bowlers they really are not among the top. Which actually makes their contribution as all rounders so special. I mean if you say someone has a batting average of 30-35 it looks like "so what?". Similarly a bowling economy less than 4.2 or a strike rate close to 40 or average like 30 again look like ordinary stats. However a combined stat like someone who can score 30-35 runs all the time bowl about 8 overs concede like 30 runs and pick at least a wicket or two every game... Now that sounds really like "wow!! I want this guy in my team." Of course the exact numbers may be different but I hope you are getting the idea.. (Contd. in the next message)

  • saurabh on September 19, 2010, 2:48 GMT

    Hey Ananth,

    Another great article, Infact I look forward everyweek for one of your statistical wonders :)

    Keep it up

  • Youvi on September 19, 2010, 0:21 GMT

    Anantha, Looks neat/less cluttered with the number key/corresponding bowlers. I hovered the cursor on the number and expected the name to come up ! Nonetheless easy on the eyes and can easily connect the number key on the graph with player. It is always felt that cricket is a batsman's game so it is really fascinating to look at bowlers'performances this way. More often than not the centuries/runs etc seem to linger but the bowlers are not lionized to that degree, imo. This is very well done. [[ Youvi Hovering and display of the bowler names is a great idea. But that requires a lot more of work on Cricinfo's part. I am not sure whether it can be done in a blog-html environment. Ananth: ]]

  • Rohit Singh Loomba on September 18, 2010, 23:32 GMT

    Anantha, why take criticism so personally. Chi and Matt seem to have a problem with the methods employed, not with you or at least that how it seems to this neutral observer. I like and share your enthusiasm and tenacity but when you are in a public forum being hot tempered hardly helps. My perspective is that these stats methods were developed for a reason- to separate the wheat from the chaff, to reduce noise in the data and to reduce various errors that creep in the analysis that at worse may distort the reality. For example, ODI’s have seen many profound changes over last decade or so. Even mishits are going for sixes, power plays turn 'caught in the deep' to free runs..free hits add on, shorter boundaries compound the problem. So their point is valid and you could respond by acknowledging that.. [[ Rohit I see your point and over two and a half years I can assure you that the number of times I have responded in strong words is probably in single figures. The problem is that this accusation seems to be continuing on despite my clarifying many times. In this particular analysis I have made it clear that it uses raw data. Then why label the exercise as pointless repeatedly. In fact you shoud consider my response to Matt as quite normal. Do I sound hot tempered there. It is only in responding to Chisquared, especially as he seems to do this very often: also in comments on other articles, that I might have come on strongly. What you say makes sense in a different type of analysis. However in this type of graph it is better to put in raw data. Adjusting the data will have its own problems. The adjustment methodology itself would be quite subjective. The idea is to draw meaningful conclusions. Compare Lee and Garner. Admire about the position of Akhtar and Younis. Marvel at the presence of Agarkar in this confedracy of greats. Bring new insights a la Alex on Ambrose and Bishop. Wonder what Bond would have ended with if he had an injry-free career. Elaborate my comment why the bowlers are dominant in the middle quadrants in the Average based analysis. Observe how under-rated Saqlain has been. And so on. Add value to the analysis instead of coming out with a negative comment such as "pointless" just because it does not bring in statistical methodology. Unfortunately these comments set a negative trend unlike the Batting article in which there were some useful comments to start with. Finally let me say that I never pull down any other method/article. In fact I go out of the way and appreciate what has been done. Have you ever seen one other columnist come out and make a positive comment on, say, one of my articles. I have done that quite often, including the last Gabriel article. Ananth: ]]

  • Faran Ghumman on September 18, 2010, 23:30 GMT

    This article shows what a great bowler Saqlain was, the most attacking spinner ever, his stats i.e strike rate prove that, shame that he was kicked out too soon by pak team and also has never been given the due by world observers that he deserves. Ananth, i hope you show these people how good Saqi realy was...

  • Abhirup on September 18, 2010, 23:10 GMT

    I believe what Jitendra meant was the ratio: Total no. of wickets taken/Total no. of ODIs played

    So some one with 150 wickets from 100 ODIS will have a wicket per ODI of 1.5 [[ Jitendra/Abhirup My miss and apologies. A good idea to do Average vs Wpm figure. Although the Strike rate is a better measure since it takes into account matches in which fewer overs were bowled. Ananth: ]]

  • Murali k on September 18, 2010, 20:09 GMT

    Use of the median to draw the quadrants is fantastic. It brings in more value to the graph/evaluation. I would be nice if the axes are normalized and plotted around the median, the graph would imply a better relative values of the players. [[ Murali What you are saying is that the quadrants will be of equal sizes but represent different sets of values.Since I have written my own program it can be done. But I have to see how it will look like. Let me see for the Test analysis. Ananth: ]]

  • jitendra on September 18, 2010, 17:23 GMT

    bond is the best bowler. see strike rate chart. in this a decent strike rate and good economy rate also. i also want a chart with wicket per odi. in all measures bond would pretty fit and lies in good performer category each time. [[ Jitendra Thge term "wicket per odi" eludes my uderstanding. Maybe you could be clearer. Ananth: ]]

  • dinesh on September 18, 2010, 15:52 GMT

    Hi Ananth,

    You surely sir.this analysis really shows who r d top performers n who wer not so. But i got a doubt regarding your next analysis for Test Batsmen because f d huge time frmae f tests(1877-2010) wer d conditions r really different n i request you if it is possible to have graphs for different eras like 1877-1920,1920-1970,1970-2000,2000-2010 as d condidtin n playin style varied a lot in des eras as in d 1st era pll wer startin to paly cricket so it was really new to d world n den d bradman era wen i tink world war did affect cricket n den richards n co era wen with d arrival f ODI's palyin style did change a lot n in d last decade wen T20's arrived playin style was altoghter different n pitches are more Flatter des days except in may b ENG n SA [[ Dinesh Yes I agree. Tests will be a real challenge. I have to make some adjustment before comparing batsmen/bowlers over a 132-year period. Ananth: ]]

  • mashood on September 18, 2010, 15:11 GMT

    New respect for Shoaib Akhtar. Never thought he was good enough to have his name in the same sentence as Waqar Younis but he has apparently been just as good an ODI bowler. All Hail Shoaib. [[ Mashood I myself was quite shocked to see the two names which were so closely superimposed on each other on most graphs. Shoaib has certainly been an under-rated bowler. Yesterday's delivery which dismissed Trott showed that the fire is still there.Reminded one of those famous couple of deliveries at Kolkatta which dismissed the great Indian pair. Ananth: ]]

  • Neeraj Raina on September 18, 2010, 14:59 GMT

    Hi Ananth,

    Once again a brilliant article. I think yours is one of the most read blog among which are posted in cricinfo based on the comments you receive. It really amazes me how you how you find time and ideas to write such articles and then even reply to the comments.

    I also liked the names being displayed in graph itself even though it was bit crammed. I know you received lot of suggestions to improve on it but I still feel that was better than this.

  • Jay on September 18, 2010, 14:37 GMT

    Amazing how Kapil Dev is the only Indian bowler ever who is not dangerously close to the "Also ran" quadrant.

  • Alex on September 18, 2010, 13:21 GMT

    Ananth: I think Lee was a true great in ODI's. It would be nice if you add Roberts: I still remember the first 4 overs bowled by him & Garner in '83 WC final. A few observations:

    1. ODI's evolved such that, on average, batsman score at a faster rate but also get out quicker. Makes sense since the earlier ODI's lasted 55/60 overs. Note how well Garner, Holding & Hadlee score over modern greats on RPO but lose out on SR.

    2. Saqlain was one of the two best spinners in ODI's.

    3. Only a next generation bowler (Younis) outdid Lillee on career SR.

    4. Ambrose's RPO is more impressive than Garner's. Ambrose's is 8% better than that of the next best frugal contemporaries (Walsh & Larsen) while Garner's is 6% better than his (Hadlee & Holding).

    Bishop is a classic "what if" case. Over his first 3 yrs, he had SR < 28. He was perhaps half a notch below Ambrose, but still! A healthy Bishop could have kept WI world champions into the 21st century with the support of Ambrose & Walsh. [[ Alex Difficult to add one who is so far behind the cut-off. Although an average of 20.3 makes him potentially an all-time great. Ananth: ]]

  • Abhi on September 18, 2010, 13:21 GMT

    Alex The issue is not whether Matt, chisquare etc have a point. The issue is how they get their point across.

  • Shehan Dias on September 18, 2010, 10:54 GMT

    Hi Ananth,

    Im a uni student doing business and I certainly must say I find the BCG graph quite attractive to look at! How did you come up with this idea? Brilliant I say!!

    There will be limitations in this kind of analysis (obviously, as studies cannot take into account of every variable like pitches, fielding positions, powerplays, type of batsmen faced)but I was pleasantly surprised to see the likes of Garner in one of the 'better' squares. Interesting to note the positions of guys like Zaheer and Sanath. Knew Dilhara Fernando had to make the list somehow.

    I would like to ask, are these graphs the finalised graphs or have you just restricted yourself to 30 bowlers due to space limitations?

    Keep up the good work and hope to see a BCG chart on Test matches if possible. Cheers

    (Ps-pretty sure that nearly 99% would be able to figure out this analysis)

  • jp on September 18, 2010, 10:34 GMT

    these are only figures and doesnot take match presepective and other situations into consideration [[ Yes. Ananth: ]]

  • Simrat on September 18, 2010, 10:10 GMT

    Maybe, I am in the minority, but I liked it better when the names were displayed on the graph (not by a key). Still, the overall analysis is very good. [[ Simrat Funny thing is that I also like the names on the graph. However it must be agreed that the fixed space taken on the graph with the reduced amount of printing has made the graphs less cluttered. Ananth: ]]

  • kianig on September 18, 2010, 7:37 GMT

    this article is ridiculous and waste of time............ [[ Thanks for the great insight. My only suggestion is that whatever extent of time you have, please do not waste it here. Ananth: ]]

  • Alex on September 18, 2010, 7:30 GMT

    Ananth:

    1. If you bring down the cut-off for #wkts to 80, your analysis can include Andy Roberts whose ODI figures may be said to represent his true worth.

    - I like the metric "SR x RPO" ... the lower the better. Roberts finishes 2nd on it (at 121) behind Garner (at 114). [[ SR x RPO is nothing but 6 times the Bowling Average. Ananth: ]]

    2. To satisfy Matt & Chi Square, who do have a point, perhaps you can replace RPO with "relative RPO" (e.g. RPO of 4.0 in a match that produced 600 runs in 100 overs has a "relative RPO" of 4 x (400/600) = 2.7). Likewise for SR. [[ Alex These are raw figures. Any amount of "adjusting" these will result in complaints in the other direction. Also no one is sure what is right. And what is "relative RPO". These are career leval figures. How does one get match figures in. The only adjustment which will make sense is Abhi's one on somehow incorporating Peer ratios. How, I am still unclear. Somehow, for tests, I must fiond a way to project the numbers to two vertical planes, duly adjusted. Ananth: ]]

    - I don't think Garner would have RPO=3.1 and SR=36 in 90's-00's but, considering what Ambrose managed, I think he probably would have managed RPO=3.3 and SR=36 in Ambrose's/McGrath's era.

    - It is noteworthy that Ambrose's figures eroded greatly after he turned 34 (i.e., over his last 4 years) when, thanks to Bishop's injuries & lack of upcoming talent, WI "attack" was limited to yet another father figure in Walsh. [[ Even if Garner went to 3.5, it is superlative. What about Lee's sub-30 strike rate. Is it only because of the openers' propensity to attack. Does it indicate genuine greatness. The problam is that there are many insights to be drawn from this type of presentation of measures across ages and readers are not looking for those. Unfortunately people seem to be nit-picking. Surprisingly this was not the case with batting graphs. Ananth: ]]

  • Abhi on September 18, 2010, 4:32 GMT

    Re. the reply to chi square etc...Well said. I've noticed this supercilious tone by some commentators on quite a few instances now. We have these PHD type stat nerds who tend to look down condescendingly on something they consider not particularly rigid. If we insist on some so-called rigid statistical models at all times with zero subjectivity (they wish!) we will end up with a dozen or so stat nerds corresponding with each other- that’s all. The regular joe cricket fan will be long gone. In any case, we all know about stats. On the previous blog, by Rogers, I have commented too that the WHOLE thing is basically based on an entirely subjective phenomenon called "form"! Sure, then we may apply rigid statistical models, but the entire framework is essentially shaky.

    I gave the eg. of injuries there and how they bring the whole house of cards down- since the stats then completely breakdown in "predictive" power.

    You simply cannot take a limp racehorse, incorporate his timings into "overall" analysis, or “LCG”s...and then make some sort of predictions of how things would turn out if the racehorse were fit sometime in the future. [[ Abhi Thanks for the nice words. I have said time and again that my objective is that my analysis must be fully understood by 95% of the reading public. Even when I do something a little more statistically-oriented, I make sure that I do not cross the line of common understanding. I myself feel that Gabriel's article is one of the best. However barring an overall understanding of the graphs the normal follower cannot understand the intricacies. There is certainly a place for well-written and well-presented statistical articles like Gabriel's. My only request to the readers is not to pull down other less-academic articles. Even now I am holding back the Test Bcg graphs for some time since I must make sure that the merging of peer ratios is seamlessly done and everyone must follow what I have done. Ananth: ]]

  • Abhi on September 18, 2010, 4:22 GMT

    Ananth, For the Test BCGs I feel “peer ratios” must be somehow incorporated as they are by far the best comparative indicator and in a roundabout way include conditions ,pitches etc since these would be similar to a group of peers. The Average career weighted bowling quality by itself does not factor in pitch quality and other conditions. So, perhaps a combination of peer averages and bowling quality may be better. [[ Abhi Thanks for at least making a positive suggestion. I thought of the same earlier. However could not find a way to fit in the third dimension. In fact your suggestion will make a lot of sense in Test matches since the number of years is quite high. Let me see. Ananth: ]]

  • Chi Square on September 18, 2010, 4:11 GMT

    I agree with Mat. Quite pointless since variables are not taken into account. While it is OK to do comparative analysis for entertainment value, it is quite essential to at least try to control for variables that may account for the differences! The analysis posted on career consistency earlier on the similar lines was much better as it had some genuine statistical principles embedded in the analysis. [[ This is a statement for the sake of making one. It contributes nothing. From the beginning it has been mentioned that the graphs are the same as for Batting, these are all career figures. I suggest try and read sometihing from the unadjusted career figures across ages instead of making a negative stamement because this does not conform to the so called "statistical analysis" expectation. Else not a bad idea to stay away. Ananth: ]]

  • Matt on September 18, 2010, 1:11 GMT

    I'm sorry, I think this is pointless. We all say you can't really compare players from different eras in Tests (pitches/balls/bats different), so how much harder is it going to be in ODIs when you look at the effect of the 30-yard circle? Great miser as he was, I don't think Garner would ever have gone for 3.11 (from memory) if he was bowling with only 2/3/5 men on the boundary ... perhaps if you can work out overall economy before 1980/1, when the 30-yard circle came in, and after, and apply the difference (say overall economy went up from 4 an over to 5 - an increase of 25%) to bowlers' figures from that time period? Then at least you're flattening out one hump in the far-from-level playing field. [[ Matt If what you say is to be followed then you ahould NEVER, NEVER compare players across ages. How can you compare the runs scored, the average or any other number. Why should one do a table of any of these measures at all. The point is that this type of analysis will make people sit up and take notice how good a bowler Garner must have been. How great Richards' average was and do on. These are comparisons across ages. Why would you not gather something from the charts. Like Lee's strike rate, like the Siamese twin nature of two great Pakistani bowlers and so on. Ananth: ]]

  • Ali Shah on September 17, 2010, 23:28 GMT

    You do some wonderful analyses ananth. Keep up the good work.

  • No featured comments at the moment.

  • Ali Shah on September 17, 2010, 23:28 GMT

    You do some wonderful analyses ananth. Keep up the good work.

  • Matt on September 18, 2010, 1:11 GMT

    I'm sorry, I think this is pointless. We all say you can't really compare players from different eras in Tests (pitches/balls/bats different), so how much harder is it going to be in ODIs when you look at the effect of the 30-yard circle? Great miser as he was, I don't think Garner would ever have gone for 3.11 (from memory) if he was bowling with only 2/3/5 men on the boundary ... perhaps if you can work out overall economy before 1980/1, when the 30-yard circle came in, and after, and apply the difference (say overall economy went up from 4 an over to 5 - an increase of 25%) to bowlers' figures from that time period? Then at least you're flattening out one hump in the far-from-level playing field. [[ Matt If what you say is to be followed then you ahould NEVER, NEVER compare players across ages. How can you compare the runs scored, the average or any other number. Why should one do a table of any of these measures at all. The point is that this type of analysis will make people sit up and take notice how good a bowler Garner must have been. How great Richards' average was and do on. These are comparisons across ages. Why would you not gather something from the charts. Like Lee's strike rate, like the Siamese twin nature of two great Pakistani bowlers and so on. Ananth: ]]

  • Chi Square on September 18, 2010, 4:11 GMT

    I agree with Mat. Quite pointless since variables are not taken into account. While it is OK to do comparative analysis for entertainment value, it is quite essential to at least try to control for variables that may account for the differences! The analysis posted on career consistency earlier on the similar lines was much better as it had some genuine statistical principles embedded in the analysis. [[ This is a statement for the sake of making one. It contributes nothing. From the beginning it has been mentioned that the graphs are the same as for Batting, these are all career figures. I suggest try and read sometihing from the unadjusted career figures across ages instead of making a negative stamement because this does not conform to the so called "statistical analysis" expectation. Else not a bad idea to stay away. Ananth: ]]

  • Abhi on September 18, 2010, 4:22 GMT

    Ananth, For the Test BCGs I feel “peer ratios” must be somehow incorporated as they are by far the best comparative indicator and in a roundabout way include conditions ,pitches etc since these would be similar to a group of peers. The Average career weighted bowling quality by itself does not factor in pitch quality and other conditions. So, perhaps a combination of peer averages and bowling quality may be better. [[ Abhi Thanks for at least making a positive suggestion. I thought of the same earlier. However could not find a way to fit in the third dimension. In fact your suggestion will make a lot of sense in Test matches since the number of years is quite high. Let me see. Ananth: ]]

  • Abhi on September 18, 2010, 4:32 GMT

    Re. the reply to chi square etc...Well said. I've noticed this supercilious tone by some commentators on quite a few instances now. We have these PHD type stat nerds who tend to look down condescendingly on something they consider not particularly rigid. If we insist on some so-called rigid statistical models at all times with zero subjectivity (they wish!) we will end up with a dozen or so stat nerds corresponding with each other- that’s all. The regular joe cricket fan will be long gone. In any case, we all know about stats. On the previous blog, by Rogers, I have commented too that the WHOLE thing is basically based on an entirely subjective phenomenon called "form"! Sure, then we may apply rigid statistical models, but the entire framework is essentially shaky.

    I gave the eg. of injuries there and how they bring the whole house of cards down- since the stats then completely breakdown in "predictive" power.

    You simply cannot take a limp racehorse, incorporate his timings into "overall" analysis, or “LCG”s...and then make some sort of predictions of how things would turn out if the racehorse were fit sometime in the future. [[ Abhi Thanks for the nice words. I have said time and again that my objective is that my analysis must be fully understood by 95% of the reading public. Even when I do something a little more statistically-oriented, I make sure that I do not cross the line of common understanding. I myself feel that Gabriel's article is one of the best. However barring an overall understanding of the graphs the normal follower cannot understand the intricacies. There is certainly a place for well-written and well-presented statistical articles like Gabriel's. My only request to the readers is not to pull down other less-academic articles. Even now I am holding back the Test Bcg graphs for some time since I must make sure that the merging of peer ratios is seamlessly done and everyone must follow what I have done. Ananth: ]]

  • Alex on September 18, 2010, 7:30 GMT

    Ananth:

    1. If you bring down the cut-off for #wkts to 80, your analysis can include Andy Roberts whose ODI figures may be said to represent his true worth.

    - I like the metric "SR x RPO" ... the lower the better. Roberts finishes 2nd on it (at 121) behind Garner (at 114). [[ SR x RPO is nothing but 6 times the Bowling Average. Ananth: ]]

    2. To satisfy Matt & Chi Square, who do have a point, perhaps you can replace RPO with "relative RPO" (e.g. RPO of 4.0 in a match that produced 600 runs in 100 overs has a "relative RPO" of 4 x (400/600) = 2.7). Likewise for SR. [[ Alex These are raw figures. Any amount of "adjusting" these will result in complaints in the other direction. Also no one is sure what is right. And what is "relative RPO". These are career leval figures. How does one get match figures in. The only adjustment which will make sense is Abhi's one on somehow incorporating Peer ratios. How, I am still unclear. Somehow, for tests, I must fiond a way to project the numbers to two vertical planes, duly adjusted. Ananth: ]]

    - I don't think Garner would have RPO=3.1 and SR=36 in 90's-00's but, considering what Ambrose managed, I think he probably would have managed RPO=3.3 and SR=36 in Ambrose's/McGrath's era.

    - It is noteworthy that Ambrose's figures eroded greatly after he turned 34 (i.e., over his last 4 years) when, thanks to Bishop's injuries & lack of upcoming talent, WI "attack" was limited to yet another father figure in Walsh. [[ Even if Garner went to 3.5, it is superlative. What about Lee's sub-30 strike rate. Is it only because of the openers' propensity to attack. Does it indicate genuine greatness. The problam is that there are many insights to be drawn from this type of presentation of measures across ages and readers are not looking for those. Unfortunately people seem to be nit-picking. Surprisingly this was not the case with batting graphs. Ananth: ]]

  • kianig on September 18, 2010, 7:37 GMT

    this article is ridiculous and waste of time............ [[ Thanks for the great insight. My only suggestion is that whatever extent of time you have, please do not waste it here. Ananth: ]]

  • Simrat on September 18, 2010, 10:10 GMT

    Maybe, I am in the minority, but I liked it better when the names were displayed on the graph (not by a key). Still, the overall analysis is very good. [[ Simrat Funny thing is that I also like the names on the graph. However it must be agreed that the fixed space taken on the graph with the reduced amount of printing has made the graphs less cluttered. Ananth: ]]

  • jp on September 18, 2010, 10:34 GMT

    these are only figures and doesnot take match presepective and other situations into consideration [[ Yes. Ananth: ]]

  • Shehan Dias on September 18, 2010, 10:54 GMT

    Hi Ananth,

    Im a uni student doing business and I certainly must say I find the BCG graph quite attractive to look at! How did you come up with this idea? Brilliant I say!!

    There will be limitations in this kind of analysis (obviously, as studies cannot take into account of every variable like pitches, fielding positions, powerplays, type of batsmen faced)but I was pleasantly surprised to see the likes of Garner in one of the 'better' squares. Interesting to note the positions of guys like Zaheer and Sanath. Knew Dilhara Fernando had to make the list somehow.

    I would like to ask, are these graphs the finalised graphs or have you just restricted yourself to 30 bowlers due to space limitations?

    Keep up the good work and hope to see a BCG chart on Test matches if possible. Cheers

    (Ps-pretty sure that nearly 99% would be able to figure out this analysis)