Trivia - bowling March 14, 2008

The bowlers who took the most high-quality wickets

In the wickets column of scorecards there is the bland pronouncement that a bowler has captured x number of wickets
130

About a month back, I had done a post on the most consistent bowlers in Tests, as part of an analysis on bowlers. I had mentioned then that there would be two measures for bowlers - the second one is on the quality of wickets taken by bowlers.

In view of the very high number of comments received, we will close the comments by evening of Friday, 21 March so that a comprehensive follow-up can be posted.

Consider three recent innings summaries:

West Indies 215 all out (Sehwag 3-33, Patel 3-51, Kumble 3-57) These numbers suggest Virender Sehwag was the best of the lot and Anil Kumble the worst. In reality, it was the other way around. Kumble took the wickets of Chris Gayle, Brian Lara and Dwayne Bravo. Munaf Patel took the wickets of Daren Ganga, Ramnaresh Sarwan and Denesh Ramdin, while Sehwag collected the tailenders - Ian Bradshaw, Jerome Taylor and Pedro Collins. Another example:

India 240 all out (Ntini 3-41, M Morkel 3-86) Makhaya Ntini captured the wickets of Wasim Jaffer, Sachin Tendulkar and Sourav Ganguly while Morne Morkel captured the wickets of Mahendra Singh Dhoni, Kumble and Zaheer Khan. For that matter, the spell of Andre Nel, who captured only two wickets - those of Sehwag and Dravid - is better than that of Morkel.

Bangladesh 259 all out (Ntini 4-35, Steyn 4-66) Here both bowlers took the same number of wickets, but Dale Steyn took the top four while Ntini mopped up the tail.

In the wickets column of scorecards there is the bland pronouncement that a bowler has captured x number of wickets. There is no information on whose wickets he captured. This analysis seeks to secure such information.

The computation is simple. Every wicket captured by a bowler in the 1865 Test matches played so far is analysed, and the sum of career batting averages of the batsmen dismissed is calculated. It is then divided by the number of wickets captured by each bowler and a Batting Quality Index (BQI) arrived at. It's a simple but exhaustive calculation, which is impossible manually.

The top ten bowlers in this list - criterion being at least 100 Test wickets - ordered by BQI is startling. (I would appreciate it if readers do not immediately write in saying "Wasim Akram is the greatest", "Who are these clowns", "Boje and Dillon could not find a regular place in their teams" etc.)

Table 1: Ordered by BQI

SNo Bowler            Bow Cty Mat Wkt  Sum of   BQI
BatAvge

1.Boje N LSP Saf 43 100 3453.0 34.53 2.Flintoff A RFM Eng 67 197 6652.0 33.77 3.Connolly A.N RFM Aus 29 102 3444.0 33.76 4.Giles A.F LSP Eng 54 143 4812.0 33.65 5.Dillon M RFM Win 38 131 4366.0 33.33 6.Collinge R.O LFM Nzl 35 116 3825.0 32.97 7.Zaheer Khan LFM Ind 53 170 5599.0 32.94 8.Caddick A.R RFM Eng 62 234 7706.0 32.93 9.Hoggard M.J RFM Eng 66 247 8118.0 32.87 10.Martin C.S RFM Nzl 37 125 4086.0 32.69

The list is headed by virtually unknown bowlers. Why does this happen?

Possibly because they do not bowl at the end, picking up tail-end wickets. The other more established bowlers get the opportunity. These bowlers tend to bowl during the middle of the innings.

The other peculiarity is the presence of the three current England bowlers. Here the possible reason is that England has played Australia and India recently and the average of batting averages for these two teams is quite high.

I would be interested in reading comments from interested readers on possible reasons for this peculiar situation.

136.Steyn D.W         RFM Saf  20 105  2655.0  25.29
137.Barnes S.F        RFM Eng  27 189  4646.0  24.58
138.Blythe C          LSP Eng  19 100  2449.0  24.49
139.Wardle J.H        LSP Eng  28 102  2416.0  23.69
140.Noble M.A         ROB Aus  42 121  2859.0  23.63
141.Turner C.T.B      RFM Aus  17 101  2291.0  22.68
142.Giffen G          ROB Aus  31 103  2229.0  21.64
143.Peel R            LSP Eng  20 102  1960.0  19.22
144.Briggs J          LSP Eng  33 118  2025.0  17.16
145.Lohmann G.A       RFM Eng  18 112  1755.0  15.67

At the other end of the table we have the pre-World War-I players, indicating very low batting averages for batsmen playing at that time. Dale Steyn is a surprise inclusion, possibly because his last 54 wickets (over 50%) have been against the weaker batting teams of New Zealand, West Indies and Bangladesh, who have lower batting averages.

For a full list, please click here.

However let us seek to address this situation by looking at two other measures. The first is the difference between BQI and the career bowling average for the bowler. While it is true that having a high BQI means that the bowler has picked up better quality wickets, it might be more than offset by a high bowling average, which means the bowler has conceded a lot of runs for each wicket captured. The difference between these two figures will give a clear indication of the bowler's quality. The higher the difference, the better the bowler.

Table 1: Ordered by Difference between BQI and Bowling Average

SNo Bowler            Bow Cty   BowAvg  BQI    Diff

1.Marshall M.D RF Win 20.95 30.06 9.11 2.Davidson A.K LFM Aus 20.53 29.51 8.97 3.Ambrose C.E.L RF Win 20.99 29.85 8.86 4.McGrath G.D RFM Aus 21.64 30.43 8.79 5.O'Reilly W.J RLB Aus 22.60 31.12 8.53 6.Barnes S.F RFM Eng 16.43 24.58 8.15 7.Laker J.C ROB Eng 21.25 29.30 8.05 8.Croft C.E.H RF Win 23.30 31.22 7.91 9.Miller K.R RF Aus 22.98 30.65 7.68 10.Adcock N.A.T RF Saf 21.11 28.17 7.07

Ha! The list looks a lot more 'normal'. This is certainly a list of the outstanding bowlers of all time. Again, no mails bringing up other bowlers' names please. These are great bowlers who will stand comparison with anyone outside the list.

136.Giles A.F         LSP Eng   40.60  33.65  -6.95
137.Yadav N.S         ROB Ind   35.10  28.14  -6.96
138.Wright D.V.P      RLB Eng   39.11  31.06  -8.06
139.Boje N            LSP Saf   42.65  34.53  -8.12
140.Venkataraghavan S ROB Ind   36.12  27.56  -8.56
141.Emburey J.E       ROB Eng   38.41  29.69  -8.71
142.Abdul Razzaq      RFM Pak   36.93  27.66  -9.27
143.Shastri R.J       LSP Ind   40.96  31.69  -9.27
144.Mohammad Rafique  LSP Bng   40.76  31.35  -9.41
145.Hooper C.L        ROB Win   49.43  31.52 -17.91

Again, one feels vindicated. Boje is at the end with a huge negative difference. There is no denying that these are bowlers of average skills and in case of Mohammad Rafique, playing for a weak team. The last in this list is Carl Hooper, a very ordinary bowler indeed.

For a full list, please click here.

Another analysis I have done is to consider the number of lower-order wickets taken by a bowler and determine a % of lower-order wickets taken.

Who is a lower-order batsman? For the purpose of this exercise, I have defined it as a batsman batting at positions 8 to 11, and averaging less than 25 [to take care of situations when a Adam Gilchrist or Kapil Dev or Ian Botham might have batted at No. 8 or lower]. Only Daniel Vettori, with a batting average of 26.39, goes out of this classification. But then who can say that Vettori, with two Test centuries, is not an allrounder.

Initially I did this analysis based on batting average. However, that distorted the complete picture since the batting averages of batsmen during pre-WW-I and recently those from Bangladesh and Zimbabwe have been quite low. Hence I have gone back to the batting position.

In addition, any nightwatchman, determined through a proprietary algorithm, is considered as a lower-order batsmen.

Table 3: Ordered by % of lower-order wickets

SNo Bowler            Bow Cty Mat Wkts LowOrder Wkts
& %age

1.Zaheer Khan LFM Ind 53 170 23 (13.5) 2.Collinge R.O LFM Nzl 35 116 18 (15.5) 3.Boje N LSP Saf 43 100 16 (16.0) 4.Martin C.S RFM Nzl 37 125 22 (17.6) 5.Edmonds P.H LSP Eng 51 125 22 (17.6) 6.Flintoff A RFM Eng 67 197 36 (18.3) 7.Reid B.A LFM Aus 27 113 21 (18.6) 8.Pathan I.K LFM Ind 28 100 19 (19.0) 9.Intikhab Alam RLB Pak 47 125 24 (19.2) 10.Ghavri K.D LFM Ind 39 109 21 (19.3) 11.Hall W.W RF Win 48 192 37 (19.3) 12.Mushtaq Ahmed RLB Pak 52 185 36 (19.5) 13.Allen D.A ROB Eng 39 122 24 (19.7) 14.Srinath J RFM Ind 67 236 47 (19.9) 15.Thomson J.R RF Aus 51 200 40 (20.0)

This is a very good table, showing bowlers whose tally of lower-order wickets is less than 20% of the career wickets. It shows the value of Zaheer Khan to the Indian attack, as also Flintoff, Martin and Pathan to their respective teams.

134.Vettori D.L       LSP Nzl  77  238    81 (34.0)
135.Gupte S.P         RLB Ind  36  149    51 (34.2)
136.Rhodes W          LSP Eng  58  127    44 (34.6)
137.Mallett A.A       ROB Aus  38  132    46 (34.8)
138.Johnson I.W       ROB Aus  45  109    39 (35.8)
139.Adams P.R         LSP Saf  45  134    48 (35.8)
140.Giffen G          ROB Aus  31  103    37 (35.9)
141.MacGill S.C.G     RLB Aus  42  203    74 (36.5)
142.Wardle J.H        LSP Eng  28  102    38 (37.3)
143.Noble M.A         ROB Aus  42  121    47 (38.8)
144.Briggs J          LSP Eng  33  118    50 (42.4)
145.Lohmann G.A       RFM Eng  18  112    52 (46.4)

At the end of the table, these are bowlers whose tally of lower-order wickets is greater than a third of their total. It looks as if these bowlers have often been brought in to clean up the tail. There are quite a few pre-WW-I bowlers. A surprise is the presence of Vettori and MacGill, especially, who seems to have been brought in to bamboozle the tail despite the presence of other fast bowlers.

For a full list, please click here.

Some possible reader queries are anticipated and answered below.

1. At what individual score does the bowler dismiss the batsman. It is true that, for the fielding team, it is better for a batsman to be dismissed at 10 rather than 100. However, that is a more complex computation and has been done in a different context.

2. Whatever happens, capturing Tendulkar's wicket, even when he is on 100, is invaluable since he is capable of scoring a lot more runs than, say, when Dinesh Kartik has scored 25. It has been assumed that Tendulkar's wicket is Tendulkar's wicket, whatever be his individual score. Also, it might be true, in certain cases, that capturing Brad Haddin's wicket at 10 might be more valuable than capturing Matthew Hayden's wicket when he is at 100.

3. What about current form? While it may be true a few matches back Sehwag's wicket was going quite cheaply, his potential, as shown in the Adelaide Test, can never be underestimated. It is also true that even when Rahul Dravid or Ricky Ponting are going through indifferent form, their wickets are extremely valuable, because of their potential to score big.

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

  • Rafael on February 2, 2013, 20:42 GMT

    This a great article on the aliticappon of analytics in sports. Another aliticappon of this approach comes in mind is to predict the winner. The difference in the runs/goals scored by two teams could be taken up as the target variable and should be regressed based on all possible relevant parameters. The p value for each estimated point within the confidence interval will predict how likely the team is to win/loose.Applications are endless.

  • Dawood on October 28, 2008, 2:58 GMT

    Wicket is a Wicket. Either of Brian Lara or of Shoaib Akhter. A good bowler is the one who terrorises the Batsman and produces awesome deliveries. Your analysis doesnt include those wrong decisions which were made by the Umpire's in favour of the Batsman. If only the Wickets of a good Batsman matters, then there should be no place for non-batters in the Team. They should just come and Bowl and not Bat. The Difference between an ordinary Bowler and a good Bowler is the Number of different Deliveries He can Bowl, His accuracy, His wickets, His average and His consistancy. Not how many wickets He has taken of good Players or bad Players.

  • dinesh on March 30, 2008, 12:53 GMT

    I think this is a excellent article. It shows who are truly the great bowlers. It has already came in my mind that Zaheer khan would top the list.

  • Ravishankar on March 21, 2008, 14:31 GMT

    - Good analysis. But, why are you penalising bowlers for picking up tailenders /batsmen with low averages. Afterall, its not their fault. Why dont you limit the analyses to wickets by batting order. That is, consider only wickets upto No.7 Bat. Discarding all rest of the wickets?

  • Nikhil Pradhan on March 21, 2008, 3:51 GMT

    Nice article. but a good bowler has to be selected on the number of top order wickets he gets.but he should also get early wickets so that the other team is under pressure from the start. Whats the use in mopping the tail and get 4-5 wickets when the top order has already scored 400 runs. Its of no help.

  • Oz Parvaiz on March 21, 2008, 0:58 GMT

    Tremendous article. The problem with statistics, however, is that it captures performance over time, essentially averaging out greatness over a number of years. Therefore, the peak of a player is not captured (unless you're Bradman of course). For instance, not matter how many wickets Warne has taken, or how high Marshall ranks on your list, I don't know anyone in history that can match Lilee and Imran between 1980 and 1984. Imran took 40 wickets in a series in India for God's sake. He could make the ball bounce and swing on quicksand (ask Gavaskar if you don't believe me). I have personally never seen any bowler dominate with speed, variation and swing the way these two did during that time period. Maybe Marshall was more effective for a longer period of time, but nobody was as good as Imran and Lilee at their peaks ... not Akram, not Warne, not Holding, not anyone. If it wasn't for Lillee's back and Imran's shin they would have captured 500 long before Hadlee.

  • Dhammachintak Neel on March 20, 2008, 20:16 GMT

    The problem, as highlighted here, is that the good bowlers tend to "also" get the lower order wickets. In that case, the average isn't the best way out. On the contrary, "slabs of batsman's averages" should be considered. For eg., BQI for for those wickets having batting average between 10-20, 20-30, 30-40, 40-50, 50-60, 60-70, 70-80,80-90, 90-100. Also, the batting average should be considered for only those batsmen who have played atleast 'x' no. of matches (say 30) so that the average gives a rather real picture. Now for each of the slabs, have a weightage. Eg. 10-20 has weightage 0.1 and 90-100 has a weightage of 0.9. Thus, summation of product of weightage and BQI over all slabs gives a far better picture.

  • Atul Bhogle on March 20, 2008, 14:23 GMT

    Very interesting analysis and some very good comments. While it is true that statistics can never give you a 'correct' picture, the beauty of Cricket is lies in such statistical games which can give us the only realistic chance of pitting players against each other regardless of the era they played in.

    While the suggestion to consider only the batsman's average upto the point of the match is correct, I think the same should apply to bowlers also. This way we would be rewarding bowlers who were match winning material right from their first match. This would pit a bowler of X caliber against a batsman of Y caliber, and only then is comparison possible.

    Though s match by match statistic of such a comparison would have too much data, a consolidated list would be interesting to compare against what we already have.

  • Greg Brown on March 20, 2008, 12:26 GMT

    I think a better calculation would be to compare the number of runs/wicket conceded to a particular batsman compared to that batsman’s average/wicket.

    So for example if a bowler dismisses Bradman 3 times in his career he would expect to be hit for 300 runs while taking those wickets. The bowler would gain a point for every run less then 300 conceded to Bradman, and lose a point for every run more than 300. Any runs conceded to batsmen where no wicket is taken is points added to the total.

    An average bowler should have a score of zero, with better than average bowlers having negative scores.

    We could also see who particular bowlers have had the most success against. I suspect that Atherton would be a good source of points for McGrath!

  • Muhamad Ajmal on March 20, 2008, 6:43 GMT

    I believe we can check a bowler's capability during a test match but in this way we can't decide whether he take top class wickets,but how often he take the the wickets of clss batsman with his own art not when he made 100+ & got out with any silly selection of short playing. if i have a fair view of the past then I will found 2 bowlers with the capability to get the batsman when ever he want then there w/b 2 names 1 is great Waseem Akram & 2nd is legendary S.warne & if I look at the present bowlers then there is no name except all the time great Murli & new chap from Pak. Asif.

  • Rafael on February 2, 2013, 20:42 GMT

    This a great article on the aliticappon of analytics in sports. Another aliticappon of this approach comes in mind is to predict the winner. The difference in the runs/goals scored by two teams could be taken up as the target variable and should be regressed based on all possible relevant parameters. The p value for each estimated point within the confidence interval will predict how likely the team is to win/loose.Applications are endless.

  • Dawood on October 28, 2008, 2:58 GMT

    Wicket is a Wicket. Either of Brian Lara or of Shoaib Akhter. A good bowler is the one who terrorises the Batsman and produces awesome deliveries. Your analysis doesnt include those wrong decisions which were made by the Umpire's in favour of the Batsman. If only the Wickets of a good Batsman matters, then there should be no place for non-batters in the Team. They should just come and Bowl and not Bat. The Difference between an ordinary Bowler and a good Bowler is the Number of different Deliveries He can Bowl, His accuracy, His wickets, His average and His consistancy. Not how many wickets He has taken of good Players or bad Players.

  • dinesh on March 30, 2008, 12:53 GMT

    I think this is a excellent article. It shows who are truly the great bowlers. It has already came in my mind that Zaheer khan would top the list.

  • Ravishankar on March 21, 2008, 14:31 GMT

    - Good analysis. But, why are you penalising bowlers for picking up tailenders /batsmen with low averages. Afterall, its not their fault. Why dont you limit the analyses to wickets by batting order. That is, consider only wickets upto No.7 Bat. Discarding all rest of the wickets?

  • Nikhil Pradhan on March 21, 2008, 3:51 GMT

    Nice article. but a good bowler has to be selected on the number of top order wickets he gets.but he should also get early wickets so that the other team is under pressure from the start. Whats the use in mopping the tail and get 4-5 wickets when the top order has already scored 400 runs. Its of no help.

  • Oz Parvaiz on March 21, 2008, 0:58 GMT

    Tremendous article. The problem with statistics, however, is that it captures performance over time, essentially averaging out greatness over a number of years. Therefore, the peak of a player is not captured (unless you're Bradman of course). For instance, not matter how many wickets Warne has taken, or how high Marshall ranks on your list, I don't know anyone in history that can match Lilee and Imran between 1980 and 1984. Imran took 40 wickets in a series in India for God's sake. He could make the ball bounce and swing on quicksand (ask Gavaskar if you don't believe me). I have personally never seen any bowler dominate with speed, variation and swing the way these two did during that time period. Maybe Marshall was more effective for a longer period of time, but nobody was as good as Imran and Lilee at their peaks ... not Akram, not Warne, not Holding, not anyone. If it wasn't for Lillee's back and Imran's shin they would have captured 500 long before Hadlee.

  • Dhammachintak Neel on March 20, 2008, 20:16 GMT

    The problem, as highlighted here, is that the good bowlers tend to "also" get the lower order wickets. In that case, the average isn't the best way out. On the contrary, "slabs of batsman's averages" should be considered. For eg., BQI for for those wickets having batting average between 10-20, 20-30, 30-40, 40-50, 50-60, 60-70, 70-80,80-90, 90-100. Also, the batting average should be considered for only those batsmen who have played atleast 'x' no. of matches (say 30) so that the average gives a rather real picture. Now for each of the slabs, have a weightage. Eg. 10-20 has weightage 0.1 and 90-100 has a weightage of 0.9. Thus, summation of product of weightage and BQI over all slabs gives a far better picture.

  • Atul Bhogle on March 20, 2008, 14:23 GMT

    Very interesting analysis and some very good comments. While it is true that statistics can never give you a 'correct' picture, the beauty of Cricket is lies in such statistical games which can give us the only realistic chance of pitting players against each other regardless of the era they played in.

    While the suggestion to consider only the batsman's average upto the point of the match is correct, I think the same should apply to bowlers also. This way we would be rewarding bowlers who were match winning material right from their first match. This would pit a bowler of X caliber against a batsman of Y caliber, and only then is comparison possible.

    Though s match by match statistic of such a comparison would have too much data, a consolidated list would be interesting to compare against what we already have.

  • Greg Brown on March 20, 2008, 12:26 GMT

    I think a better calculation would be to compare the number of runs/wicket conceded to a particular batsman compared to that batsman’s average/wicket.

    So for example if a bowler dismisses Bradman 3 times in his career he would expect to be hit for 300 runs while taking those wickets. The bowler would gain a point for every run less then 300 conceded to Bradman, and lose a point for every run more than 300. Any runs conceded to batsmen where no wicket is taken is points added to the total.

    An average bowler should have a score of zero, with better than average bowlers having negative scores.

    We could also see who particular bowlers have had the most success against. I suspect that Atherton would be a good source of points for McGrath!

  • Muhamad Ajmal on March 20, 2008, 6:43 GMT

    I believe we can check a bowler's capability during a test match but in this way we can't decide whether he take top class wickets,but how often he take the the wickets of clss batsman with his own art not when he made 100+ & got out with any silly selection of short playing. if i have a fair view of the past then I will found 2 bowlers with the capability to get the batsman when ever he want then there w/b 2 names 1 is great Waseem Akram & 2nd is legendary S.warne & if I look at the present bowlers then there is no name except all the time great Murli & new chap from Pak. Asif.

  • Don on March 20, 2008, 6:23 GMT

    Another suggestion. The players average coming into the match does give an indication of how accomplished that player is. However, you have tried to establish how VALUABLE that wicket is. On paper, sure a batting average is an indication of the importance of the player, but in the field it is often the wicket breaking a large partnership that is the most important. It is well known that wickets often fall in bunches. If you have an established pair in the middle, it is very difficult to pry them out. However, once you have a new player in, often another wicket will soon fall. Instead of taking batting AVERAGES going into the match, why not try taking the worth of the partnerships for that wicket in the match. If the match is a run-fest, the bowlers will not be taking many wickets anyway, so their stats will not be that great. The true value of the bowler will be in breaking those bothersome partnerships that can potentially win the match.

  • Sam "Madlib" on March 20, 2008, 5:30 GMT

    i true representation of the best bowlers of all time would be to interview every player who scored a qualifying level of runs say 4500 at test level and ask them the best bowlers they faced giving a quantification of points.... then you would get a picture of the bowlers who dominated their era's. Surely a bowlers level of skill can only be determined by his opposition... of course this would elimate Jim Laker who was a legend but would shed sum sweet light on the modern era

  • Dan on March 20, 2008, 0:00 GMT

    being an avid kiwi and therefore a NZ fan, gutted not to see shane bond on that list as he consistently dismisses quality batsmen, esp against india and australia, pity about the 100 wickets qualification

  • Naveed on March 19, 2008, 23:50 GMT

    Interesting analysis. One thing that one should keep in mind is that sometimes a mediocre bowler gets a high profile scalp b'ce of the pressure of top class bowlers on the other end and batsman's attempt to score of at the other end. The other thing is that wraping up the tail is an art too which should not be taken lightly too. There are bowlers who are masters at that with some fierce diverly wepon up their sleves which can not be so easily done by an average bowler. In conclusion their is no perfect way to gauge bowlers and old fashioned way probably reflect the best picture.

  • Kotesh on March 19, 2008, 23:32 GMT

    I am assuming that you have included the batting average of each batsman dismissed by a bowler only once in the "Sum of batting average". If my assumption is right, the result could be skewed based on the number of different batsmen the bowler dismissed and how may times a bowler dismissed the same batsmen. The higher the number of times a bowler dismisses the same batsman, the less likely he will show up in this list. The bowlers who have dismissed more number of batsmen will have a better chance of featuring in this list. Let's say Wasim Akram dismissed Tendulkar 10 times and Tendulkar's batting average is included in the "Sum of batting Average" only once, the resulting "Sum of batting Average" would be less by 500 than what it would have been had Akram dismissed 10 different batsmen with an average of 50. If we multiply the number of times a batsman was dismissed by the bowler with the batsman's average and inlclude that in the "Sum of Batting Average", we would get better results.

  • Nishant on March 19, 2008, 16:56 GMT

    Very interesting work Ananth. As someone already pointed out, this proves that there is no one measure that defines a bowler's greatness. I wonder if such an analysis would be useful way to look at a bowler's performance in a specific series for the use of captain and team management and somehow take this into consideration while choosing teams. I realize that it can make the process very complex but I will be curious to have such an insight. Once again, very thoughtful and impressive work. Thank you.

  • Yasir Hasan on March 19, 2008, 11:05 GMT

    Your try is good, but lets add one more factor i.e. how much batsman has scored before getting out. Like a bowling out an opener for 0 is much more valuable than bowling him out on 50 or 100.

  • Mohsin Khan on March 19, 2008, 10:43 GMT

    Good read but I think the analysis is flawed. Using career batting average is meaningless as a lot of batsmen tend to get better with age. Dravid,Inzimam, Yousuf....etc pre 2000 vs the three post 2000 is a big difference.

    A better analysis would be the batsman's average at time of dismissal.

  • CKR on March 19, 2008, 10:15 GMT

    Yes I agree, finishing of the tail is important and it was the biggest problem Indian bowlers have always had bcos of which many tests have drawn that could have been Indian victories.

  • Rohit on March 19, 2008, 10:09 GMT

    I think the impact a bowler makes during a test match is not just decided by whether they take top class wickets,but how soon they take those wickets.So if a batsman makes 150 runs and gets out trying to slog a spinner like Boje hoping to get a declaration,it should not be worth the same as say Mcgrath dismissing a batsman when he is on a single digit score.So in my opinion you should add (Batsman's career average - Runs scored by the batsman in the innings) for every wicket taken by the bowler to compute the final BQI.This might lead to negative BQI's for mediocre bowlers,and might penalize some bowlers for ordinary bowling by others or not getting a spell soon enough,but it will be a fairer indication of a bowler's strike capability,which is the ability to get great batsmen out cheaply.A final subtraction from the bowling average might then not even be required.

  • hastagiri on March 19, 2008, 9:33 GMT

    one reason why table one might not be 'normal' is these bowlers who are not all time greats might be used as change bolwers and the batsmen took off concentration while facing them or tried to attack them and got out.

    an interesting way to analyse further would be to calculate averages of batsmen against the bowler who dismissed them and compare it with their career avarage and this gives how well the bowler has bowled. for example if batsman A with career avarage of 50 has scored 700 runs against bowler B getting out to him 10 times his average against that bowler is 70 and this -20 gets added to the bowler aggregate. the total aggreagate of this per batsman average difference divided by number of wickets talks about the effectiveness of the bowler. note that for every batsman a bowler has bowled against and not dismissed, the batsman's career average must be added to the bolwer's aggregate.

  • Heath on March 19, 2008, 9:17 GMT

    i wonder what the list would look like if it was for bowlers with over 200 wickets? I think this would help clear out some players who arent greats. I like the list of the difference between average and BQI but it still does penalise players who have taken a lot of tail end wickets - possibly including walsh and warne, as two examples who were good at clearing the tail - a skill any captain would value.

    I guess all this proves that there is no perfect method - but i do love the concept of trying to develop a comparison between the bowlers of different eras. That is the beauty of the difference list - it is less skewed towards the modern era. Obviously a huge part of the reason that the overall list features lesser players from today is that for a variety of reasons batting averages are higher these days. I very much doubt that this somehow proves modern batsmen to be better, and hence, it doesnt prove modern bowlers to be better.

  • Ganesh Murthy on March 19, 2008, 7:56 GMT

    Again, taking the wicket of a top class batsman need not say anything about the effectiveness or greatness of a bowler. You may take his wicket after he has already hit a good score. So a real good measure for an effective bowler would be the strike rate for the wickets of top class batsmen, that is, the number of balls bowled by the bowler per top class batsman wicket.

  • Hareesh on March 19, 2008, 4:03 GMT

    I liked this analysis, although seeing Boje and Giles top the first list almost put me off. I agree with most of the metrics and appreciate the effort and idea. Some more quantifiable metrics I would like to see are using batsmen averages in that country (getting Inzy out in Aus is easy), preceding partnership score (maybe biased against new ball bowlers).

  • abc on March 19, 2008, 3:22 GMT

    Mopping tail is an art. You can't penalize a bowler for that. You are actually rewarding bowlers like Zaheer Khan who can't mop up the tail.

  • Rajan Srinivasan on March 19, 2008, 0:47 GMT

    extremely well analysed with interesting perspectives; further improvement difficult because it would require something on these lines: a) runs scored byof batsman dismissed as percentage of runs scored in innings b) runs scored by batsmen dismissed as percentage of runs scored by team in series. This would improve standing of bowlers in the 'Golden Age' of uncovered wickets, when innings totals were less than in the inter-war period of dead wickets and tall scores. Quality of batting in the earlier period ws ranked much higher by even Frank Woolley who said that only Hammond of the inter-war greats could really bear comparison with any of the great English batsmen of the pre-WWI era (Hobbs overlapped both ages). So the quality of the batting needs to be factored in with greater equalization. Great job! I loved the analysis. Keep it up.

  • Adrock on March 18, 2008, 23:44 GMT

    Another aspect you can never take into account is the quality of the bowling attack as a whole - if competition is fierce for wickets (think McGrath, Warne, Gillespie, Lee days) this affects how many of those prized top order wickets X bowler CAN get each game. Also a bowler who excels far beyond the attack (Murali anyone) and consequently ends up getting a lot of wickets of all sorts soaks up those lower BQI players too due to LACK of competition for them, dragging their average down. Food for thought, but otherwise it's a much more realistic analysis then the normal bowling average :)

  • Aaditya on March 18, 2008, 22:57 GMT

    This is a very interesting way to figure out the quality of a bowler. You guys have developed several criteria for calculating the ability of a bowler. How about bringing together all your new criteria, along with the aldready present bowling average, and create a table that incorporates EVERYTHING. I believe that doing this could really bring out the cream of the crop when it comes to evaluating bowlers.

    PS- Your 'improved' batting average is so good that it should be implemented in official cricket statistics!

  • Onkar Walavalkar on March 18, 2008, 20:09 GMT

    One of the comments proposes an interesting formula for BQI (sum of averages of batsmen dismissed / bowling average). I don't completely agree with this because this biases it towards great bowlers (with lower bowling averages). We do want to rate the top-order specialists at the top of the list (In some cases a bowler might not possess a very good overall average but consistently dismisses top order batsmen). Which is why I don't think we should use bowler's average or strike rate (that introduces a bowler's overall ability as a parameter). I also don't agree with your last table of percentage of lower order wickets dismissed, since it punishes people for taking lower order wickets. We want to award people for taking top order wickets, but not punish them for taking lower order wickets.

  • Onkar Walavalkar on March 18, 2008, 19:58 GMT

    Another interesting way of calculating this is to count the averages of batsmen the bowler has gotten and divide it by the number of innings in which the bowler has bowled. So in my earlier case the BQI of A will be 300, B will be 340, C will be 40 which will give us the correct order of B,A,C. The number of innings in denominator will ensure that we do not bias it towards bowlers who have played many matches. It may bias it against the pre-WW bowlers though.

  • Onkar Walavalkar on March 18, 2008, 19:51 GMT

    Consider this - Let's say each of the top 6 batsmen of a team averages 50. Each of the next 5 averages 10. Consider 3 scenarios (where no. 11 remains unbeaten each time). Bowler A gets the top 6, giving him a BQI of 50. Bowler B gets all 10, giving him a BQI of 34 (50 * 6 + 10 *4 / 10) Bowler C gets the last 4, giving him a BQI of 10. Your 1st calculation will list them in the order of BQI in the order A,B,C, whereas what we are really looking for is B,A,C. In short we are punishing B for taking the last 4 wickets.

  • Jaspal Singh on March 18, 2008, 13:50 GMT

    In some ways the analysis confirms what I thought about Wasim Akram, that he was rated rather too highly and struggled to dismiss the best on a consistent basis. The system here does tend to favour fast bowlers over spinners and disproportionately favours the more modern bowler in an era of high batting averages.

    How do we factor in the kind of pitches the bowlers played on ? Imran and Hadlee had very similar averages, but the former bowled, for the mosty part, on dead subcontinental tracks which even Lillee could barely perform on, whereas Hadlee had tracks tailor-made for medium-pace, Ilford seamers i.e. the ratings favour pace bowlers from countries with pitches that are either bouncy or those that afford rather more lateral movement.

    Factoring in the bowling at the other end and the team being played for would make the analysis even more convincing.

  • George on March 18, 2008, 13:12 GMT

    Very interesting article. Top effort. It would be interesting to see stats of teams that clean up/struggle in getting the tail/batsman out, i.e. stats of team loosing focus/going for the kill.

  • Max on March 18, 2008, 11:40 GMT

    This is wrong, because the better batsmen will tend to play the better bowlers with more caution and not get out , whereas they will try and hit out or mentally relax against 'weaker' bowlers like Boje. Hence you see the skewed average.

  • Hamza on March 18, 2008, 10:06 GMT

    Do you consider the batting average of batsmen dismissed up till that innings or for the career. If you take the career average then you do not get the quality of the player at the time he was dismissed.

  • Ranadurjay on March 18, 2008, 9:21 GMT

    For BQI calculations, when you sum the averages of batsmen dismissed, you should take not their career averages, which you seem to have done here. Instead, you should consider their average till the match when they were dismissed.That way, the wicket of the same batsman in different phases of his career gets different weightages, as should be the case.

    Also, if you have taken the difference of bowling avg. and BQI for your final list, why don't you also consider strike rate as a factor? Surely, how cheaply you get your wicket is as important as how many deliveries you bowl to get your wickets, isn't it?

    You can maybe do an average/median of bowling avg., BQI and strike rate. That might throw up some interesting results.

  • Qamar on March 18, 2008, 7:59 GMT

    My main problem with the analysis, albeit very good, is that it is based on a bowler's 'luck' at taking wickets. I believe this fundamental premises is inadequate. I have seen bowlers bowl excellent spells, where they could have taken 3-4 wickets, but ended with none. And sometimes, its been (often mediocre) bowler at the other end that ended up with 2-3. I guess what I am saying is that the number of wickets is not necessarily the best criteria to judge a bowlers ability, and sometimes the only justice a better bowler gets is 'bagging the tailender wickets'. Regardless, your analysis is definately an improvement over the simplistic view. I wish someone re-writes the batting averages into something more meaningful; maybe a 'modal range', or something based on a weighted index of current form, or a index that takes the Team total into account, so that your average is a percentage of Team's total score.

  • Mani Ramaswami on March 18, 2008, 6:36 GMT

    I think batting average against the specific bowler, divided by career batting average would give a different, but more meaningful indication of how good the bowler has been. Or a difference (instead of the quotient) between the career batting average and average against this bowler. It would be difficult for an average bowler to look good in such a "relative efficacy index."

  • HLANGL on March 18, 2008, 5:08 GMT

    Whatever the stats may be, Marshall & Ambrose were far superior bowlers compared to the other much hyped ones of modern era. The early part of Waqar (say his initial 30-31 tests) may have promised even greater but he deviated from that afterwards. Anyway, a great qualitative analysis on part of Ananth!!!.

  • Andrew on March 18, 2008, 2:57 GMT

    When I first read the headline it I thought it would be a measure of the quality of the dismissal not the batsman dismissed which may explain Boje's constant appearance. This ignores the fact that bowlers often feed off other bowler's pressure so while some may bowl unplayable balls that miss the edge by a fraction, a batsman like Lara or Ponting may see Boje as a chance to attack and end up mistiming a long hop to midwicket, hardly a wicket that suggests quality bowling. KP's dismissal of Taylor in the first test is an example of this.

  • GD on March 18, 2008, 2:05 GMT

    Best results: Ordered by Difference between BQI and Bowling Average

    This would be the best ever list for mine.

  • PMcG on March 18, 2008, 1:35 GMT

    Answering my own earlier comment, it appears that the list undervalues spinners. In a team placing the emphasis on its good spinners, it is likely that those spinners will bowl throughout the innings and so pick up a larger number of tail-enders, so placing them lower on the list. Similarly, Australia often used McGrath in combination with Warne in the fourth innings of a match to finish off the tail.

  • PMcG on March 18, 2008, 1:25 GMT

    Thank you for the analysis. I query whether there is any analysis which doesn't penalise the members of the great WI teams of the 70's and 80's in which there was an abundance of great fast bowlers. There were so many of them that they had to share the wickets and so no one bowler could dominate. I am also curious as to the reason why Indian bowlers are so under-represented.

  • Charles Davis on March 18, 2008, 0:26 GMT

    Interesting analysis, Ananth. One important point you or your commenters haven't empahasised is that the highest percentage of tailend wickets will mostly be in teams that bowl their opposition out effectively. Note that the majority of bowlers with high tailend % come from pre-World war I when scores were lower and declarations almost unknown.

    For this reason you will never find a Bangladesh bowler, even a spinner, with lots of tailend wickets - they don't bowl to tailenders often enough - but guys like Shane Warne and Stuart MacGill have very high proportions.

    Bangladesh bowlers bowl their opposition out in only 36% of the innings they bowl, whereas Australia in recent years has done so in 83% of their innings.

  • bullfrog on March 18, 2008, 0:21 GMT

    Excellent post. One criticism - the computations favour fast bowlers over spinners. Often when spinners come on to bowl several of the top order batsmen have already been dismissed thus minimising their ability to dismiss top order batsmen.

  • Ralph on March 17, 2008, 19:28 GMT

    This blog fascinates me in one particular way: the quality of any particular analysis seems to be largely judged on how closely it reflects existing opinion, and the analysis is nudged and nurdled until it reflects it pretty closely! I think the test of a really good, thought-provoking analysis would be one which throws a cat amongst indisputable pigeons! For example, you might have four legends averaging ~20 and someone averaging 30 but with some other quality (maybe someone along the lines of Rafique, but a less extreme case).

  • cricBoy on March 17, 2008, 18:58 GMT

    Another simple way could have been:

    Bowler Quality Index (BoQI) = (sum of career batting averages of the batsmen dismissed) divided by the bowler's average.

    The higher the BoQI, the better the bowler.

  • Goatman on March 17, 2008, 17:20 GMT

    Another classic from my favourite column!

    Two unresolved issues though:-

    1) The "difference" number does not really mean anything in game terms, being the difference between a batting average and a bowling average, which are computed from different data. Could a replacement for the BQI be found by analysing how difficult bowlers find removing the batsman - find all bowlers who have bowled to them, sum the number of dismissals, sum the number of runs conceded and divide as in a bowling average. Numerically this may make little difference, but logically it is important.

    2) The "difference" term is surely elastic, dependant on the batting average of the day? Low contemporary batting average will lower both bowling average and BQI, thus lowering potential difference. Sidney Barnes was a genius, but would probably average in the low 20's today. Normalising for contemporary batting average (increasing Barnes's, decreasing Hoggards) may help. This also informs Point 1.

  • Mark on March 17, 2008, 16:51 GMT

    Top post - sure to spark lots of debate (as it has).

    But it doesn't take into account ability of bowlers to spark a wicket to fall - how about something involving wickets taken during a bowler's spell (i.e. bowler creates pressure when bowling, wicket falls at other end; and also run outs taken while a bowler is on).

    Also, in reply to earlier posts, if you limit the range to 200+ wickets you also miss out on very good bowlers who gave their all to the cause but came off injured (e.g. Simon Jones).

  • Haridas Rao on March 17, 2008, 13:01 GMT

    the article does ticckle the readers interest.however such complcated stats are not needed to know the true standing of ashley giles , nicky boija and merv dillon vis-a-vis shane warne , freddie trueman anil kumble and almost any other bowler regarded as truly great . the article simply proves the point that statistics can be used to reach any desired result

  • KM on March 17, 2008, 12:51 GMT

    A really innovative approach to quantify a bowlers rating...at the end of the day...would agree that what happens in the field would matter.

    Like a couple of other readers have mentioned, should the accountability of the partnership broken be an important criterion?

  • vaseem on March 17, 2008, 10:02 GMT

    I agree with one of the other comments. If a bowler takes a 5-fer, including the top three and two tail-end wickets, he will have a lower rating on your table than one who took just the top three. This is an anomaly that explains why Flintoof rates so highly and Akram (+ other greats) so low - it especially discrimates aganist bowlers who are effective against the tail. This can be corrected either by having a sliding scale, or by looking at the proportion of wickets that a bowler (with over 200 wickets please) has of the top order. Or perhaps a better way would be to look at some stat such as strike rate x average x proportion of top order wickets; which effectively looks at both the quality of wickets without disadvantaging those who were great against the tail

  • Rosh on March 17, 2008, 9:39 GMT

    Great stuff Ananth. SKchai and Nalin's takes are also v. interesting. In fact I like Nalin's analysis and of course the final result ie. Hadlee followed by Marshall - no howls there. Hadlee - lone shining beacon and Marshall - shone above the other beacons. Brilliant!! Personally, statistic's aside, for me, Dennis Keith Lillee was the mightiest of them all. By the way P.J. Joseph you are surely saying Murali's "overs are rated", er, very high. Well they are - that is why the Aussies and some other no-gooders take to whining away when he gets to tour Australia and when he nears milestones. It happened just last November. Its their way of facing the Murali-threat ie. affect his mental make-up. Once again its "Murali's-overs-are-rated-high" - hope you got it!!! Good luck with future messages.

  • Rosh on March 17, 2008, 9:35 GMT

    Great stuff Ananth. SKchai and Nalin's takes are also v. interesting. In fact I like Nalin's analysis and of course the final result ie. Hadlee followed by Marshall - no howls there. Hadlee - lone shining beacon and Marshall - shone above the other beacons. Brilliant!! Personally, statistic's aside, for me, Dennis Keith Lillee was the mightiest of them all. By the way P.J. Joseph you are surely saying Murali's "overs are rated", er, very high. Well they are - that is why the Aussies and some other no-gooders take to whining away when he gets to tour Australia and when he nears milestones. It happened just last November. Its their way of facing the Murali-threat ie. affect his mental make-up. Once again its "Murali's-overs-are-rated-high" - hope you got it!!! Good luck with future messages.

  • UM BAJWA, Islamabad on March 17, 2008, 7:27 GMT

    A good effort indeed. You have tried tackling a very complex task. Your analysis, however, is devoid of any of the following important factors while rating all-time bowlers, which I do concede as extremely cumbersome to incorporate:

    1. The importance of a wicket taken in a special situation vs. an ordinary situation. eg. a top order wicket during the ashes series decider or a regulation wicket against Zimbabwe/Bangladesh.

    2. It does not take into account the type/quality of pitch/surface on which the wicket was taken. Surely, someone getting Ponting out on a belter of a pitch bouncy/seamy(Perth) as against taking his wicket on Lahore's batting friendly patch shall not be given the same weightage.

    3. Age factor. Getting a five-for haul at age 33 shall weigh more than getting the same figures at age 23.

    4. The overall number of tests played by the bowler. The more the tests played the better the bowler, it shows his consistency in being selected for more series over longer period.

  • Ghufran Ahmad on March 17, 2008, 5:26 GMT

    The reason why we don;t see to names like Malcom Marchal, Garner, Roberts, Waseem, Waqar in the list is in my opinion is that the teams like Pak and West Indies and Australia had more than 3 quality bowlers sharing wickets among them, whereas for other teams like India, Bangladesh there was only one main bowler capable of taking quality wickets although the runs given by him were enormous for those wickets. So their names appear in the list. The quality of bowler does not depent on getting a quality batsman out. A mediocre bowler might be punished all day giving away a lot of runs and at the end of second day same bowler may get him out because batsman was too tired. This will not make him a great bowler.

  • Binu Thomas on March 17, 2008, 4:48 GMT

    Doesn't the table 1 shows one thing clearly, at least for a statistically minded person like you- that this method is inherently wrong? May be this is simply not the way to calculate the quality of wickets taken by bowlers.

  • James on March 17, 2008, 2:03 GMT

    A point I haven't seen in the comments above (apologies if I missed it): early bowlers such as Lohmann are causally responsible for the batsmen of the opposing side having such low averages, since the fewer Test sides there are, the higher proportion of an opponent's batting time will be spent against any one team. So it seems more than a bit harsh to use the low averages of these batsmen as a reason to then discount Lohmann's bowling quality. Why not instead make your weighting based not on averages but on the relative ranking of the batsman in, say, a five year period?

  • amirali on March 16, 2008, 23:34 GMT

    Why not a simple analysis of the strike rate/average of a bowler against batsmen, broken down by their averages? E.g, strike rate/avg against batsmen with average 50-55, strike rate/avg against batsmen with average 45-50 etc.

    The measure as it is currently, is also quite good. It's reasonable to accord less value to a lower order wicket than upper order....I would prefer if instead of alloting his average though, it would allot the average value of the partnership that batsman is involved in. That would still accord less value to lower order wicket (as is sensible) but also take into account the contributions an established batsman at other end would make with him (on average).

  • Shailesh on March 16, 2008, 17:45 GMT

    As a student of statistics myself, it is a very interesting exercise you have done and tried to open up a topic which many may have missed.

    However, when a Lillie or a Thompson is bowling at you, (their) statistics do not matter.

  • Alfred on March 16, 2008, 15:28 GMT

    This is amazing stuff! However I agree that it is wrong to penalize a player for taking a lower-order wicket.

    Why not make it simple and direct: calculate each bowler's average number of top order wickets per match?

  • saurabh somani on March 16, 2008, 14:48 GMT

    ananth, ive not agreed with a lot of your analyses, but this one is superb.... great work. and i understand totally the need to keep it simple for a wider understanding, but perhaps you could also do a more complex analysis for greater cricketing integrity? you could put those up on your website!

  • genehackman on March 16, 2008, 14:25 GMT

    while computing the sum of batting averages to determine BQI, do you make sure you include a batsman's average mutiple times if the bowler got him out multiple times? if not, the BQI number is skewed.

  • Nalin on March 16, 2008, 13:09 GMT

    Even though this analysis is not perfect it will give some clear indication about the quality and greatness of the bowlers. For example take Murali(M),Warne(W),Hadlee(H) & Marshall(Mar).... diff between BQI & Boling Ave. M=6.01 W=3.26 H=6.79 Mar=9.11 .... % of lower order wickets M=27.0 W=32.6 H=27.8 Mar=22.3... BQI M=27.78 W=28.68 H=29.09 Mar=30.06. So this will say that In order of quality Marshall> Murali> Hadlee> warne... Marshall & Hadlee did not bowl to Zimbabwe & Bangladesh while Murali & warne did. Marshall & warne had good support while Hadlee & Murali did not. Warne did not bowl to the great Aussi Batting line while Marshall for the great WI batsmen. Both Hadlee & Marshall Bowled well against Australia. So who is the greatest ?? Australians say Warne..but he isn't.. Sri lankans say Murali but he isn't either.. so it should be a fight between Marshall & Hadlee. Since he has bowled well against Aus, WI, Eng, INd & Pak.. I like to Pick Hadlee as the best.

  • Yogesh on March 16, 2008, 12:29 GMT

    Here is an even better measure than my earlier comment : Instead of averages which is a period dependent thing why not take ICC ranking points. I guess now these are available for all matches & for all batsmen. Plus these will also exclude tailenders averaging high early in the career sometimes due to the no: of not-outs. These points are considered to be a true indication of batsman's quality. So why not add up the batting points of each batsmen upto the match he was dismmised ?

  • Eric M on March 16, 2008, 12:17 GMT

    You should also consider how early the bowler knocked off the batsman. If openers play until 40th over, I am sure 6 or 7 wickets will go down in the next 10 because of agressive hitting. Whereas if 3 or 4 wickets fall in the first 10 overs, the middle order will play off the next several overs to restore some sense in the batting.

  • ram on March 16, 2008, 10:15 GMT

    One thing that occurs to me is that you shouldn't just consider individual bowlers performance in isolation. It should be based on a relative performance based on the bowlers that have played in the match.

    On a flat pitch where 500 plays 500 a bowlers who gets 2 wickets for 80 runs is a creditable performance. This may not be a good analysis in a bowling friendly conditions. Your analysis goes someway towards that but it might be useful to see some measure of relative performance of the bowlers with regards to the conditions they play in and opposition they face.

  • Zeeshan on March 16, 2008, 9:50 GMT

    While considering the value of tail enders wicked, we could consider the following - time he spent on crease - the size of the partnership he participated in - the balls he faced or the runs he scored.

    I reckon these three factors summarize the value of a tail enders wicket.

    Ananth , could you please run a query using those three factors and show us a list of most valuable tail enders ?

    once we determine this , we could easily figure the BQI and then come up with a list of most effective bowlers in the history of cricket

  • Owais2010 on March 16, 2008, 8:19 GMT

    I think attack bowlers will always figure higher in this analysis, obviously Shane Warne, Macgill, Anil Kumble, Muralitharan, Abdul Qadir are not usually bowling when no 1, no 2, no 3 or no 4 are still fresh at crease ! Zaheer Khan will figure higher in this list compared to Saqlain Mushtaq though the later has truly been a match winner - remember Asian test championship and Pak-India series of 1999 ?

  • David Barry on March 16, 2008, 8:02 GMT

    Shantanu Dutt: The comment by skchai gives the results of what you suggest. Weighting each wicket by the batting average of the batsman dismissed and calculating A = (runs conceded / weighted wickets) is equivalent to finding B =(BQI / bowling average).

    B = overall_avg / A.

    Take the overall_avg as about 31.7 and apply it to skchai's numbers, and you have the figures you want.

  • Sushamna on March 16, 2008, 7:02 GMT

    I see one problem in calculating BQI, batsman's average is a variable, and not a constant, it keeps changing as his career progresses. e.g. when Sachin was first dismissed, his average was 15 and if he's dismissed now, his average would be ~ 55. Is this taken into consideration?

    One more suggestion, instead of just taking into account batsmen's average according to position, how about considering their contribution to total inning's score and then ordering them? a top order batsman may not be in good form always..

  • murali on March 16, 2008, 5:10 GMT

    Great analysis! I have a small doubt/query: suppose a lower order batsman gets together with a top order batsman and puts on 100+ rather irritating runs (recent examples - Symonds and Hogg in the Sydney test, Misbah and Sami in the Delhi test) wouldn't the main concern be to dislodging the partnership? In the sense, the tail-ender's wicket would be as vital as the top order batsman's.

  • Arpan on March 16, 2008, 1:32 GMT

    An alternative would be to sum up the averages of the batsmen dismissed by a bowler (for each wicket)and divide this by the number of tests/innings played by the bowler (instead of the number of wickets). This would give us an idea of the bowling average and also the quality of batsmen dismissed by a bowler. The problem is that batting averages keep varying over the decades.

  • Shantanu Dutt on March 16, 2008, 0:52 GMT

    Post 3 of WQ metric:

    Finally a bowlers average can be computed as his WQ/(runs given) [instead of wickets/(runs given) as is done currently].

    Ananth:

    It would be fascinating to see how the WQ metric pans out. At the risk of sounding a bit biased towards my own metric, let me say that it is a very good one plus simple to compute. Would it be possible to program it in (and its associated metrics given in my posts) and let us know the results? Thanks.

    Shantanu

  • Shantanu Dutt on March 16, 2008, 0:44 GMT

    Post 2 of WQ metric: Now consider that in a match bowler A has taken 1 wicket of a batsman w/ an ave of 50 and another wicket w/ an ave of 10, & that another bowler B has taken only 1 wicket of a batsman also w/ an ave of 50. Clearly, bowler A has done better than B. However, the BQI metric for A is (50+10)/2 = 30 while that for B is 50. Now w/ the WQ metric: Assume Ave_all = 20. Then, the WQ metric for A is 50/20 + 10/20 = 3, and the WQ for bowler B is 50/20 = 2.5. This is much more reasonable (indeed correct).

    Note also that the WQ metric can accommodate some other reasonable suggestions. Thus if we want to determine batsmen's worth by eras w/ significantly different averages (say, different by at least 20%), then AVE_all can be replaced by AVE_all_era-X-Y which is the ave of all batsmen from years x to Y (of course, the batsman has to belong to that era :-)). Similarly, instead of a batsman's career average, his average at the time of his wicket in some innings can be considered.

  • Shantanu Dutt on March 16, 2008, 0:29 GMT

    Ananth has made a reasonable attempt at quality of wickets. But, in doing so he has transformed the "wickets" unit to the "runs" unit leading to the flaw that some have pointed out: it penalizes bowler A if he has taken x top-order batsmen (w/ say ave runs of "m") plus y lower-order batsmen w/ ave runs of alpha*m where alpha < 1 (by definition) compared to bowler B who has also taken x top-order batsmen w/ the same ave of "m" runs but no lower-order wickets.

    The way to solve this is keep the unit as "wickets" but to have a sum of "weighted wickets" as the quality metric. A batsman C's value (or weight) is calculated as the "normalized average( Norm_ave)" = (ave of C( / (ave AVE_all of all batsmen). So if AVE_all = 20 and C's average is 50, C's wicket is worth 2.5 average batsmen's wickets and if D's ave is 10, then D's wicket is worth 1/2 that of an average batsman. Then a bowler's "wicket quality (WQ)" = sum of the Norm_ave's of all the batsmen whose wickets he has taken. (Contd.)

  • TK on March 15, 2008, 22:24 GMT

    Wasim Akram is the greatest bowler ever!!!!! well done great article

  • Warwicks fan on March 15, 2008, 22:23 GMT

    Very interesting analysis, yet it does seem that something is not quite right. Table two seems the most 'reasonable' one of the three, based on knowledge of cricket and it's history, if only because Boje is very high in tables 1 and 3, and Lohmann is very low. It is obviously impossible to get a perfect measurement, and some of the suggestions above for improvement are either ridiculous or impossible. Several people have suggested a qualification of 200 wickets. This might be helpful for modern players, but would exclude everyone before Grimmett including the S F Barnes, widely considered one of the greatest bowlers ever. It might be interesting to have separate tables for different eras - for instance pre 1914, between the wars, 1945 to perhaps 1980? and post 1980. This could be applied to most analyses.

  • Warwicks fan on March 15, 2008, 22:22 GMT

    Very interesting analysis, yet it does seem that something is not quite right. Table two seems the most 'reasonable' one of the three, based on knowledge of cricket and it's history, if only because Boje is very high in tables 1 and 3, and Lohmann is very low. It is obviously impossible to get a perfect measurement, and some of the suggestions above for improvement are either ridiculous or impossible. Several people have suggested a qualification of 200 wickets. This might be helpful for modern players, but would exclude everyone before Grimmett including the S F Barnes, widely considered one of the greatest bowlers ever. It might be interesting to have separate tables for different eras - for instance pre 1914, between the wars, 1945 to perhaps 1980? and post 1980. This could be applied to most analyses.

  • Seshan Patamadai on March 15, 2008, 22:18 GMT

    While I appreciate the innovation behind the analysis, at what point do we stop attempting to appreciate landscape or poetry through geometry and grammar? The wizardry of a Dennis Lillee or a Wasim Akram and the impact of finishing off a team's batting order cannot be captured in analysis. Cricket as a game goes beyond figures to the connoiseur. However, I do concede that in modern times, power counts for more than elegance, numbers more than beauty and results more than how the game was played. With that in context, maybe an additional variable can be considered - how many of the bowling performances reflected above have resulted in a win for the team?

  • Jonathan on March 15, 2008, 20:59 GMT

    I agree with Yogesh: you've assumed a batsman's wicket is worth the same throughout their career. Apart from fluctuations in batting average, which we can just about live with, a more serious problem is that the two factors (taking a batsman's wicket and their batting average) are not independent of each other.

    Take Glenn McGrath. McGrath dismissed Atherton 20 times, or whatever. Your model thus awards McGrath 20 times 38 points (or whatever Atherton's average was.) But what might Atherton's average have been if he'd never faced McGrath? A darn sight higher, I expect. Thus by taking Atherton's wicket with such regularity, McGrath was effectively penalising himself with respect to his BQI (I suspect he would have been able to live with this.)

    So more accurate than awarding Bowler A a Batsman X's average at the time the wicket was taken, would be awarding him the X's batting average excluding innings in which A bowled.

    Yes, it's cumbersome, but somebody here must have time...

  • eddy on March 15, 2008, 18:58 GMT

    McGrath took Laras wicket plenty of times but often after Lara had scored plenty of runs at a great average.....just check out statsguru. so What does this articule tell us?

  • skchai on March 15, 2008, 18:17 GMT

    I do have one quibble with the statistic - wouldn't BQI / average be a better measure than BQI - average? The latter measure would seem to favor bowlers of the modern era, where far more runs tend to be scored than in the 19th and early 20th centuries, thus inflating the difference. Using the former measure does not change the list wholesale, but it does lead to some interesting reordering, with Sidney Barnes now being at the top of the table. The top 25 would be:

    1.496 Barnes S.F 1.456 Lohmann G.A 1.437 Davidson A.K 1.434 Marshall M.D 1.422 Ambrose C.E.L 1.406 McGrath G.D 1.378 Laker J.C 1.376 O'Reilly W.J 1.372 Turner C.T.B 1.339 Croft C.E.H 1.334 Adcock N.A.T 1.333 Miller K.R 1.317 Trueman F.S 1.315 Donald A.A 1.314 Blythe C 1.304 Hadlee R.J 1.290 Imran Khan 1.285 Lillee D.K 1.281 Pollock S.M 1.279 Bishop I.R 1.276 Muralitharan M 1.270 Garner J 1.268 Bedser A.V 1.263 Lindwall R.R 1.260 Verity H

  • skchai on March 15, 2008, 18:06 GMT

    While most of criticisms here of the BQI-adjusted average seem pretty spurious and based upon basic misunderstanding of this statistic and what it is supposed to represent. Aren't tail-end wickets important? Of course they are, but they also tend to flatter a bowler's average, and this is what the the BQI-adjusted average is meant to correct. Shouldn't you take into consideration the nature of the wicket, the nature of the game situation, etc.? Not only does data not exist for these characteristic; it would be hard to define any practical quantitative indicator that is feasible to measure and where there is reasonable consensus regarding validity. This is not an attempt to create a perfect measure of bowler effectiveness (were such a thing possible); it is simply an attempt to use readily available data to generate a more valid indicator of bowling average, one that takes into account the quality of the opposing batsman.

  • Amirali on March 15, 2008, 17:18 GMT

    This measure suffers from penalizing bowlers who clean up the tail often. One has to recall, cleaning up batsem averaging 5 and 10 runs is NOT worth just 5 to 10 runs to the side. If you let taileneders block out for hours for a few runs, a Gilchrist or a Laxman can score heavily at the other end. In that sense, lower-order wickets also stop scoring potential of batsman at other end. @Ranjan. I don't have the stats for it, but if you take the average runs scored for last 4 wicket partnerships, it would most probably exceed the combined averages of last 4 batsmen by at least 30-50%, because of the effect of having an established batsman at one end. Hence that means lower order wickets might be "undervalued" by this measure. Of course, that's just speculation on my part; I don't have the hard facts.

  • D C on March 15, 2008, 16:56 GMT

    Very good attempt at capturing bowling - though as someone mentioned - pitches and teams you are playing for matter a lot. McGrath supported by a battery of fast bowlers and a class spinner playing most of his matches on pacy pitches should have a different quality than say Mashrafe Mortaza or Heath Streak who hardly have big totals to defend and bowler friendly pitches to bowl on.

  • Joaquim Da Silva on March 15, 2008, 12:20 GMT

    The more we try and analyse the statistics the more confusing it gets. What about wickets that were extremely lucky?

    What about one bowler bowling an extremely devastating spell and his partner picking up the wickets? What about the bowler getting 3 wickets because of brilliant catches or umpiring errors?

    Stats are great but you can prove just about anything with them.

  • Yogesh on March 15, 2008, 12:14 GMT

    I have three alternate ways which are better:

    1) A minor but significant variant of your method. Instead of taking batsmen's average, why not take batsmen's average upto the match ? That is a fairer indication. Simple example, bowlers who took Ponting in late 1990's are same as someone who takes his wicket today. And someone who takes Tendulkar's wicket today benefits equally as someone who took it in late 1990's. While any reasonable cricket follower can distinguish which is more valuable ?

    2) I would love to see the following. Consider the number of runs a batsman has scored fromn the time bowler has started bowling before he is dismissed by the bowler. Subtract this from his average at that point of time.. Sum it up for all the batsmen dismissed. If it is highly positive then, the bowler has dismissed the batsmen much early. But this is a very rough idea. I am sure there are going to be anamolies. But this is yet another measure.

  • Ranjan Kant on March 15, 2008, 11:36 GMT

    Why can't one simply look at the following to get the quality ?

    1) # of top order wickets / match 2) Average of the bowler against top order only

    This should give a fair assessment of the Bowler's class in my opinion, no ?

  • raj on March 15, 2008, 11:23 GMT

    Nice math but you obviously know this isn't even close to a decent system. 1. Tail Enders' wickets not important? Try telling to India, who many a time in the 9's used to knock off 7 wickets and then concede another 200 runs 2. Jason Gillespie. No explanations required. Okay, i guess it is. He may not have a big average but he was difficult to dislodge and was a valuable wicket. 3. Dismissed batsman's average? So, is it average at the time of dismissal? Average at the end of the career? So, this is an exercise in retrospect? If you do the same exercise next year, it will change the numbers. See what I mean? Sometimes, you have got to believe your plain eye which would tell you that Akram was better than McGrath and Warne is better than Kumble.

  • Zeeshan on March 15, 2008, 11:10 GMT

    "a bowler like wasim relied on pure skill. the joy one derived from watching him bowl was tremendous"

    Interesting point... Personally I feel that a bowlers craft lies in beating the batsman and getting those bails off .. what a treat to eye is when those two stumps are uprooted by swinging yorker...

    so could you please come up with a BQI which takes account or gives more credit to the bowlers who have hit stumps more ..ant average joe can take Tendulakars or Dravids wicket when they mistimed a slog .. however how many bowlers can creep into Dravids defense ??

  • srin on March 15, 2008, 10:58 GMT

    Determining quality of the batsman by career batting average is suspect. I think it is more relevant to use his average at the time the wicket was taken, as a fairer measure of the "value".

  • Karthick on March 15, 2008, 10:41 GMT

    Brilliant post Ananth. I had posted during your last analysis that this list could be useful and it indeed is. I was always under the impression that Gillespie was the best against top order batsmen, but your analysis has dispelled my notion. But I am indeed glad to know that our own Zaheer Khan is the best in getting out the better batsmen (this era's beefed up batting averages aided). Your 2nd list which takes the bowling average into account is a good move to put the boje's and dillon's in their place. But I have my doubts here. What is a bowling average? Isnt the bowling average calculated by consolidating all his wickets irrespective of top order or lower order. The classic example is Murali who has a BQI of 27 but helped by a monstrous bowling average - no doubt improved by tail-end and BAN/ZIM wickets - manages to come in the 23rd position. I agree with Karthik Agaram on computed bowling averages. I know Ananth would say this wasnt meant to be that complicated, but just curious.

  • Phil on March 15, 2008, 9:14 GMT

    It disadvantages the players in the great teams of their generations. The current Australian (OK, recent Australian) team haven't been bowling againist their own batsmen, likewise the opposite analysis for batsmen. Of course this is cricket, the beauty of which is that we could ananlyse it forever (and indeed probably will)

  • Xill-e-Ilahi on March 15, 2008, 8:15 GMT

    I like the suggestion that bowlers with less than 200 test wickets should be left off the calculation - at least for bowlers in the post-WW2 era, considering that they probably played a lot more test matches than their predecessors.

    Another basis might be percentage of wickets taken as opposed to the remainder of the team (and the average at which they were taken) - and build this BQI thing into that. This would show the true value of the bowler to his team.

    Also, is there some way to account for the significance of a bowler taking 5 for 47 while defending 160 as opposed to one taking 5 for 47 while defending 360. The pressure of a small target should ideally be accounted for. That might put people like Warne and Younis right at the top where they should be.

  • sridhar on March 15, 2008, 7:54 GMT

    superb. extremely innovative stuff. keep it coming. table 3 was an eye opener! Zaheer khan is indeed ver valuable

  • Anonymous on March 15, 2008, 7:25 GMT

    Since we are talking about quality of wkts here, does this analysis assume that a batsman's quality remains constant thru out his career. For e.g. if a bowler has taken a batsman 10 times, did you add his avg. at each of those 10 occassions or was it just 10*current avg?

  • Philip John Joseph on March 15, 2008, 7:17 GMT

    Good analysis Ananth, confirmed by the fact that Muralitharan is NOT in the top ten. This proves that Muralitharan is overrated. To those who insist on talking about Wasim Akram, all I can say is PLEASE. Waqar Younis Maitla SCHOOLS Wasim Akram any day of the week, and HOW! Just because Waqar isn't a media-talking superstar doesn't mean he isn't the best. Since everyone is into making suggestions, I think I'll provide a suggestion too. Maybe a list based on which bowlers exceeded the team bowling average for a given match, versus those who underperformed the team bowling average. As in, if a team is bowled out for 400 runs, then the bowling average to decide the relative performance of the bowlers would be set at 40, or 40 runs per wicket. That said, that analysis has probably already been done on this site, seeing as I haven't been surfing this section for that long.

  • Anonymous on March 15, 2008, 7:12 GMT

    This has to be a joke. Would be interested in knowing Boycott's comment on his favorite bowlers being in the list.

  • R.Narayan on March 15, 2008, 6:18 GMT

    It is not surprising that the list of bowlers with a high percentage of tailenders include some big names. Most of them are spinners, who would be expected to be using the old ball when the tail was in,

  • Peter Della Penna on March 15, 2008, 5:38 GMT

    This is a very interesting and useful analysis. It proves what I have thought over the past year, that Dale Steyn is incredibly overrated and Brett Lee is a far better bowler than he is. Any halfwit can rip through New Zealand, West Indies and Bangladesh. I want to see what Steyn does against India, England and Australia over the next 12 months. Then we will see how good he actually is.

  • Sam Singh on March 15, 2008, 5:36 GMT

    I think you'll be better off calculating how bowlers fared against top order batsmen. Create two different tables - a. Bowlers to top 5 batsmen with runs conceeded, wickets taken, strike rate to get the bowling average calculated for just the top 5 b. repeat this for bottom 6 batsmen while this might be really difficult to calculate for all time, for recent years it'll give you a far more useful perspective Sam Singh

  • Ryan on March 15, 2008, 5:22 GMT

    This analysis is great. To tackle your own suggestion, how about adding in the number of runs the batsman has scored below/above his average at the time of dismissal? Adjusting this for his actual average, you could get a measure of the bowler's ability to get batsmen out early and cheaply. I suspect this will favour quick bowlers who would get to bowl at top order batsmen (with high averages) under favourable first-day morning conditions, but it could be interesting.

  • Mohan on March 15, 2008, 4:48 GMT

    I thought this article will be more thorough after I got here when i clicked thru cricinfo. But I kinda agree with Clarke here..

  • Ali Akbar on March 15, 2008, 2:51 GMT

    Superb analysis. But i think only bowlers with more than 200 wickets should be taken into consideration. A

  • Srinivas on March 15, 2008, 1:13 GMT

    I think a modified Table 2 after using a multiplying factor depending on the career number of wickets will represent the best bowlers of all time.

  • Mani on March 15, 2008, 0:42 GMT

    fantastic idea! just a couple of suggestions:

    1/ does normalizing the sum of batting averages by the team batting average help - so dismissing dravid/tendulkar/sehwag is as valuable as dismissing smith/gibbs/kallis which is the same as cook/vaughan/bell (say) in a match context

    2/ normalize the batting average of wickets captured by a bowler to total team score in each innings - might be too computationally intensive

    thanks

  • abbas on March 14, 2008, 23:51 GMT

    only 1 criteria should be used to evaluate the greatest bowlers and batsmen..how many matches they have won...dont confuse ureself with stats and # crunching look at the bottom line..look at the man of the match awards..be smart :)

  • Jay on March 14, 2008, 23:43 GMT

    Great analysis. Can anyone answer how India has managed over the years with hardly any Indian bowlers making it to the list?

  • Jamil on March 14, 2008, 19:09 GMT

    I can appreciate the extensive use of cricinfo stats to calculate BQI but I don't think it is a good indicator because we cannot consider the wickets the matches were played on, the conditions (of course you will have english and australian bowlers in the list), the medium of the bowlers (spinners sometimes don't start until 30-40th over), and effect of own team's batsmen (if own team scored low then there is no pressure on other side to play strokes etc). However, it is just one view of looking at the stats with no apparent verdict. Just wondering though, if Wasim Akram didn't take the top order wickets for 15years, then who did for Pakistan? Tauseef Ahmed??

  • Geoff Bethell on March 14, 2008, 18:44 GMT

    I totally agree with Abeer. What would be interesting is an analysis of tail performance against teams containing bowlers in your list. With a good leggie in the team nobody after #6 or #7 ever gets runs - they are simply too much for these players. A seam-only team will struggle to dismiss the tail in good conditions, when tired, and with an older ball.

  • satish ganesan on March 14, 2008, 18:32 GMT

    now we should have an analysis into why south indians love numbers so much.

  • Raj Mondhra on March 14, 2008, 17:49 GMT

    I like your analysis of difference between BQI and Bowl Average. I would've liked it further if you excluded bowler uner 200 wickets. To me, that list really represents the best bowlers in Test cricket.

  • eddy on March 14, 2008, 17:47 GMT

    M Marshall, the greatest seam/fast bowler ever; 2nd only to Warne as the greatest bowler ever.

  • Abeer Agrawal on March 14, 2008, 17:27 GMT

    Not only does your analysis of the bowlers with the least lower-order wickets indicate a proclivity to a dismiss the top order batsmen, it also indicates an inability to dismiss the tailenders, with Zak and Pathan being typical cases in the point.

  • Kartik Agaram on March 14, 2008, 17:16 GMT

    How about a simple threshold? Compute bowling averages taking into account only wickets of batsmen above some average, say 40.

  • nangaswaami jee on March 14, 2008, 17:14 GMT

    well nice analysis but i wud say swaami jee that WASIM AKRAM IS THE GREATEST BOWLER OF ALL TIMES and SECOND COMES ABDUL QADIR... sweet dreams......

  • siddharth on March 14, 2008, 16:56 GMT

    brilliant analysis...its shocking to see wasim akram at 68 in the list..but you should also consider the different bowling conditions in which the bowlers took their wickets.. mcgrath bowled mostly on the bouncy tracks down under..a bowler like wasim relied on pure skill. the joy one derived from watching him bowl was tremendous. So even though you did a terrific job with the statistical analysis, it doesn't show th ereal picture..

  • Arpan on March 14, 2008, 16:26 GMT

    Interesting comparison, however your BQI measure penalizes bowlers who take lower order wickets, no matter what other wickets they have taken. If bowlers A and B take 2 top order wickets each, and bowler B takes 1 lower order wicket, bowler B's BQI suffers (although B is as good as A, if not better).

  • B M on March 14, 2008, 16:10 GMT

    Dude - where do you come up with such analysis? All very wonderful stuff. Sheds new light and offers un-heralded bowlers some sunshine :-). Keep it coming.

  • arnab on March 14, 2008, 16:01 GMT

    as I was reading the first para, I thought the total (or average) of the batting averages would be the denominator. The lower the number the better the bowler.

    Maybe you would need to normalise somehow by doing (10 - wickets per test) / average of the averages

    Otherwise

  • Aaron on March 14, 2008, 15:49 GMT

    A valuable analysis here. I was a critic of one of your earlier articles but this one is excellent. Great job.

  • Sumit on March 14, 2008, 15:40 GMT

    Great Post: BQI - bowling average is a very scientific method of capturing bowler effectiveness. The "real" effectiveness though is (BQI - bowling. average)*Wickets/inning. (Essentially saying a 60/3 performance is better than a 40/2 performance). However, unlike batting where a batsman's innings is independent of other players in the team, the "real" effectiveness will depend on how many overs a bowler bowls which depends on what are the other bowlers like.

  • Roelof on March 14, 2008, 15:38 GMT

    I would say that a useful attempt would be to combine your list of batting positions dismissals with the bowling average.

    Determine a coefficient x by giving a bowler say (arguably) 3 points for a top order batsman and 1 for a tail-ender, then dividing the total points by number of dismissals.

    Then (bowling average)/x should provide an interesting stat similar to your Average-BQI, but removing the problems of variable batting averages. Of course how much more worth a top order batsman is compared to a tail ender in the calculation of x is up for debate.

    It could also be interesting to consider strike rate, and see which bowlers got the best batsman out fastest!

    Thanks for the interesting column! Roelof

  • Elayaraja Muthuswamy on March 14, 2008, 15:22 GMT

    Your Table-1 looks great. It has got all time great bowlers in the list. I think it is a nice way to figure out the quality of wickets.

  • Elayaraja Muthuswamy on March 14, 2008, 15:21 GMT

    Your Table-1 looks great. It has got all time great bowlers in the list.

  • kk125 on March 14, 2008, 15:02 GMT

    This indeed shows that there is no perfect statistical measure to decide how capable a bowler is.

  • kk125 on March 14, 2008, 15:01 GMT

    This indeed shows that there is no perfect statistical measure to decide how capable a bowler is.

  • kk125 on March 14, 2008, 15:00 GMT

    This indeed shows that there is no perfect statistical measure to decide how capable a bowler is.

  • Michael Clark on March 14, 2008, 14:25 GMT

    The problem with this analysis is that you're punishing bowlers for taking tail-end wickets. A bowler who takes all 10 wickets in an innings will have a worse BQI than one who takes just the top three!

  • No featured comments at the moment.

  • Michael Clark on March 14, 2008, 14:25 GMT

    The problem with this analysis is that you're punishing bowlers for taking tail-end wickets. A bowler who takes all 10 wickets in an innings will have a worse BQI than one who takes just the top three!

  • kk125 on March 14, 2008, 15:00 GMT

    This indeed shows that there is no perfect statistical measure to decide how capable a bowler is.

  • kk125 on March 14, 2008, 15:01 GMT

    This indeed shows that there is no perfect statistical measure to decide how capable a bowler is.

  • kk125 on March 14, 2008, 15:02 GMT

    This indeed shows that there is no perfect statistical measure to decide how capable a bowler is.

  • Elayaraja Muthuswamy on March 14, 2008, 15:21 GMT

    Your Table-1 looks great. It has got all time great bowlers in the list.

  • Elayaraja Muthuswamy on March 14, 2008, 15:22 GMT

    Your Table-1 looks great. It has got all time great bowlers in the list. I think it is a nice way to figure out the quality of wickets.

  • Roelof on March 14, 2008, 15:38 GMT

    I would say that a useful attempt would be to combine your list of batting positions dismissals with the bowling average.

    Determine a coefficient x by giving a bowler say (arguably) 3 points for a top order batsman and 1 for a tail-ender, then dividing the total points by number of dismissals.

    Then (bowling average)/x should provide an interesting stat similar to your Average-BQI, but removing the problems of variable batting averages. Of course how much more worth a top order batsman is compared to a tail ender in the calculation of x is up for debate.

    It could also be interesting to consider strike rate, and see which bowlers got the best batsman out fastest!

    Thanks for the interesting column! Roelof

  • Sumit on March 14, 2008, 15:40 GMT

    Great Post: BQI - bowling average is a very scientific method of capturing bowler effectiveness. The "real" effectiveness though is (BQI - bowling. average)*Wickets/inning. (Essentially saying a 60/3 performance is better than a 40/2 performance). However, unlike batting where a batsman's innings is independent of other players in the team, the "real" effectiveness will depend on how many overs a bowler bowls which depends on what are the other bowlers like.

  • Aaron on March 14, 2008, 15:49 GMT

    A valuable analysis here. I was a critic of one of your earlier articles but this one is excellent. Great job.

  • arnab on March 14, 2008, 16:01 GMT

    as I was reading the first para, I thought the total (or average) of the batting averages would be the denominator. The lower the number the better the bowler.

    Maybe you would need to normalise somehow by doing (10 - wickets per test) / average of the averages

    Otherwise