How consistent or otherwise was Tony Greig? The answer is certainly memorable

How consistent was Tony Greig as a batsman?

© Getty Images
84

Stats feature

Who are the most consistent players in Test history?

Bradman? Sachin? Warne? Hadlee? Go on, find out. You'll be surprised

Anantha Narayanan |

Measuring the consistency of Test players has always been a tough proposition. The statistical methods normally followed give us results that are seemingly fine but prove unsatisfactory in the end. Let me give a simple example.

Batsman A has scores of 50, 50, 50 and 50 in four innings. There is no doubt that he is a statistician's delight and the epitome of consistency. Batsman B has scores of 100, 0, 0, 100 in four innings. He is a thoroughly inconsistent player and no one knows which B will turn up on the day. However, it is eminently possible that his two hundreds have contributed to wins for his team. The real problem is that if we post a high statistical variance on Batsman B and brand him inconsistent, it goes against cricketing logic. In a pure cricketing sense, all scores above the mean value are always welcome and should not have a negative impact on the consistency values.

The other problem is that in almost all score distributions, the mean (runs per innings) is around 15 to 25% of the high score. The distribution is always lopsided. If a batsman plays 100 innings and scores 5000 runs, his mean is 50. He is likely to have 30 scores between 50 and 200 and 70 scores below 50. We must realise that 0 and 200 are both part of the game. However, the 200, while it may be four times the mean, should not be penalised. The variance component for the 0 is 50 and for the 200 it is 150.

I have done a lot of work using an innings or a Test as the unit of analysis. While the results have been acceptable, I have always felt that this was a riddle waiting to be solved. How does one arrive at a method to measure consistency keeping in mind the cricketing context? The thinking has to be out of the box - that much is certain. Finally, I think I have the answer.

I came across this idea when I did work on performance streaks for players. I realised that a streak of X innings/spells/Tests gave me a well-defined block of player activity to work with. After a lot of trials and evaluations, I have come up with the following methodology.

The cornerstone of this analysis is that consistency should only be measured using data of the player being considered. A batting average of 40 might be way below the par performance of, say, Herbert Sutcliffe, while the same average might be way above the par performance of Ian Botham. Hence, we will measure the consistency of each player against the overall career standards he has himself set.

It should be noted that the statistics in this piece are as of Test No. 2264, last month's magnificent Colombo Test between Sri Lanka and Zimbabwe, which swung between the two teams so many times that the participants and followers never knew where they were at any given point of time.

Muttiah Muralitharan: No. 1 on the list of Test wicket-takers, but No. 25 by consistency

Muttiah Muralitharan: No. 1 on the list of Test wicket-takers, but No. 25 by consistency © AFP

Methodology

The basic unit of measure will be blocks of ten innings or spells.
By "spell", I mean a bowler's total number of overs in an innings, not a continuous bowling stint, as understood normally. A block of ten is long enough to represent between five and eight Tests over about three to six months. This allows a player the opportunity to make up for failures, bad decisions, bad luck, unplayable balls and so on in other innings/spells within the block. The blocks traverse home/away Tests, series, opponents, and even years. Therefore, there is no inbuilt bias. Why ten, why not eight or 12? No particular reason: ten represents an easily workable unit. I have made trial runs with eight or 12 innings/spells as the basis. There are only minor variations in the results. The majority of the players stay in their respective areas.

An innings is established when the batsman faces a ball and a spell is recorded when the bowler bowls a ball.
This may seem unfair if a batsman scores 5 not out or a bowler bowls three wicketless overs. However, these things even out across ten innings/spells. The cut-offs are 3000 runs and 100 wickets respectively for batsmen and bowlers. In all, 184 batsmen and 177 bowlers qualify. The minimum number of blocks for batsmen is eight and for bowlers it is four.

Do I measure the player performance against an imaginary norm?
No, this is a method to measure a player's consistency and all comparisons have to be within the player's ambit. Performance analyses should be across players but consistency analyses should be within the career of the concerned player.

What is the base performance we work with?
We will take the career figures of the player and determine the average runs scored in ten innings or wickets taken in ten spells. For batsmen, the question of batting average does not arise. Instead, the actual runs per innings will be used.

How would above-average performances be handled?
If the mean for a batsman is 400 runs per block and he scores 700 runs in a particular block, what do I do? This is the key decision. In keeping with cricketing logic, I will not penalise the batsman but cap this performance at 100%. This is a key decision in the analysis. Do I detect a few academic purists shaking their heads? But I can also see many more cricket followers nodding.

How would below-average performances be handled?
If the mean for a bowler is 22 wickets per block and he takes 15 wickets in a particular block, what do I do? I will leave this performance at 68.2% (15/22). After all, these are the performances we are looking for to measure consistency.

How do we handle the last few innings/spells of a career?
Don Bradman played 80 innings and Richard Hadlee bowled 150 spells. No problem with these players. However, what do I do about Clarrie Grimmett, who bowled 67 spells, or Tony Greig, who played 93 innings? I tried leaving the seven and three as part blocks and adjusted the numbers accordingly. But it often backfired. Greig's last three innings were 0, 43 and 0. Players should not be penalised for ending their careers at certain points. Lean blocks would always lead to problems like this. So I decided that if the last block had five or more innings/spells, I would take it as an independent block. If the last block had four or fewer innings/spells, I would merge it with the penultimate block. In both these cases, tweaking is done to take care of the number of innings/spells in the final block.

Finally, what is the index of player consistency?
It is the average of the block performance percentage values across a career. This average is called the Consistency Index. The higher the average, the more consistent the player. Let me define player consistency here as "the ability of players to keep their sub-par performances to a minimum and close to their mean values".

Let us see how the process works using the data for a very consistent batsman and a very consistent bowler.

Summary of Tony Greig's career
Greig scored 3599 runs in 93 innings. His Block Mean thus works out to 387.0 runs (3599/9.3). Since the last block has three innings, the penultimate block is deemed to have 13 innings. Thus, he has nine blocks. The mean for the last block is 503, which is derived by extrapolation of the mean (387*13/10). The block analysis is explained below.

Block Innings Runs Mean % of Mean % - Adjusted Consistency Index
1 10 356 387.0 92.0% 92.0%  
2 10 463 387.0 119.6% 100.0%  
3 10 393 387.0 101.6% 100.0%  
4 10 486 387.0 125.6% 100.0%  
5 10 385 387.0 99.5% 99.5%  
6 10 370 387.0 95.6% 95.6%  
7 10 302 387.0 78.0% 78.0%  
8 10 440 387.0 113.7% 100.0%  
9 13 404 503.1 80.3% 80.3%

 

          845.4% 93.9%

Greig's troughs were not low: the three 90-plus values helped a lot. Also, the remaining two values are around 80%. His final average of 93.9% indicates a very high level of consistency.

Summary of Bill O'Reilly's career
O'Reilly took 144 wickets in 48 spells. His Block Mean works out to 30.0 wickets (144/4.8). I have selected O'Reilly so that the other method of last-block handling can be explained. Since his last block has eight spells, it is left as it is, and he has a total of five blocks. The mean for the last block is 24.0, which is derived by extrapolation of the mean (30*8/10). The block analysis is explained below.

Block Spells Wickets Mean % of Mean % - Adjusted Consistency Index
1 10 31 30.0 103.3% 100.0%  
2 10 29 30.0 96.7% 96.7%  
3 10 25 30.0 83.3% 83.3%  
4 10 29 30.0 96.7% 96.7%  
5 8 30 24.0 125.0% 100.0%  
          476.7% 95.3%

O'Reilly had two of his five blocks at 100% and two others at 96.7%. These led to a very high Consistency Index value of 95.3%.

Most consistent Test batsmen in history

Batsman Team Runs Innings Blocks Block Mean Consistency Index
Tony Greig Eng 3599 93 9 387.0 93.9%
Darren Bravo WI 3400 89 9 382.0 93.3%
Herbert Sutcliffe Eng 4555 84 8 542.3 92.5%
Ian Redpath Aus 4737 120 12 394.7 91.8%
Arjuna Ranatunga SL 5105 155 16 329.4 91.7%
Roy Fredericks WI 4334 109 11 397.6 91.2%
Alan Knott Eng 4389 149 15 294.6 91.2%
Kevin Pietersen Eng 8181 181 18 452.0 91.1%
Gordon Greenidge WI 7558 185 19 408.5 91.0%
Saeed Anwar Pak 4052 91 9 445.3 91.0%
Linsay Hassett Aus 3073 69 7 445.4 90.9%
Joe Root Eng 4875 102 10 477.9 90.5%
Ijaz Ahmed Pak 3315 92 9 360.3 90.4%
Allan Border Aus 11174 265 27 421.7 90.2%
Greg Chappell Aus 7110 151 15 470.9 90.0%

Greig is the most consistent Test batsman ever. The illustration above indicates how he achieved his amazing Consistency Index value of 93.5%. His Block Mean value is 387.0, and his minimum block compilation is 302. This is within around 25% of the mean, and that has helped produce such a high Consistency Index value.

Darren Bravo has an almost identical Block Mean value as Greig, and his block runs of 372, 403, 564, 310, 362, 377, 292, 349 and 371 (nine innings) has secured an almost similar Consistency Index value of 93.3%. However, Bravo's is a currently running career, and as and when he returns, the figures will change.

Herbert Sutcliffe had an average of 64 at the end of his first Test. He played 53 more and never once did it go below 60. So his position at No. 3 on this list is not a surprise. His eight blocks are 813, 535, 556, 526, 530, 665, 459 and 471 (14 innings), for a Consistency Index value of 92.5%. His Block Mean value of 542.3 is the highest in the top 15.

The elegant Australian batsman of the '60s and '70s, Ian Redpath, is next, with a Consistency Index value of 91.8%, followed closely by Arjuna Ranatunga, at 91.7%. The attacking West Indian opener Roy Fredericks is a surprise placement, bracketed with another surprise, Alan Knott. The latter had an expectedly lower Block Mean value of 294 runs but achieved this figure across 15 blocks to Fredericks' 11.

Who would have thought that the mercurial batsmanship of Kevin Pietersen would merit eighth place, at 91.1% - and across 18 blocks at that? In one block, Pietersen scored 303 runs for a 67.0% value. Otherwise, he was always above 75%. Gordon Greenidge is in ninth position, with 91.0%, achieved over 19 blocks. His penultimate block was a disaster, at 127 runs and 31.1%. But for this, he would have finished higher. Saeed Anwar completes the top ten, with a Consistency Index value of 91.0%.

We all suspected that Joe Root is a consistent batsman, and that is reinforced by him being in 12th place on this list, with 90.5%.

A graphical depiction of the most consistent batsmen looks like this.

© Anantha Narayanan

Readers can note the almost flat nature of the curves. Greig and Sutcliffe had below-average last blocks, while performing at or above their par values for the better part of their careers. Darren Bravo's career has come to a stuttering halt; however, there is a clear upward trend recently. After a few dips, Redpath's career graph is almost totally flat. In the second half of his career, he has performed at or above his par.

Least consistent Test batsmen in history

Batsman Team Runs Innings Blocks Block Mean Consistency Index
Dennis Amiss Eng 3612 88 9 410.5 77.2%
Victor Trumper Aus 3163 89 9 355.4 78.2%
Gautam Gambhir Ind 4154 104 10 399.4 78.6%
Sanath Jayasuriya SL 6973 188 19 370.9 79.6%
Navjot Sidhu Ind 3202 78 8 410.5 80.0%

Let us now have a look at the five least consistent Test batsmen ever. Dennis Amiss had a poor start to his career, scoring 151 runs in his first ten innings. He ended almost as poorly, with 175 runs in his last eight. In between, he had periods of deluge and drought. This is reflected in his Consistency Index value of 77.2%. He had block accumulations of 151, 189, 603, 472, 774, 425, 241, 582, and 175 (eight innings), as against the Block Mean of 410.

Victor Trumper had a turbulent career: 282, 235, 347, 595, 197, 200, 377, 774, and 156 (nine innings), which led to a Consistency Index value of 78.2%. Gautam Gambhir was similarly inconsistent; he followed blocks of 875 and 794 with 154.

Others among the 20 least consistent batsmen in Test history include Denis Compton, Virender Sehwag, Jimmy Adams, Dilip Vengsarkar, Marlon Samuels, Keith Fletcher, Aravinda de Silva and Herschelle Gibbs.

© Anantha Narayanan

There is no better visual depiction for inconsistent batsmen than these graphs. The crests and troughs appear with monotonous regularity, especially for Sanath Jayasuriya.

Most consistent Test bowlers in history

Bowler Team Wickets Spells Blocks Block Mean Consistency Index
Chris Cairns NZ 218 104 10 21.0 96.3%
Geoff Arnold Eng 115 61 6 18.9 96.2%
Yasir Shah Pak 149 50 5 29.4 95.6%
Bill O'Reilly Aus 144 48 5 30.0 95.3%
John Snow Eng 202 93 9 21.7 94.9%
Ryan Harris Aus 113 52 5 21.7 94.7%
Neil Wagner NZ 130 61 6 21.3 93.8%
Neil Adcock SA 104 46 5 22.6 93.6%
Lasith Malinga SL 101 59 6 17.1 93.3%
Darren Gough Eng 229 95 10 24.1 93.1%
R Ashwin Ind 275 92 9 29.9 92.9%
Andy Caddick Eng 234 105 11 22.3 92.9%
Vanburn Holder WI 109 72 7 15.1 92.9%
Mohammad Asif Pak 106 44 4 24.1 92.8%
Mohammad Rafique Ban 100 48 5 20.8 92.8%

Chris Cairns is the most consistent Test bowler ever. Over the past few years, he has appeared more in courtrooms than on cricket grounds or in broadcasting boxes. But he was a terrific allrounder, and while his bowling numbers are impressive - 218 wickets at 29.40 - it is his consistency that is astonishing.

Let us look at his block performances: 21, 19, 19, 23, 21, 24, 21, 21, 25 and 24 (14 spells). His Block Mean is 21 wickets. His three below-par values are 90.6%, 90.6% and 81.8%. That is some consistency, and has resulted in a Consistency Index value of 96.3%.

Geoff Arnold was almost as consistent as Chris Cairns and falls short by only 0.1%. His Block Mean value, however, is lower at 18.9. None of Arnold's blocks is lower than 90%.

Who would have guessed that the third bowler on this list would be Yasir Shah? Yasir moved into the top ten after his sterling performances in Bridgetown and Roseau earlier this year. His Block Mean is a high 29.8, and in his five blocks he has taken 30, 35, 25, 28 and 31 wickets. Never has he gone below 84%. It is true that having bowled only 50 spells in his career has helped, but it is remarkable how he makes up for barren Tests with resounding performances in others.

O'Reilly is in fourth place, and John Snow completes the top five. Three modern bowlers - Ryan Harris, Neil Wagner and Lasith Malinga - find spots in the top ten, with Consistency Index values exceeding 93%.

It is not a surprise that the Consistency Index values for bowlers are slightly higher than those for batsmen. The main reason is that the percentage of below-par blocks for bowlers is 51.4%, whereas for batsmen it is 53.4%. This is the order of difference. Two per cent more bowler performances are capped at 100% levels.

© Anantha Narayanan

Look at the graph for Cairns. A virtual straight line, barring a couple of blips.

Least consistent Test bowlers in history

Bowler Team Wickets Spells Blocks Block Mean Consistency Index
Wilfred Rhodes Eng 127 90 9 14.1 70.5%
Carl Hooper WI 114 145 15 7.9 73.9%
Trevor Bailey Eng 132 95 10 13.9 77.4%
Mushtaq Ahmed Pak 185 89 9 20.8 77.8%
Monty Noble Aus 121 71 7 17.0 78.1%
Alf Valentine WI 139 63 6 22.1 79.7%

Wilfred Rhodes is the least consistent bowler in the history of Test cricket. Since he took only 127 wickets in 90 spells, his Block Mean value is a low 14.7. But just look at his wildly varying block wickets: 27, 32, 17, 3, 16, 1, 8, 7 and 16. His Consistency Index is a miserable 70.5%. There are reasons for this wild variation. After his first 20 Tests, Rhodes ceased to be a front-line bowler but kept on bowling.

It is not surprising that Carl Hooper, the West Indian allrounder who was more batsman than bowler, finished with a bowling Consistency Index of 73.9%. Almost a similar situation exists with Trevor Bailey, who has a Consistency Index value of 77.4%. Mushtaq Ahmed, Monty Noble and Alf Valentine complete the bottom six, all with Consistency Index values below 80%.

There are four specialist bowlers in the bottom ten: Mushtaq, Alf Valentine, Rodney Hogg and Jerome Taylor. Valentine, who started his career with hauls of 39, 28 and 29 wickets and finished with a horror sequence of 13, 14 and 16 (13 spells), has a Consistency Index value of 79.7%. Hogg started with a bang, taking 40 wickets in his first block, and then tailed off to 18, 16, 11, 15, 16 and 7 (six spells). Taylor followed a block of nine wickets with one of 24, a block of 25 with one of seven, and a block of 22 with one of eight. This explains his inconsistency.

© Anantha Narayanan

There are nice variations even among these four inconsistent bowlers. No single obvious pattern can be observed.

Top players' standings
Most cricketers with long careers are in the middle two quartiles. Let us look at how some well-known players fare in the consistency stakes.

Don Bradman is in 52nd position, with a Consistency Index value of 88.1%. This is in the second quartile.

Sachin Tendulkar is in 100th position, with a Consistency Index value of 85.9%. This is in the lower half.

Ricky Ponting is in 146th position, with a Consistency Index value of 83.5%. This is a fairly low number and puts him in the fourth quartile.

Brian Lara is in 53rd position, with a Consistency Index value of 88.1%, just behind Bradman. This is in the second quartile.

Sunil Gavaskar is in 123rd position, with a Consistency Index value of 85.1%. He is in the third quartile.

Allan Border is in 14th position, with a Consistency Index value of 90.2%. This is very close to the top ten. Border is the best placed among the big scorers.

Muttiah Muralitharan is in 25th position, with a Consistency Index value of 92.1%. This is in the first quartile. Muralitharan is the most consistent among the top wicket-takers.

Shane Warne is in 81st position, with a Consistency Index value of 89.1%. This is in the second quartile. This is somewhat unexpected, considering that Warne performed very well around the world.

Anil Kumble is in 73rd position, with a Consistency Index value of 89.4%. He is comfortably in the second quartile.

Glenn McGrath is in 35th position, with a Consistency Index value of 91.7%. This is in the first quartile. This level of consistency was probably expected from McGrath.

Kapil Dev is in 165th position, with a Consistency Index value of 82.8%. Kapil is in the bottom 15 of the table.

Richard Hadlee is in 45th position, with a Consistency Index value of 90.8%, which places him in the top quartile.

Standard statistical measures
In order to validate the results, I also worked out the Standard Deviation (SD) and Coefficient of Variation (CoV, calculated as SD/Mean) values. To present the CoV tables in this article would take too much space and would be superfluous. However, I can affirm that there is a very good correlation between the Consistency Index and the CoV values. Just an illustration: the lowest CoV among Test batsman, 0.165, belongs to Tony Greig.

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

 

RELATED ARTICLES

 

LOGIN TO POST YOUR COMMENTS

  • POSTED BY arun on | August 18, 2017, 12:14 GMT

    thanks for the really nice and thoughtful conversation! federer is indeed a role model on more ways than one. It is probably safe to assume that he would have retired if he had not done well, or if he thought he could not...still, retiring from a beloved game is such a dicey decision., i dont want to over-judge a sportsperson on that basis. to me, that chase for 100 100's was a real blot. In a way, srt's average does justice to the fact that while he was king of the ODIs (arguably with Sir Viv), in tests, he was an equal rather than a greater mortal when compared to his peers (in my opinion)...
    [[
    I get the feeling that at least half a dozen current players are capable of reaching 55 in Tests. Look at the way Pujara is moving up. So SRT & Lara could very well go off the top-20, Lara first.
    Ananth
    ]]

  • POSTED BY arun on | August 18, 2017, 11:03 GMT

    thanks for publishing! Also, top sportspersons are competitive and they, i assume, would keep fighting to regain form. but it does harm the team. you can dot it when you are 28 30 and have earned the right to have a dip in form. but when you are 38 and in the twilight, its harmful to the overall cause as well...there is good reason for an observer to call that kind of thing "selfish"...maybe srt cud not see the bigger picture...it really was also up to the selectors to nudge. i would say that they did a shoddy job; too much hero-worshipping. I agree that he unfortunately missed his timing completely there, esp in ODIs. What a chance it was to go out in glory. What more did he want to do in ODIs and what was the point?! He anyways quit a year later! And his last 15 tests certainly blot the rest, decreasing his avg. by nearly 3 points, but I think some some leeway can be given there. (Of course, if there commercial interests, the whole point is moot.)
    [[
    I would say, sitting on the outside, ill-advised by others. I have no idea who are the others.
    Compare that with how Federer is handling his twilight years.
    He called off 6 months during 2016 because he knew he had a serious injury and had to get surgery done. There was no wavering. But he practiced like no one has ever did.
    He came back in 2017 with a bang, winning one GS and two Masters. Then no vacillation. He knew that the Clay season could kill him and he was off for over 2 months.
    He came back with another bang. Two titles, including Wimbledon.
    Off straightaway. No small tournaments. Very successful at Montreal other than the last hurdle. Immediate;y he is off one of his favourite Master's at Cincinatti. It does not matter that the no.1 is at stake. He would not bother if he did not get the no.1 again. He knows when to rest.
    Now US Open. I hope he competes. But would not be surprised if he takes off another month, to concentrate on the indoor season.
    Note the clarity of thinking. So much so, the others have started following.
    Oh! I know, Tennis is an individual game. But SRT was the king when it came to Indian Cricket. No one to touch him. Maybe the selectors could have discussed a little more.
    But the waters have already reached the sea.
    I get irritated when I see him 17th in the Batting Average table, that is all.
    Ananth
    ]]

  • POSTED BY arun on | August 18, 2017, 10:26 GMT

    Very nice and fun article. Sad that many people (?) missed the point and spirit.. Anyways, completely off-topic, but there were some comments, and hence... But totally fair if you decide to not publish it. Sportspersons' desire to retire rests on so many factors, that it may be unfair to judge. Lets assume what we know and that no commercial interests figured in SRT's decision. He had an extremely successful 2009-2011. Would'nt he assume that he is going to maintain form and continue in the same vein? It is a reasonable expectation. Isnt it then difficult for him to cast away his life so easily? A conundrum that many face since they cannot predict their future form. Dravid, for example, had a 3-y lean period. Perhaps he should have retired then? But he didnt and he had a great tour of England afterward. Was he lucky to rediscover his form? I mean, who knows, who can tell? Its not quite as pat. That said, chasing 100 100's was really sad (Ind lost B'desh ODI partly due to that!)
    [[
    I agree that it is the player's own decision and no body else's, when he quits. In a way, I agree with you. When someone is going through a prolonged poor patch and does not quit, there is a lot on the line. If he recovers, as Dravid did, we hail the decision to rough it out. If he does not, he find fault. Maybe unfair to the player concerned.
    Somewhat similar to a reverse sweep. Looks beautiful when it comes off. Looks ugly if the batsman perishes. Maybe a commentator should have the guts to take the batsman to task when he succeeds or refrain from criticizing if he fails.
    But like Kapil's last 15 Tests, these last 15 Tests also hurt when one peruses the careers of these two wonderful players.
    Ananth
    ]]

  • POSTED BY GV on | August 18, 2017, 3:02 GMT

    Ananth, this is a nice approach, but in my way of thinking, one of the pieces of work on batsmen you had done earlier - my favourite piece of work - was on classifying all innings played by a batsman as belonging to easy, middle or tough groups on the basis of bowling strength. I would look at the consistency of averages across those groups. If i remember correctly Mark Waugh averaged 40 regardless of whether the bowling was easy or tough, and that to me is an important indicator of consistency. Also on Randolf - where were you all these years? You really have brought tears to my eyes. But don't forget - no one else has the unique distinction of the 100th 100 and highest test score against the same opposition - Bangladesh. You may also want to look up Tendulkar's position in the tough group analysis I have mentioned above.
    [[
    Yes, Giri, I agree. That was, arguably, the best piece of work I have done so far. Especially because the reader participation was terrific. And the concept of BPQI which was developed as we went along on our discussions. Maybe I will re-do it in a simpler manner. Where the article gets published is a big question mark.
    Ananth
    ]]

  • POSTED BY Ashraf on | August 17, 2017, 9:26 GMT

    "Anantha Narayanan has written for ESPNcricinfo and CastrolCricket" Only could have been written by an Indian to not give us a cricket view but and indian view of it. Why tell us the position of Kapil Dev and Anil Kumble but not those of Imran Khan and Wasim Akram?
    [[
    My apologies for the strong words used, which have been removed.
    I have given Wasim Akram's and Imran Khan's details in response to a comment of aliasg3619310.
    But you belong to an elite reader group of one. The only one to accuse me of a bias towards Indian players. Normally it is the other way around.
    Ananth
    ]]

  • POSTED BY randolf on | August 16, 2017, 9:19 GMT

    Ananth, the reason why I disagree that SRT should have retired at the end of 2011, is the fact that it was not even an option as far as he was concerned at that time. I think it's well documented that he batted just for himself; and "to score a 100"! And at that time, the elusive 100 was still at large (the so called 100th 100). Hence, at that stage of his career he was most motivated to score a 100. So, he could not have; neither would he have stopped at age 38! Remember too, he and his fans kept fooling themselves that he was "a kind of Bradman" or "greater"! But though late, they discovered how much they were fooling themselves; because, the great Sir Don, in the process of retiring at age 40, had made "fifteen 100s during the last 40 inngs" of his career - but Tendulkar on the contrary, with all the positive inspiration and motivation in his guts, did try and also retired at age 40, but couldn't score a single 100 in his last 40 inngs; batting for all of 3 consecutive yrs (36 mths)
    [[
    I may not agree with all of you what have written but cannot but accept that the futility of those last many months will remain forever a black spot. Maybe he was ill-advised. Mybe not just his career was at stake but commercial considerations of others.
    Ananth
    ]]

  • POSTED BY Harsh on | August 15, 2017, 15:23 GMT

    Great batsmen comprising the top 10 are greenidge,Sutcliffe,Chappell ,Border and Anwar with only 3 averaging above 50.Really surprised not to find Dravid,Gavaskar ,Tendulkar,Headley,Steve Waugh ,Sobers and Miandad.Allwere an epitome of consistency.Amongst those who averaged under 50I felt Boycott,Kanhai or Peter May should have been right there.Most deserving in your 10 are Border.Sutcliffe and Chappell.
    [[
    Again, I am quite surprised. How do you conclude that Peter May SHOULD be there. Do you have his numbers readily available. Strewn in his career are 56.1% and 38.8%. However, despite these two really awful blocks, he has managed to get in at 75th (87.1%) - very close to Gary Sobers, Greame Smith, Graeme Gooch and Ganguly.
    Ananth
    ]]

  • POSTED BY GV on | August 15, 2017, 14:31 GMT

    For Ananth - I tried sending this article to you by email, but am not able to - please email me so that I have your correct email address... https://www.theguardian.com/sport/2017/aug/13/scariest-test-england-ever-played-terror-west-indies-cricket-1986-patrick-patterson See the youtube link mentioned in the article from 48:20, and you will get some idea of why one must occasionally look beyond numbers, just for fun.
    [[
    Giri, Where did I give the impression that I only look at figures.
    Ananth
    ]]

  • POSTED BY Harsh on | August 15, 2017, 12:40 GMT

    Great work ananth.Howver really suprising that the likes of Gavaskar,Sobers Headley,Dravid,Miandad Tendulkar or even Kanhai do not figure in top 10.The only all time great batsmen are Border,Sutcliffe and Greg Chappell.Ifeel the rating system may be unfair to batsmen with high averages above 50 as Miandads average never fell below 50 ever in his career.Lastly hard envisaging Geoff Boycott not there .It almost boils down to rating the greatest of batsmen as not up there with most consistent.
    [[
    Miandad has 9 blocks below 75%. To me, that is not the consistency promised by his being always above 50%. Let us not forget that his home and away figures do vary widely. Miandad is 10th (84.5%).
    Boycott is better with four blocks below 75%. This leads to a CI of 87.7% and position of 60th (87.7%).
    I will ask you to think over why Gavaskar,Sobers Headley,Dravid,Miandad Tendulkar or even Kanhai should figure in the top of the table. Headley does not aualify.
    Ananth
    ]]

  • POSTED BY sathya1169118 on | August 14, 2017, 18:37 GMT

    Seems the general public (myself included) are more at ease reading pieces which are either poetic free flowing articles or bundle of statistics (like the ask Steven type). But you have woven together a nice approach, an algorithm (if it can be termed so) to find out a few grains of truth!!! It would take more than a little patience to understand your approach, analyse the results, and may be appreciate it. But your job doesn't end with putting together the article, and results - also need to keep your energy reserves for replying to the comments!!!
    [[
    I am normally not that patient a man. However, somehow, God has given me the required balance when I handle the different types of queries.Maybe 1-2% of the queries - I see red. But the others, I try to understand.
    Thanks for nicely wording the entire philosophy behind such analysis-centric articles.
    Ananth
    ]]

  • POSTED BY hattima@yahoo.com on | August 14, 2017, 16:15 GMT

    Dear Ananth, Thanks for your reply. I must admit that I did not read most of the other comments. However, reading some of the other comments now, I completely understand your indignation. I do understand the rationale behind most of your analyses, and I do understand that you are skipping some of the technical details to allow it to be understood by the general audience. However, I hope you would appreciate that given the high quality of your analyses, it may attract some statisticians, who may wonder about these technical details. Since you have already done this subsidiary analysis like robustness check, how about putting some of it somewhere else for the interested group, with a link to that here?
    [[
    In fact, I had shown the top-10 CoV values in a table in my article. However the Editor, correctly, felt that it would take the emphasis away from the main theme. I will see what can be done. Maybe upload into my DropBox folder and give a link here. Give me a day or two.
    Ananth
    ]]

  • POSTED BY randolf on | August 14, 2017, 15:55 GMT

    Ananth, nice that @GURU toned down his rhetoric and apologized for his tirade. When I saw it, I thought that he was being grossly unfair to you, when he accused you of "…coming up with fodder for Tendulkar haters…" @GURU needs to know that every public figure has "critics", not necessarily "haters". And, as a career long critic of Tendulkar, I beg that you allow me to also tell @GURU that neither you nor any of the other brilliant professionals within your fraternity provides "fodder" for me. All critical thinkers need not to just accept information being fed to them - they should themselves verify it. Hence, in Sachin's case, it's he who provides me with the tons of fodder which I share, for his correct placement as a test batsman. Eg: @GURU has to know that SRT batted in 40 innings, for 3 yrs (36 mths) between Jan 2011 and Dec 2013 and couldn't score a 100 - that too, showed up in his CI. I disagree that he should have stopped at 184 tests; he was still young and enjoying his game.
    [[
    I did a special calculation. If SRT's last 39 innings are removed, his CI moves up to 88.1%. This would place him in the 51st position, almost in the first quadrant, 49 positions higher. Before anyone jumps up, let me say that those 290 innings are more than any other player played in his career. So it is clear that Tendulkar lost 3 in his Batting Average and 2.2% in CI.
    I doubt whether he was enjoying his game.
    Ananth
    ]]

  • POSTED BY hattima@yahoo.com on | August 14, 2017, 8:29 GMT

    Dear Ananth, Sorry for getting back so late, I was extremely busy the last few days. Let me explain my position on 10 innings not being natural. For example, suppose for one person 21st to 30th innings are the Ashes. For him maintaining consistency during that period is simpler compared to his teammate, for whom the same Ashes tests may cover 25th to 34th innings, while 21st to 24th innings might be in India. Hence, I am not sure everybody has a common ground in this comparison. In statistics, we refer to this as robustness: would the results change dramatically if we shift the window of analysis slightly, or if we move from 10 innings windows to 9 innings windows or 11 innings windows? If that happens (I do not means to check that), then perhaps we need to think of something else. Capping at 100 is also a similar issue. We know that in our mind 100s are special. But would the results change dramatically if we cap at 110, or 120, or 95?
    [[
    I know that if we change the basis from 10 to 9 or 11, the results will change. I have tried this out also. However, barring minor churnings, the collection of the top-10 players is almost there. That is one reason why I take issue with people who think I am presenting this as a list of great batsmen. I have not done that. I have also not tried various options and selected one in which my favourite batsmen are doing well. I have selected 10 and left it at that.
    Say, a batsman's career consists of 130 innings. These would have been played home, away, neutral, very strong opponents to weak opponents (and in between), batting paradises to minefields (and everything in between), covered pitches, open-to-the-elements pitches and so on. The 10 innings should just be taken as a microcosm of the career. It might not include all variations but I would feel happy if it includes, say, 6 home and 4 away innings or vice versa.
    So the final methodology might not pass the strict statistical validations. But it is a common-sense based methodology which will be understood by everyone, especially the non-technical lot. When they see Greig's career details, they can see for themselves, why he is at the top, something a CoV of 0.165 will not confirm immediately.
    By no means do I decry the pure technical methodology. I myself have enough statistical expertise to have done the entire analysis in a different way and presented multiple CoV tables. But then I would have lost many of the readers.
    Why 10? No real reason except that it makes some calculations easier and anyhow, it is the same for all players. It is just a sufficiently long career slice in which players are allowed the chance to make up for temporary losses of skill/form and for dropped chances/ great catches/unplayable deliveries/lack of opportunities et al.
    Ananth
    ]]

  • POSTED BY kiedar1864107 on | August 13, 2017, 19:41 GMT

    [[
    What do I do with an uninformed person's juvenile comment.
    In addition to being uninformed, the commenter has not bothered to read the article also.
    The only option is to blank out the comment completely and publish.
    Then at least will people think a little before doing something like this.
    Maybe a good thing will come out of this. Such people will not visit this space.
    Ananth
    ]]

  • POSTED BY victoria on | August 13, 2017, 12:40 GMT

    @Ian: Thanks for this massive compliment! I'm cock-a-hoop to hear that my opinion mirrors that of the great David Lloyd. However, @"Victoria" is absolutely not a pseudonym for David Lloyd. Incidentally though, I have the greatest respect for David Lloyd as a cricket expert. For me, he's the most honest of any of those guys who usually take up the microphone to describe cricket the game, and the people who play it. And why you can rely on the honesty of such people as Ananth and Lloyd is the fact that they don't say things just to protect their economic survival - they say it as objectively and unbiased as they see it - they're not members of the "supper-sopranos". KP Pietersen didn't achieve a 50 runs avge, but his name is mentioned with the 4 greatest of All Time; because, if he was not delved that dirty blow by his bosses, he would've eventually got there. He was unfairly axed at the 'tender' age of 33 - at the age when I'm sure his maturity as a batsman would've been most prosperous.
    [[
    When I see that three English middle order players of the current team would not have found place in the English 'A' team a few years back, I feel that somewhere the egos should have taken a backseat and Pietersen should have played for the past three years. Even today, no one from English top order, other than Root, can come anywhere near KP. But all the water has long gone past the bridge and reached the sea. Thanks, @Vick, for one of the nicest compliments I have received.
    Ananth
    ]]

  • POSTED BY pramathesh on | August 13, 2017, 11:56 GMT

    Instead of 3000 runs and 100 wickets, how do your results change if u use tighter cut-off of min. 5000 runs or min. 200 wickets in tests? Was expecting that only a consistently performing player will be selected to play for more tests and will prove to be the horse for a long race. Ravi Ashwin & Ravi Jadeja are yet to prove themselves in tests in Eng, SA, NZ and Aus. For all-rounders like Kapil, Imran, Kallis, etc, do the CI values take performance in batting as well as in bowling ? One day cricket gained popularity from 1980's decade implying that players had to give time to tests as well as odis-this can take a toll on fitness. What about CI value for odi performance?Sachin Tendulkar was selected to play for India in 200 tests and 463odi's in a period of 24 years from age 16 to age 40 because he managed to score more than 15K runs in both versions respectively & also kept himself fit to play for those many years.
    [[
    As I have mentioned in one of my responses, I will give below a set of players who will miss the bus.
    Graveney, Dhoni, Simpson, Andy Flower, Redpath, Sutcliffe, Peter May, Weekes, Mohinder Amarnath, Martyn et al.
    Laker, Prasanna, Barnes, Davidson, Ajmal, Akhtar, Sarfraz, Lock, Tayfield, Miller et al.
    Nothing more needs to be said.
    CIs for ODIs is worth exploring.
    Ananth
    ]]

  • POSTED BY Ian on | August 12, 2017, 17:43 GMT

    @POSTED BY VICTORIA ON | AUGUST 10, 2017, 12:25 GMT: Are you really David Llyod under a pseudonym, by any chance? Here's a clip of David Bumble Lloyd talking of his favorite cricketers, amongst those that he has SEEN over the 50+ years of his interest in the game. https://www.youtube.com/watch?v=sxw8YmaWyQ0 Guess which players he named: Gary Sobers, Viv Richards, Brian Lara and.....Kevin Pietersen! For the record I completely, absolutely agree with your selection of Bradman, Sobers, Richards, Lara and Pietersen. Bradman and Sobers played before I was born, and Richards, stopped playing before I developed any real interest in cricket. But in my time seeing cricket, I have not seen any batsmen turn the game single-handedly against Gun-attacks like Lara and Pietersen have.
    [[
    Bradman, Sobers, Richards, Lara and Pietersen it will be. I may add Stan McCabe and Neil Harvey to the list though.
    Ananth
    ]]

  • POSTED BY victoria on | August 12, 2017, 15:52 GMT

    Ananth, it's good to see that you're not being sidetracked by some of these wild responses. All this great piece of statistical work has done is, to show cricket enthusiasts how consistent different players have been, at meeting their established expectations, "right throughout their careers" - nothing else. You have also hit the nail on its head within the article itself, and in reply to a few subscribers that, "because Player A has a higher career average than Player B doesn't necessarily mean that Player A is more consistent than Player B. Let's use the Shane Warne vs Sangakkara example - purely as batsmen. Warne might have accumulated a block of 10 scores at a range between 16 and 19, giving an avge of 17.3. Sangakkara on the other hand may have 10 scores of: 7, 10, 253, 8, 10, 09, 10, 10, 6, 233, giving an avge of 56. Suppose they're in the same team playing a series of 5 tests, against a second team that's evenly matched with accurate precision, which batsman will you pick?
    [[
    No one gets picked on consistency. People get picked on performance.
    Ananth
    ]]

  • POSTED BY Shehan on | August 12, 2017, 13:45 GMT

    According to this analysis Sangakkara is very inconsistent. But today I found that Sangakkara features twice in the list of most consecutive 50 + scores. [http://stats.espncricinfo.com/ci/content/records/283043.html] How can this happen ?
    [[
    Sangakkara has played 233 innings. Out of these, he has two streaks of 7 and 6 innings respectively in which he crossed 50. Do these 13 innings convey anything to us about his other 220 innings.
    In the 23 blocks, Sangakkara scored below 50% of the block mean no fewer than 4 times. This sort of huge under-performance has pushed his CI to 84.7%. Kindly don't immediately make this out to be a criticism or pulling down of Sanga, who happens to be one of my favourite players.
    Ananth
    ]]

  • POSTED BY Guru on | August 12, 2017, 7:39 GMT

    1. I admire Ananth for his openness. I honestly didn't expect any of my comments to be published. 2. These are ananth's own words : "Sangakkara is quite inconsistent. His ci is 84.7 percent". So he does imply 85 percent is not consistent. 3. I initially didn't pick Warne. In fact, I initially didn't even pick Tendulkar. I used an example of Tony creig and Herbert Sutcliffe to clarify that consistency here doesn't mean better. Ananth agreed. But that didn't stop haters from using this analysis for defaming Tendulkar and his fans (at least two comments after my innocuous Frist comment). That's why i had to pick an extreme case to highlight what Ananth already mentioned in the article. 4. Sorry Ananth, for the tirade. I don't mean any disrespect. Will stop with this.
    [[
    I only compared 84.7 to much higher numbers. That is all. Like in my response to Dravid's consistency, the numbers do not indicate so.
    I know that there are those who will pick specific comparisons to put down SRT just as there are others who would use such numbers to push up SRT blindly. I will publish any comment provided they do not villify me, another reader or any player.
    I have told this many times. I have my favourites, I can say that. But I will not do anything just to glorify them and put down others. And I felt strongly about SRT's extending his career (or almost conned by vested interests to extend) by easily 20 Tests. I would have been be happier if he had quit at the end of 2011 with figures of 184 Tests, 15183 runs at 56.03. But that does not mean I do not admire his technical skills, his role model behaviour, his never forgetting his middle class background and what he did for the game in India.
    And let us agree that there can be total consistency at an average of 10 and total inconsistency at an average of 50. That will never make the first batsman a better one. Make that disastrous 2 in 10 innings to 10 in 10 innings (two streaky fours), Chris Martin would have crossed even 90.0. It would have only made him a very very consistent but worst batsman ever.
    I can see that you are not the typical SRT fan, ready to bad mouth me. You wanted to make a point and you made it. My apologies if I came on too strongly.
    Ananth
    ]]

  • POSTED BY randolf on | August 12, 2017, 6:04 GMT

    "But what haters see is 'Tendulkar is not consistent, so his fans are idiots'" (@GURU ON | AUGUST 11, 2017, 14:56 GMT). @GURU, you are absolutely correct - only an "IDIOT" would think that anybody who sees a batsman with a consistency index (CI) of "85.9%" is "not consistent"! Which school did you go to? In nearly every proper school, 85% is an "A" or better. In fact, you are the one who seem not to understand what Ananth's analysis has done here. Nowhere in the article has he implied that Tendulkar was not consistent. All that he is instead saying is that 99 other batsmen were more consistent than SRT in producing at the best standard that they set for themselves. It is clear that you are the only one who is wrongly interpreting "more consistent" to mean "a better batsman than". Now, the question to you @GURU is: "Why pick on Shane Warne who is not a batsman, to make your foolish point concerning Tendulkar? Why not choose one of the MANY batsmen who are superior to him?
    [[
    You have captured the problem with gentlemen like Guru perfectly. Thank you.
    When he asked for the details of Warne, I provided the figures little realizing that the purpose was devious. Just to see if Warne had a higher CI% than SRT and come out with the tirade. I have no problems. If someone asks for the figure of Chaminda Vaas, I will give. If they misuse my openness and misuse the same, they are the losers, not I.
    Guru talks as if I made a statement "Harbhajan and Agarkar are better batsmen than Chetan Chauhan since they have scored Test hundreds and he has not". I have presented numbers without making a single comparison. Unfortunately, that does not suit some people. So they invent a statement and talk as if I made that statement.
    Ananth
    ]]

  • POSTED BY bhavan5206519 on | August 12, 2017, 3:02 GMT

    Sir, I read through your article and have a hard time grasping the results that your analysis shows. Couple of things to ponder is what is considered a good test average for a batsman and a bowler. To me a batsman should have a career average of 40 plus and a bowler 30 minus. That was the holy grail when I was growing up. What happens when you add that criteria to your 3000 runs? You will see a totally different set of players emerge that to this day we consider them greats. Case in point is Rahul Dravidian. The man never had an inconsistent series till the very last one and the only reason he is down the totem pole is there is no batting average cutoff. Is it possible for you add the 40 plus and 30 minus criterion and run this data?
    [[
    1. Dravid had 6 blocks in which he scored fewer than 70% of the block mean. So you are on the wrong track when you say that he was very consistent. You have perceptions only. I have figures.
    2. So you want to exclude Fletcher, Vijay, Alec Stewart, Tamim Iqbal, Colin McDonald, Dileep Sardesai, Clem Hill, Victor Trumper, Majid Khan, McCullum, Dhoni, Michael Atherton et al from the analysis.
    3. Also exclude Alf Valentine, Prasanna, Srinath, Hoggard, Lawson, Doshi, Brett Lee, Southee, Titmus, Vinoo Mankad, Harbhajan Singh, Abdul Qadir et al from the analysis.
    Ananth
    ]]

  • POSTED BY Bennett Mendes on | August 11, 2017, 19:41 GMT

    At first, I was thinking that this analysis could expose the Flat-Track Bully (FTB) batsmen for being inconsistent - i.e. those who score many runs when the conditions are good, yet fail when they are not good. Then I saw Victor Trumper's name as one of the most inconsistent batsman. He was known for throwing away his wicket when the conditions were good, but would fight/score when they were not good.

  • POSTED BY Guru on | August 11, 2017, 15:20 GMT

    Shane Warne is 86.6 percent likely to score 17.3 runs. Tendulkar is only 85.9 percent likely to score 53 runs. So, Warne is a more consistent batsman than Tendulkar. That's your analysis. b
    [[
    At times it is totally futile to even try to communicate with people who are seeing things through five layers of coloured glasses.
    Ananth
    ]]

  • POSTED BY Guru on | August 11, 2017, 15:08 GMT

    I know Shane Warne's batting average is 17.3. Is there anyone in the list with a lower batting average? Can you give a distribution of the averages to the first half? Like, how many have average less than 20, how many 20 to 30, between 30 to 40, and how many 40 plus..

  • POSTED BY Guru on | August 11, 2017, 14:56 GMT

    The problem for me with your articles is that you keep coming up with fodder for Tendulkar haters (although, I agree, that's not your intention). While most of articles do have some truth, this one is purely misleading. As per this analysis, Shane Warne is a more consistent batsman than Tendulkar, Dravid, Ponting, kallis, Sangakkara, Viv Richards, imzamam up haq, Mohammed Yousuf, Azharuddin, Kohli, Smith, Amla. But what haters see is "Tendulkar is not consistent, so his fans are idiots", without understanding what "consistent" here means. I don't blame you for that, I'm just frustrated. Btw, I did the math for Warne's batting, and I got 89.6%. so he should be ranked in the 20s, even better than the Don

  • POSTED BY Usman on | August 11, 2017, 13:59 GMT

    Great analysis!! I have a question though. If a batsman plays 170 innings with a career average of 30; and each block he is very near to the 30 average; he would be ranking very high in this type of list. But we all know 30 average at any era is a not so good batman. Consistent in scoring 30. We can conclude that Tony Craig is the most consistent player in test history in scoring 38-40 runs per innings.
    [[
    Let us consider three statements.
    Batsman A averaged 40 but was very consistent.
    Batsman B was quite inconsistent but averaged 50.
    Batsman C was extremely inconsistent. He averaged 60.
    That is all. You draw your own conclusions.
    Ananth
    ]]

  • POSTED BY randolf on | August 11, 2017, 12:14 GMT

    Ananth, great work as usual! It leaves out nothing, taking every ball bowled and faced into account. This is great because, a bowler can take a wckt with the first ball he bowls; and a batsman can get out with the first ball he faces. I take consistency as the most important factor in batting. It's the outcome of the other factors such as technique, temperament, et al. I also note once again, maybe for the millionth time, that credible, unbiased and objectively analysed data shows that Tendulkar is not necessarily among the first 25 batsmen who played test cricket. Everybody knew that he had a good technique, and that's what was blowing away most of his fans. He scored more runs and 100s than any other player, just because he played many more games than them - PERIOD! Cricinfo has a flawed statement about him saying that he's "the most complete batsman". I don't know what that means! Eg. Dravid played every type of bowling as well as SRT, but Rahul was few notches better against pace.

  • POSTED BY Jitendra on | August 11, 2017, 10:43 GMT

    A cap put above 100 %. A cap should be there for lower performance. Like Imran khan has not bowled much in last matches of his career. So his wicket per block will be less. For lower scores a cap should be there.
    [[
    I am afraid you have not understood the explanations and the need for capping. And we can only cap high performances, not low ones.
    Ananth
    ]]

  • POSTED BY aliasg3619310 on | August 11, 2017, 9:30 GMT

    I would like to know about, Mohd Yousuf (80.8%), inzamam up haq (84.0%), Younis Khan (89.1%), Javed Miandad (84.0%), VVS Laxman (87.5%), Imran Khan (88.9%), Wasim Akram (88.6%), Waqar Younis 83.2%), Saqlain Mushtaq (89.7%), Javagal Srinath (87.3%) and Venkatesh Prasad (only 96 wkts).
    [[
    The numbers are interjected.
    Ananth
    ]]

  • POSTED BY Swapnil on | August 11, 2017, 9:16 GMT

    I somehow disagree with the method. Consistency should be measured with some reference. How can we put Don Bradman so low ??? If he was not consistent .. how can he maintain >99 average. I think for batsman ... progression of batting average over various time windows should determine if he was consistent or not. Let's say, for Sachin Tendulkar when he started his test batting average was less bet. 30-40 ... later for almost 10 years, it was between 55-60 .... then at the end of his career, it reduced to 50-55. If you consider 50 as benchmark batting average, we can say he was consistent most of his career. Variation should not matter in long run. How will you rank the following batsman in 10 innings - 1) 30, 40, 50, 30, 40, 50, 30, 40, 50, 30 2) 100, 30, 70, 100, 50, 20, 30, 50, 100, 10 3) 50,50,50,50,50,50,50,50,50,50 4) 300, 30, 20, 10, 0, 50, 70, 90, 30, 40 For me as a cricket fan, rank will be --> 2, 3, 1, 4. Let me know your thoughts.
    [[
    Your statement "how can he maintain ........." is flawed. A batsman can score 200, 0, 200, 0 and get a 100 average.
    And why make another flawed statement "How can you put Bradman so low". In the 'Runs scored' table, Bradman is 48th.
    I am not going to answer your contrived question. But I will say that if my life depended on a batsman scoring 50, I would any day take Batsman 3 and stay as far away from 2 and 4. Reduce the number to 30, I will take 1 and 3.
    Ananth
    ]]

  • POSTED BY Desmond on | August 11, 2017, 9:07 GMT

    Good analysis. One question: Is(Are) your measure(s) statistically sound? Unbiased, consistent, efficient, sufficient, etc.?

  • POSTED BY Guru on | August 11, 2017, 8:48 GMT

    I would like to know Shane Warne's Ranking in this analysis as a batsman. He has 3154 test runs.
    [[
    Warne: 85th (86.6%).
    Ananth
    ]]

  • POSTED BY Himanish Ganjoo on | August 11, 2017, 8:28 GMT

    Great way to analyse consistency! Especially the capping of high performances at 100%. This ensures that high peaks do not add to variance and the final consistency index is only affected by lapses in form.

    The graphs were also a great way to easily see consistent performers. Along those lines, have you thought about maybe computing the differences in averages of the blocks and taking the mean of that? It measures the fluctuations between block performances, and matches nicely with the information that is presented in the line graphs. A player with a low mean difference is more consistent.
    [[
    The CI is the mean of the % values (100% or below). As such it represents the consistency quite effectively.
    Ananth
    ]]
    It might match very nicely with your original measures, and I can do a computation sometime next week to check this new metric.

  • POSTED BY Piet on | August 11, 2017, 7:41 GMT

    Dear Anantha, thank you very much for a most insightful article. A few questions: Is there somewhere that I could access your complete list? Also, how would one measure all rounders or fielders? I tried to work out Jacques Kallis's Consistency Index and arrived at 85.5% for batting, and 84.8% for bowling. Would it be possible to get a combined score on his consistency as a player?
    [[
    Your numbers are more or less correct.
    Re all-rounders, as I have already mentioned in a response, we have to get the thinking done from ground up. Did the player deliver as a batsman, bowler or both. Does an all-rounder get condoned for delivering only as a batsman (or bowler) and not doing anything on the other discipline - a la Imran Khan? How do we handle situations where a player has only bowled and not batted (and vice versa). And so on.
    I will think over it and if I can come out with a viable solution, come out with an article. Through which outlet is uncertain now.
    Ananth
    ]]
    Once again, thank you for your great articles. Piet

  • POSTED BY mujju14889803 on | August 11, 2017, 7:03 GMT

    great work, I would like to know about Rahul Dravid & Mohammed Azharuddin
    [[
    Dravid's figures have already been given.
    Azharuddin: 85.3% (117th).
    Ananth
    ]]

  • POSTED BY rukman2441469 on | August 11, 2017, 6:29 GMT

    Well done. Cricket is a performing sport and players must be paid as per performance. Now that 10 IPL seasons are over plenty of performance data is available. Can you come up with statistically sound plan by which a player,s fee comprises a base pay of say 25% and the rest should depend on agreed measures of performance. We have seen many instances where so called star players with high prices performed poorly while many young players with very low fees performed very well. I am sure with your expertise and all the available data you can come up with a model which can be reviewed and refined.
    [[
    Unfortunately or (fortunately for me), I stay off IPL completely.
    Ananth
    ]]

  • POSTED BY S on | August 11, 2017, 5:17 GMT

    Tony Greig's entire test career happened to coincide with the period in which I was a complete cricket fanatic. I remember being amazed by Greig's consistency - not just as a batsman but as an all rounder. Every score card for any test he played would show a player who would somehow muster about 70+ runs, about 4 wickets, and about 3 catches per match. Its wonderful to see that impressionistic memory is backed by the numbers, at least when it comes to batting. Thanks as always for a cerebral and interesting analysis.
    [[
    Unfortunately, Greig left the Test scene in a cloud. His tour-de-force was at Trent Bridge against New Zealand during 1973: 4wickets+139 off 178+3 wickets. England won by only 38 runs. Greig came in at 24 for 4 in the second innings. And at Bridgetown the next year. 148 + 6 wickets + 25.
    Ananth
    ]]

  • POSTED BY Sundar Raman on | August 11, 2017, 4:51 GMT

    Instead of doing all the crazy stats, why can't we simply find the difference between Mean and Median. Both are measure of central tendency but also reveal the impact of punch above the weight. 10 innings can be misleading as Most of the Ausi and Eng players especially batsman played 5 test match Ashes series with minimum 7 to 8 innings. If a team bowlers don't know how to get him out, they have no way to get him out through the series. there are many examples for that, For example, Clark and Smith in Austrialia against India . Indian team bowler has no clue how to get them out. They played at least 6 innings in that series. Either it should be at least 20 tests to find out consistency period for a player.

  • POSTED BY Abhivadan on | August 11, 2017, 3:31 GMT

    Hey, Thoroughly thought through piece as usual. When I read the title, my first thought was going by moving averages in blocks of 5 or 10. Here's a small flavour: 1. I really like the idea of capping the high performances, since they are not supposed to add to the variance. 2. Moving averages then could give us a better view of the purple patches or the slumps e.g. A batsman hit a low, in innings 7-13, but was otherwise alright. In the current analysis, he probably would turn up alright in both the blocks, but if we look at the a moving average we will surely pick that slump. 3. Half blocks not an issue. (67 innings or 63 innings) 4. The series of moving averages can serve the purpose of studying consistency as well as the position of that consistency viz. Mean and Variance of the moving average series 5. Moving average as a concept is easy to understand. And with your way of explaining things, it will be understood by a lot. Your feedback is appreciated. Thank you
    [[
    I understand and appreciate the idea. It removes the rigidity imposed by taking in an arbitrary number such as 10.
    However one of the strengths of my analysis is that it would be very easy for any reader to get a batsman career data into an Excel sheet and determine the CI by himself. A moving average determination might be tougher.
    Ananth
    ]]

  • POSTED BY Matt on | August 11, 2017, 2:47 GMT

    Anantha, in an earlier comment you ask for an example of a batsman that is selected base don location. Well, as of this year that unfortunate soul is Usman Khawaja.
    [[
    I can only admire at your inventiveness. I agree. As (If) and when Khawaja crosses 3000 Test runs and gets into this analysis, we will take his CI with a pinch of salt. Can you think of one older batsman who was selected to play at home but not to travel?
    Ananth
    ]]

  • POSTED BY Matt on | August 11, 2017, 1:59 GMT

    Gee Anantha, some people don't get the point. Yes readers, consistent does not equal best or most valuable, but that is why the article does not even imply that to be the case. It is just an interesting look at which players tended to play to their potential year in year out.

    And fantastic to see the analysis on Chris Martin the Binary Man.

    I would be interested to see the index for the following players if you have the time: David Boon, Mike Hussey, Chris Rodgers.
    [[
    Boon: 86.6% (82nd).
    Mike Hussey: 87.0% (78th).
    Rogers did not qualify.
    Ananth
    ]]

  • POSTED BY Matt on | August 11, 2017, 1:45 GMT

    Nicely done. I like it a lot. Interesting to consider reasons for inconsistency in certain players. For example Shane Warne had a lean period when he bowled with shoulder injuries. Herbert Sutcliffe is amazing when you consider that in those days you could get caught on a sticky wicket and a wet summer should have destroyed your figures. It makes sense that bowlers are more consistent than batsmen. Why? Well it is the nature of the game. For a bowler, if you bowl badly there is always the next delivery, the next over, the next spell (in the more traditional sense). Lose your radar with the new ball and come back with the old. For batsmen, one error in judgement and it is goodbye. Imagine how bowlers would fare if every time they bowled a wide they were off for the innings!
    [[
    Good point. 20-0-80-0 could easily become 22-0-90-4.
    Ananth
    ]]

  • POSTED BY Cricinfouser on | August 11, 2017, 1:04 GMT

    Sorry if I come across as biased but very few W\west Indians r in this analysis. I would l like 2 see the results for Shiv. Gayle n Sarwan, Roach those should b pretty erratic. also maybe Bishop vs Walsh vs Ambrose, or maybe 2 make things really interesting compare the bowlers from the golden era of Windies fat bowling. Croft Roberts holding garner n Marshall
    [[
    Chanderpaul: 81st (86.7%).
    Gayle: 94th (86.2%).
    Sarwan: 41st (88.5%).

    Croft: 92.3% (23rd)
    Ambrose: 90.1%
    Roberts: 89.5%
    Garner: 89.2%
    Marshall: 88.9%
    Walsh: 88.5%
    Holding: 86.9%
    Roach: 84.7% (151st).
    Ananth
    ]]

  • POSTED BY heathq1437344 on | August 11, 2017, 0:32 GMT

    Really, why are you statisticians so preoccupied with diminishing the contributions of our greats. You cannot beat the batting and bowling greats true stats, so find ways to obscure results to fit into your own idealisms. Good work (not).

  • POSTED BY Vinayakaram Nagarajan on | August 10, 2017, 23:52 GMT

    Very interesting; but let's take Ashwin for instance. He is picked only on turfs suited to him and he is bound to be consistent. There are articles around how his average is vastly different from home and away. So there is never 'the' statistic that determines a players all round consistency. These outliers are hard to spot.
    [[
    Your example is primarily a bowler. Can you think of one batsman who is selected based on location?
    Ananth
    ]]

  • POSTED BY Ruchit Khushu on | August 10, 2017, 22:29 GMT

    I think consistency alone can be used as an adjective in this cases and needs to be qualified with great,good,ordinary and poor i.e consistently great, good, average,ordinary and poor !! For example Tony Grieg can be consistently good but not in the category of consistently great -for a which player has to be great basically like Bradman, Sobers, Sachin,Lara etc..which then begs if this comparison/analysis makes any sense!! No it doesn't

  • POSTED BY victoria on | August 10, 2017, 22:06 GMT

    Ananth, don't worry with that 'sour grapes' crew who usually curse every cricket statistical analyst for not "HAND PICKING" their individual hero(es) and just "PUTTING" them in top positions that they don't deserve. Yes, when some of them look at the objectively tables of merit and discover that their particular heroes don't top the charts, they condemn excellent work that is credible, objective, unbiased and truthful. Your work reveals exactly what it was set out to do. It unveils "the most consistent cricketers ever" - meaning, "the players who have been most consistent throughout their entire careers - not if they were inconsistent at the beginning, then super consistent in the middle, and inconsistent at the end; or whatever pattern of consistency they developed - the most consistent are those who have been consistent right through their career", and endorsed by graphics which show that players whose consistency index is closest to a straight line (100% CI) are the most consistent.
    [[
    Thank you. Through the years I have battled such readers. However, I see the point of some of the dissenters, only a couple for this article, which is that the standard statistical measures could be used. That is fine.
    Ananth
    ]]

  • POSTED BY Cricinfouser on | August 10, 2017, 20:17 GMT

    Don't doubt the stats, but in a team game the aim should be to get players who play the way the circumstances require? And "playing for the needs of the team" would seem to be a bit contradictory to "consistency".
    [[
    The same can be said of the players playing 'outlier' innings. In my opinion any so called 'selfish' innings is likely to benefit the team. Also where is there any reference in the article to value to the team or team player etc.
    Ananth
    ]]
    Having said that, & to contradict my own argument, from memory Tony Greig did seem to be pretty much the ultimate "team player"

  • POSTED BY mzrahaman on | August 10, 2017, 19:02 GMT

    I prefer the eyeball test, i.e., you know a good batsman or bowler when he is out there on the field performing. It's easy to just look at the stats 10 or 20 years after and not know about the circumstances under which the player performed. How do you judge a player who was given out unfairly or was run out by his partner or was told to bowl a certain way to entice a batsman to take chances by hitting out? I love Tony Grieg and I think that he is one of the most underrated players, but who would take him above Bradman, Richards, Tendulkar, Lara, pointing, et al. And is that not the point of trying to figure out who is/was the best. Holding once said of Viv Richards that the only reason why Richards did not score more runs was because he did not have to. This is one of those cases when the author of the article is trying to make something out of nothing just to prove to us how smart he is.
    [[
    This is the sort of conclusion which I warn against but you guys keep on doing. TELL ME, WHERE HAVE I MENTIONED THAT GREIG IS BETTER THAN ANY ONE OF THE PLAYERS MENTIONED?
    Ananth
    ]]

  • POSTED BY Christiaan Kriel on | August 10, 2017, 18:31 GMT

    Had a look a Tony "off the meat of the bat" Greigs' stats after reading this article. 58 matches with 8 hundreds a 20 fifties kind of stands out.

  • POSTED BY Six on | August 10, 2017, 18:15 GMT

    Very interesting stats re: Pietersen. It's more interesting to me as I've been arguing with people for years that Pietersen was very consistent. People tend to lie and talk nonsense in cricket to fit their agenda and this stat concerning Pietersen makes me laugh. A comment about fitting agenda... Bradman is by miles the best ever batsman in test cricket whether x, y or z likes it or not. His average is almost 40 clear of everyone else, he played a decent number of tests, no helmet, worse bat and on uncovered wickets. In current conditions a batter would need to be averaging 150+ at least to compare to Bradman in theory and Bradman simply can't be compared to anyone. He was on another level to all of them. I will read the article properly at a later time, but I imagine Bradman's enormous average and ridiculous stats are working against him when it comes to the measurement for consistency. Interesting article.

  • POSTED BY Anshul on | August 10, 2017, 17:46 GMT

    After reading the nice article, I myself tried a similar analysis on a few batsmen with different methodology. 1. Calculated a running average of 10 innings for each batsmen. 2. Then calculated the std-dev of running averages against the batsmen actual average. Here's what i got: Tony Greig - 8.5 Darren Bravo - 10.4 Dennis Amiss - 18.1

    Seems like the lower the deviation, more consistent is the batsmen.

  • POSTED BY Aatif on | August 10, 2017, 15:51 GMT

    I would like to know of Kallis,Viv Richards,Walter Hammond,Steve Waugh and Haneef Mohammad.
    [[
    Kallis: 111th (85.5%).
    Richards: 112th (85.4%).
    Steve Waugh: 103rd (85.8%).
    hammond: 161st (82.5%).
    Hanif: 136th (84.4%).
    Ananth
    ]]

  • POSTED BY kuldeep on | August 10, 2017, 15:33 GMT

    Very happy to read a article from you.I know it will be very interesting. Dear Sir,What if we see sangakkara as a batsman.means we count only his tests after 2006 When he give up wicketkeeping. also where Abd stands? also my favourite Virat kohli. Dear Sir,one man also devised a method for rating called impactindexcricket.com i think you will like it and you can write a precious article on it. Sorry but asking too much where s.smith,amla stands in your index? waiting for your reply sir. Thanks for superb article. also once again sachin od proved a not one of greatest in test.The method i tell impact index also put sachin Out of Top 30. No one can challenge lara.
    [[
    ABD: 113th (85.4%).
    Kohli: 146th (83.6%).
    Steve Smith: 99th (85.9%).
    Amla: 129th (84.1%).

    Ananth
    ]]

  • POSTED BY Ved on | August 10, 2017, 15:12 GMT

    NIce work! I am actually surprised by Tendulkar's position. I always thought (feelings can be misleading!) that in blocks of innings he appeared to be consistent. Is his consistency rating adversely affected by the last two years of his career?
    [[
    Out of Tendulkar's 33 blocks, he is above the mean in 16 (capped at 100%) and below in 17. Out of the 17, 3 are below 50%, and that really hits hard. Another 5 are below 70%. So all these have pulled the numbers down.
    Of course, the last five blocks are all below 100 and the last two below 65%. SRT would have been much higher if he had closed his innings at, say, 300 innings.
    Ananth
    ]]

  • POSTED BY Guru on | August 10, 2017, 15:00 GMT

    Are you not punishing batsmen with higher average.. for example, Tony creig is 94% likely to score 387 runs in 10 innings. Whereas, Herbert Sutcliffe is gauranteed to score at least 459 runs(his minimum block). But according to this analysis, Tony is more consistent. Theoretically,may be he is. But I don't see the point of such argument.
    [[
    There is no argument here. If you wanted to select a team, you will not select based on consistency only. You would use other factors such as Average and you would, of course, select Sutcliffe.
    Ananth
    ]]

  • POSTED BY Carbon on | August 10, 2017, 14:35 GMT

    This article is a prime example of why math gets a bad name in the popular culture. First off, what problem is being solved here. If a selector needs to pick a team, he will not be looking at these measures. Has the author tried his measures in a fantasy cricket tournament. There are perfectly good measures of consistency available in the literature that have been tried in every sphere of life and more importantly have been vigorously reviewed by professionals. The author introduces an entirely new measurement system with barely a nod to the richness of existing work and even there he says that the existing measures give the same results. So why bother with something new. When pressed in a follow-up comment, he summarily rejects the notion by saying that such measurement would be beyond the grasp of the average reader. So first the author is insulting the process of scientific inquiry and then the abilities of the reader.

  • POSTED BY anupam4457605 on | August 10, 2017, 14:20 GMT

    Hi Ananth,

    Brilliant, as always! Great stuff.

    A thought that I wanted to bounce off though. Who is more consistent: Batsman A: 30, 30, 30, 30 Batsman B: 30, 80, 30, 60

    Inherently, using their career mean, pushes up players with low averages in terms of consistency. Therefore, I was wondering if the Mean used should be the same for all batsmen or possibly a mean adjusted for the era they played in.

    Happy to be challenged and apologies if I missed something in my understanding.

    Best regards Anupam
    [[
    Ubnfortunately not a good comparison since their means are different.
    Ananth
    ]]

  • POSTED BY hattima@yahoo.com on | August 10, 2017, 14:19 GMT

    Dear Ananth, I must admit that the idea you have presented here is very nice. However, I was wondering if something could be done to reward extraordinary performances, rather than using the capping you suggest. Earlier you suggested runs per tests as a measure of consistency in an article last year, which did not need such capping. How about looking at series wise consistency, considering 3 match series as minimum, and clubbing the other stray tests with one of the earlier/ later series, whichever happens on a closer date? Based on these also, I believe, a modified version of your proposed index can be computed. I prefer this as it is more natural than 10 innings windows, which may happen in very similar or very different conditions, and hence may be harder to compare/interpret.
    [[
    The capping is the most important feature of this methodolgy. A series base will not work well now. It might have worked until, say, 1980 when 5-Test series were in vogue. Now we have 1, 2, 3, 4 and 5-Test series and it is difficult and not worthwhile to group these. Your last point is the exact reason why I feel the 10-innings block is correct. Across opponents and locations, the player is expected to delivering as many blocks as possible or come close. It is a very fair method.
    Ananth
    ]]

  • POSTED BY bharat7055017 on | August 10, 2017, 13:59 GMT

    I skimmed through the content. I have to appreciate the time author has spent to come up with this analysis, but sorry, this seems to be a complete waste of time and effort though. Analysis is needed on a subject to evaluate the worth of the subject when it is difficult to gauge the subject using basic metrics. Cricket has never been the sport that needs complex analysis to measure the value of a player :) Batting/Bowling - Home/away - batting/bowling avg- StrikeRate/RPO along with couple of straightforward metrics are all we NEED & USED to pick players at any level. Other stats won't mean nothing, they are as useless as the by-products that are released into dump sites when manufacturing products. But I am kind of happy it is a livelihood for some folks!

  • POSTED BY John on | August 10, 2017, 13:57 GMT

    An interesting read - there are many ways of doing this but the methodology sound quite reasonable. One comment about Greig. He entered the test team as quite a mature player (he had to qualify for England). He left when at the top of his game. His entire career only spanned 5 years and 2 months, so although there are a good number of tests (58) they were in a very short period. It is not surprising that within that rather small window he was consistent - he didn't have the upward and downward trends many players have a the beginning and end of their careers (as you note with Amiss).
    [[
    That is a very perceptive and invaluable insight. What you say is true. The SA origin and Packer at the end made sure that he had a very fruitful, but short career. I probably missed mentioning that. Many thanks.
    Ananth
    ]]

  • POSTED BY Venkatesh Venkatesh on | August 10, 2017, 13:57 GMT

    Only few handful cricketers stands out among so many cricketers .It is most unfortunate Tony Greig played is not with us after Sir Garry Sobers he was best all rounder then .

  • POSTED BY Jury on | August 10, 2017, 13:35 GMT

    Where's Sanga MR consistent and Dravid ??? better than above all
    [[
    Both already answered. Pl see earlier comments.
    Ananth
    ]]

  • POSTED BY Luka on | August 10, 2017, 13:33 GMT

    This analysis is quite useful in answering the exact question the author poses. However, in the bigger picture, I feel this sort of analysis rewards mediocrity. It is much easier to consistent, but average, than to dazzle for long periods of time. Also, a player who starts their career slowly (Smith, Kohli) cannot really get into the first quartile because of the 100% cap (which I guess is correct, again because of the purpose of the article). We do see some of the modern stars in the Top 15 bowling list though - cricketers like Ashwin, Yasir and Harris - who have had stellar careers already, however, they are bound to fall short in the long run. (Harris played 2 matches in SL, and none other in SC, getting most of his wickets against Eng and SA) In summary, it was a good read, but I am glad this metric is not used more often to rank players!
    [[
    You know what is the problem. In an article in which I have ONLY talked Consistency and have not made a single reference to the quality or value, you have yourself brought in these terms. Look at the batsmen top. We have at least 4 great players. So it is possible to score big and be consistent.
    Ananth
    ]]

  • POSTED BY Mohan Patel on | August 10, 2017, 13:23 GMT

    Sir , what about rahul dravid?? popular perception is that he was one of the most consistent batsman but i guess data amy not say that as he had a significant lean patch in his career

  • POSTED BY Ishan on | August 10, 2017, 13:00 GMT

    Hi Ananth, Great analysis. I loved your overall piece. However a couple of things: 1) Number of blocks: They will always have a negative effect on the final consistency. I think you could normalize for that. Some with just 10 spells (your definition) or 10 innings is likely to be on the extremes (first or fourth quartile) against people with longer careers. I guess Allan Border makes for an outlier and a great in a different sense (which is also a testament to your analysis). But I believe you could work on this angle. Maybe create segments beyond your thresholds of 3000 or 100 wickets.
    [[
    Not really. Pl see the reply sent to reader couple of comments back.The top-20 has 7 batsmen with over 150 innings.
    Ananth
    ]]
    2)An add-on: I know you are talking about pure consistency. If you could weigh the % consistency with block mean, it could lend a whole new dimension of how "valuable" was the player. Did he deliver often AND did he deliver big? The idea isn't novel and I'm sure you have your reasons not to do it.
    [[
    I have already replied to this. I do not want to confuse consistency and values. Delivering big is something else.
    Ananth
    ]]
    3)Analysis of Curves: Not for an article maybe but i think studying all the curves would be fascinating.

  • POSTED BY Guru on | August 10, 2017, 12:56 GMT

    I am not sure if this is addressing the question of reliability that well. This article is comparing one batsman's consistency in making 50 and another batsman's consistency in making 40. 90 percentage of 470 is better than 93 percentage of 390.
    [[
    Consistency should ALWAYS be measured only within the concerned person only. Else it does not become the correct metric.
    Ananth
    ]]

  • POSTED BY Arvind on | August 10, 2017, 12:56 GMT

    Ananth, 2 questions: 1] Rahul 'the wall' Dravid - where does he sit in the table? 2] It would be interesting to plot the Consistency Index as a function of number of innings/spells independent of the player in question. It will tell us a general trend whether playing longer has any impact on the consistency. Intuitively one would say 'yes', but a plot would be useful to see.
    [[
    Dravid is 101st, in the third quarter, with a CI of 85.9%.
    Probably no. The top-10 has a mix of 69 to 185 innings. I get the feeling that, barring the unfortunate sub-par ending of careers of some players, as they move on they probably are good enough to perform at the expected level, across 10 innings.
    Ananth
    ]]

  • POSTED BY Howard on | August 10, 2017, 12:43 GMT

    Why not just use CoV, as you said it gives similar results and is a standard measurement?
    [[
    Good question. Maybe 20% of the readers would understand CoV, another 20% would make efforts to 'wiki' CoV. The others would just move on. As I have mentioned elsewhere, I want my work to make sense to over 90% of the readers. But I have no problems in using something like CoV if nothing else was available.
    Ananth
    ]]

  • POSTED BY victoria on | August 10, 2017, 12:25 GMT

    Ananth, this is a statistical master piece - good enough to win you the 2017 Nobel Peace Prize, for statistical brilliance! I am blown away by the profoundness of the arithmetical intricacies herein presented. I think though, that "consistency" is just ONE of, but the most important element of a batsman's performance repertoire to be considered the greatest batsman of All Time. However, when all the other measurable comparison ingredients are added, and we are talking about really great batting of such exceptional epic standards to rival the greatest movie of All Time, we see 5 exceptional ones in our time: Don Bradman, Vivian Richards, Gary Sobers, Brian Lara and Kevin Pietersen; but because of their unmatched exploits, we know that either Bradman or Lara is the greatest of All Time. But I noticed that they both scored the same number of points in the consistency index as calculated (88.1); so how did you determine which of them is 52 and which is 53? Your work says it all - no BIAS!
    [[
    Thanks for the nice words.
    When I print these tables, I use up to 3/4 decimals and order the players perfectly. However, when I create the tables, I stick to one (rounded) decimal. Bradman's CI was 88.132 and Lara's 88.107.
    Ananth
    ]]

  • POSTED BY canthe3430335 on | August 10, 2017, 11:15 GMT

    Hello Anantha. Nice work. The blocks of 10 innings looks fine but every 10 innings may span home and away tests. A purple patch at a 10 innings home block may yield a player 100% (capped) but a 10 innings slump in away from home will damage the player's index badly (provided both are exclusively separated by chance). Did you make any adjustments to correct for home and away proportions in the 10 innings block?
    [[
    No, once the 10-innings block is fixed I did not want to make any adjustment. The 10 innings block should be a microcosm of the player's career. Then only will my analysis prove to be sound. Maybe a long time back the 10-innings would have been part of 5-Test series but nowadays it is likely to have Tests played all over the place.
    Ananth
    ]]

  • POSTED BY naresh on | August 10, 2017, 10:50 GMT

    LOL, I always liked Greig, but ......."if you torture data long enough, it will confess to anything".
    [[
    Data torturing or not, I suggest you look at Greig's distribution and be ready to be astounded.
    Ananth
    ]]

  • POSTED BY J.R. on | August 10, 2017, 10:26 GMT

    The true resolution of the paradox (what is the difference between the batsman who gets a pair followed by two hundreds and the one who gets 50 four times?) is that the more variable batsman is almost always more valuable. He contributes match-winning performances. On the other hand, the modal score for all batsmen is roughly 0, so when he gets 0, this is only what you would expect and probably not damaging to the overall team score (except where it contributes to a collapse). A high consistency index may not be characterising the most valuable attribute, as results with actual players confirm. The frustration of Joe Root at his low 50/100 conversion rate is the right reaction. A fine player like Root wants to be influential, not just reliable. There is also a problem with stats that span whole careers, in that a player may have clusters of form, while being merely reliable most of the time. An example would be Michael Vaughan, mostly mediocre but with purple streaks.
    [[
    One thing has to be understood. The capping compresses the high values into the mean value. As such batsmen with widely varying blocks of scores will not fare so well. But the purpose of this analyses is not to measure the value to the team but how consistent they have been. These two are not comparable. I have a fantastic measure analysing the Contribution of players to team causes. That would measure the influence.
    I agree that the top players should be influential but must also be consistent. Kohli is influential in many ways but he will be hurt by his recent inconsistency. I am talking of a sub-analysis within the 10-innings block. I am sure he would like to convert some of his sub-20 scores into 50+ scores.
    Root should be very happy to see that his sub-100 scores have been very valuable and more than made up for the poor quality of half the top order.
    Ananth
    ]]

  • POSTED BY Kartik on | August 10, 2017, 10:14 GMT

    Anant; Taking the total scores by blocks does not account for disparity between innings. Especially if a run of low scores has been off-set by higher scores due to a break (being dropped or different series).Instead, why not just take all the innings of a batsmen and plot it against a normal distribution? # of innings which show greater than one deviation from the mean (one both sides) can be used to measure the %inconsistency. Removes the necessity of blocks ....
    [[
    I have always thought out of the box and stayed off the normal statistical methods. It may be my greatest strength and possible weakness also. But my analyses have to be understood and appreciated by everyone who reads the articles.
    Ananth
    ]]

  • POSTED BY Dirk Laurie on | August 10, 2017, 9:45 GMT

    What about all-rounders? One might find that some usually deliver with the ball when they fail with the bat, whereas others do both well or both badly.
    [[
    The thinking has to come straight from ground up. I will think about it and have to find a suitable outlet to post the article.
    Ananth
    ]]

  • POSTED BY Grant on | August 10, 2017, 9:42 GMT

    What I want to know is how Chris Martin fared as a consistent "batsman."
    [[
    Now that this terrific request has come in, I will certainly do the required work and post the results. After all, the cutoff has to be lowered - almost to the ground !!!
    Ananth
    ]]

  • POSTED BY Grant on | August 10, 2017, 9:18 GMT

    I remember another analysis that Anantha did a few years ago on Cricinfo, I think it looked at exceptional performance in series. One of the outcomes was that Tendulkar, despite his tremendous average and number of runs, appeared in surprisingly few of these series, especially when compared to fellow uber-batter Lara, suggesting that Tendulkar was the more consistent, but Lara scaled greater heights. It is interesting that by this index, Lara is more consistent than Tedulkar.

  • POSTED BY Hitesh on | August 10, 2017, 8:40 GMT

    Hi Ananth. That's a great analysis. Can you also provide names with top consistency levels for each block of Batting average (30-35, 35-40, 40-45, 45-50 and 50+) and bowling average (30-35, 25-30, 20-25, 15-20 and 15-).

  • POSTED BY Paaji on | August 10, 2017, 8:29 GMT

    Hello Ananth, Good to see you back. Nice work. And it makes sense not to penalize players for going over their average. Just a thought w.r.t the bowlers. Wouldn't it be better to look at wickets/run (bowling average essentially) rather than just wickets? The batting is taken care of to a good extent with a 10 innings split but just saying the bowler produced 30 wickets wouldn't have much meaning unless it is known if it is for 200 or 500 runs given.
    [[
    I have gone on the cardinal principle in measuring Test bowlers that 5 for 75 is better than 4 for 30. A generalization, I agree, but let us agree that capturing wickets is a sure way to win Tests.
    Ananth
    ]]

  • POSTED BY James C Birbeck Dar on | August 10, 2017, 7:09 GMT

    I started reading this, and was a bit surprised not to see Mitchell Johnson among the least consistent bowlers! Bottom quartile? If I recall correctly, Michael Manley in his History of West Indies Cricket, says Valentine lost his zip. Greig's numbers suggest he might have been past his best, though his "retirement" from test cricket wasn't a personal choice!
    [[
    Surprisingly, Johnson has been reasonably consistent. His CI is 89.5%, he is in 71st positin and second quartile. His below-par blocks have been reasonably well handled. There is just one block below 50%. The others are above 70%.
    Ananth
    ]]

  • POSTED BY udendra on | August 10, 2017, 7:09 GMT

    Since their names are not mentioned, I'd like to know Kumar Sangakkara and Sachin Tendulkar's standing in this index.
    [[
    Tendulkar has been covered in the article.
    Sangakkara is quite inconsistent. His CI is 84.7%, he is in 129th position and in the third quartile.
    Ananth
    ]]