An over-simplified definition for consistency.

I think your analysis is completely.. Skewed..Stats or no stats how can you leave out bowlers like Marshall, Ambrose, Warne and Mcgrath? .. If your math does not take them in to account... ur analysis is scrwed and skewed... get real sir... Thanks

32.Warne S.K RLB Aus 145 250 232 92.80
39.Ambrose C.E.L RF Win 98 164 151 92.07
50.Marshall M.D RF Win 81 147 134 91.16
85.McGrath G.D RFM Aus 124 233 206 88.41

The above bowlers did not fall into the top 25 whose details were displayed. The analysis is not "scrwd" or "skewed". In this particular measure there are other bowlers who are ahead of these modern greats. That is fact. It does not also reduce their greatness an iota.

Ananth

where is shane warne in the list?

Pl see response to Mr.Ganesh.

Ananth

the real stand out on that list seems to be allan donald, along side murali and bedi

unfortunately for bond and jones, they are always injured...they could have been the two best quicks going around...

mervyn dillon up there is a bit of a joke
--------------------------------------------------------------------------
Not just Dillon, but also Miller and Tauseef. But it gives us a chance to look at these lesser bowlers in a new light.
Ananth

In response to Mr. Ganesh's observation: I am neither for or against the analysis put forth by Ananth Narayanan. However, Mr. Ganesh, if your figures are correct, the four bowlers you have listed have a consistency Index of less than the top 25 listed by Ananth Narayanan. This is precisely the point Ananth Narayanan is trying to make. The 25th Man (MacGill) has a consistency index of 93.42 while all four greats on your list have less than that. I hope you have got the stats. understood now. Coming to think of it, this stats seems interesting !

u cant left out the great waqar younis, glenn macgrath and shaun pollock when u talk abt consistent bowlers. urs observations r incomlete. thnx

I agree with the above commenter that this is too simplistic. For instance, your method takes Shane Warne's debut figures of 45 overs, 1/150 and gives that a tick because he took a wicket. But his spell of 10.2 overs, 4/15 against NZ in 2004/5 is ignored.
--------------------------------------------------------------------------
I am sorry that I have to correct you. His 4-15 will not be ignored. If you read the article carefully, this spell, even though it is below 11 overs will be included because wicket(s) were taken.
Ananth

You're right that statistics can't lie. But they also don't show the whole picture. Bowling is about partnerships. It often takes two bowlers (Murali apart) to create pressure and therefore wickets and its not necessarily the more dangerous bowler of the pair who gets the wicket.

In response to Ganesh, this is not a measure of bowling brilliance. Shame Warne et. Al. are undoubtably outstanding bowlers, however they did have "off days". Shane Bond on the other hand has had less off days. This doesn't mean he is a better bowler than Warne, it just means he is more consistant. I think this is an interesting statistic, and it would be good to see if Simon Jones can maintain this if he ever makes a comeback to the England side.

Interesting analysis Ananth. The 'greats' which do not make the top 25 are because they have a higher proportion of wicketless spells (out of their qualifying spells). The fact that they bowled a lot could be a factor i.e. the more spells you bowl the more you are likely not to hit the mark. I would however be interested to see the stats of the 'greats' i.e. Waqar, Warne, Garner, Marshall, Holding, McGrath, Kapil Dev.

I must argue with your definition of a successful spell. If a bowler bowls with an economy rate considerably lower than the innings run rate, he is helping to put the batsmen under pressure and conceivably helping bowlers at the the other end get wickets. Also, if his team's score is being chased, and the opposition has only a limited number of overs, his spell could be crucial in securing a draw. How about factoring this in your calculations, difficult though it may be?

PS:Even the Bapu Nadkarni spell you talk about was, in a way, 'successful'. Had he had not bowled THAT economically England might well have won that test.

I think the people who seem to be coming up with names that have been "left" out of the list are misunderstanding the point of the analysis, which tries to compare bowlers according to the statistic put forward by Mr Ananth. If they are left out, they are left out by the statistic, no one is implying they are not great bowlers. Although personally, i think we need to consider the number of matches played as well, since we might find share warne has been far more consistent over a greater period than someone like Simon Jones who might not have such flattering figures if he played as many test matches. Also someone like Murali, whose teammates might not be as effective as Warne's (Mc Grath and co.) has a higher chance to take more wickets in a spell. Of course, that being said, its a lot tougher to take wickets when your bowling "partner" is being creamed all over the park.

The statistic does not seem to provide much information. If a bowler were to consistently take just 1 wicket in each innings and all his spells are relevant (obviously 1 spell per inning), would that count as 100 %success ?

My only issue with this analysis is the length of spells under consideration. 11 overs as a benchmark is rather long for fast bowlers. I think that you should reduce it to 7 for seamers and retain the figure of 11 for spinners.

correct me if I'm wrong, but it looks like this is a measure of consistency for consistency's sake - it doesn't matter if a bowler gets one wicket or 5 wickets in the same innings - the performances would be weighted equally. This in turn would mean that as long as you kept plugging away and got a wicket at the end of it, you would be considered on par with someone who ran through the innings. For example, since Merv Dillon hardly had any fellow bowlers in the WI team capable of taking wickets, he had a pretty good chance of getting at least one wicket at some point in an innings as long as he kept bowling, thus resulting in a higher 'consistency' score. On the other hand, bowlers like Marshall and Holding, or even Waqar Younis, would be at a disadvantage because any one of their colleagues might finish off the opposition first. Perhaps you could add a column to the above table on strike rates to differentiate the strike bowlers from the stock bowlers.

Are you saying that a bowler, whose average figures look like the following -> 20 overs, 2 maidens, 98 runs, 1 wicket

I think you might need to introduce many more variables to accurately capture this statistic.

Anand.. Good one this.. But don't you think consistency should also be a function of the number of matches played by the player. Do you think its fair to weigh achievers in 30 innings and achievers in 150 innings on the same scale..

It would make for interesting reading if factors such as caught behinds, caught in slips, lbw's etc. could be taken into account. I know the analysis wouldnt be very objective.
Since we are talking of test cricket only and the strategies adopted are fairly straightforward.
Can we not normalise all different ways of getting out and create a scaling of sorts?
Different criteria for fast bowlers and spinners. I fear comparing the two would be tougher.

Am not surprised to see the occasional average bowler come up on the list - especially those who havent played in too many tests (so that the luck factor in playing only on a few bowler-friendly pitches can play in their favour). In fact, I would have expected to see folks like Hirwani, Bob Massie and Bob Holland on this - people who had phenomenal strike rates in their first couple of tests and then faded away. Or did you have a minimum no. of tests cutoff? If you didnt, I would suggest having that - because true consistency is consistent wicket taking while playing multiplied by consistent playing over a span of time and across different geographies. I suspect your warnes, mcgraths, imrans, hadlees etc will then show up and the millers and dillons wouldnt

First of all thanks to Ananth for an extremely simplified solution. And I agree with his explanation about the 4 greats as well.
I do have a suggestion. We know it is very easy to get tailend wickets. And bowlers like Murali frequently get cheap tailend wickets. It would be really interesting to see the same analysis performed for top order wickets only. That would show the real performance of the bowlers. For e.g: A guy like Jason Gillespie who had close to 80% top order wickets comes to my mind.

Though it complicates calculations a fair deal, i wonder if you could somehow include strike rate in the equation.

The stats are interesting and there are advantages to having a simple model, but one cannot simplify reality to such an extent that the model ceases to provide any insight.

The model treats 4/15 off 10.2 overs and 1/150 off 45 overs equally. Why not divide the total number of wickets a player has taken by the number of relevent spells he has bowled and multiply this number by the consistency index?

Anantha - I too am not convinced with your definition of consistency - I believe it is too simplistic. For example, you have not accounted for the fact that cricket is played on different continents and how does each bowler compare with himself and others when playing at home and away. The other thing that you have overlooked is longevity of a bowler - for example, if Narendra Hirwani had only played one test match, his consistency index would have similar to Bond's or may be even better (don't know if that is possible), while Shane Warne's index would have been far lower than what it may be now. Besides, to get a wicket, there are other parameters involved as well - like the fact that a bowler my produce an edge, but if the fielder doesn't catch it, then consistency goes out the window. Please let me know your thoughts on my reflections.
Thanks,
Dhaval Brahmbhatt

Interesting analysis. But I have just a simplistic thought: what if "weights" are attached to each spell based on statistical data to the number of wickets taken in that spell (or may be even better - the strike rate for that spell, or for that matter, even the ratio of wickets captured by the bowler to the total wickets falling in that spell, since that will "sort of" portray the pitch conditions. You are likely to have lower # of total wickets on batting paradises, but higher # in greentops/turners). This will sort of shift the definition from "a minimum level of consistency" to "consistently good".

BTW, I assume that when you talk of a spell, it's not the "spell" that commentators talk of (i.e. the bowler bowling unchanged from one end), but the entire innings instead.

The stats interesting. Of course, real 'consistency' means bowlers can bowl week in and week out in all sorts of conditions without getting injured every second match. My only other thought is... imagine the differnce for England and New Zealand had Bond and Jones been fit for long periods of time! Wow! They would have been far more successful!

i thought it was quite interesting. some people are a little over defensive of there heros. he wasnt saying those bowlers were the best by any means some were quite average but its interesting to see stats like this. even the greats can have bad spells guys some more than others but then show us their greatness in the next game(or innings) by taking 7 wickets

Interesting analysis. We should also find out the length of each "spells". We know that Murali or Kumble can bowl longer spells (even 20 overs unchanged) whereas McGrath, Ambrose and Waqar used to bowl shorter spells.
I would go with the traditional definition of strike rate - which is how many deliveries do you need to bowl before you get a wicket.

In the top 25 only 7 bowlers have played more than 50 matches. So, the list is more representative of the bowlers who could have been in the ranks of greats but missed it somehow.
And yeah those bowlers who have played more than 50 matches are really great and it is a further confirmation of thier greatness and the amount of advantage they bring to the team by their presence

Maybe because I live in a country where I cannot see live first class cricket I have missed something obvious. I don't understand your comment that "Bapu Nadkarni in 1964 in Chennai with a spell of 32-27-5-0 would today be booed off". If McGrath or Murali pruduced bowling figures like this wouldn't they be applauded? I would appreciate it if you could expand on the reasoning behind your comment.

I don't think this is a good indicator of consistency. Taking wicket in a spell is counted even though the bowler may get hammered in the spell. A better indicator of consistency is the bowling average of each innings or test match and then determine the standard deviation of the bowlers bowling average. The lower the standard deviation them more consistent the bowler (This could mean consistently good or bad) perhaps a system that combine average and standard deviation (like average/st.dev) would be much better then your system as this would determine how often the bowler perform well in a career or whether it is inflated by a few good performances.

A suggestion to Ananth: I think to make this a bit more meaningful (or appease the crowd:P ) it would make sense to have measure of outliers in a bowlers career (wrt consistancy acc to your def) and then fine tune it to make more sense (ignore 3 divs from mean). This will not effect Shane Bond but will boost those who have had a few bad games here and there in a long career.

This will be consistent with your definition of consistency. As right now if someone gets 9 wickets in 10 overs and then no wickets in next 8 matches will get a full score but if you remove the outlier (9 wickets) then the real consistency will show.

Hi Ananth,

Good job in working the above out and it is an interesting method. And I would make a point to all the readers is that the operative word is consistency defined by the method devised by Ananth here and not Greatness.

Greatness is a very subjective concept in sport and cannot be easily defined or analysed. And yes statistics fall short of accomodating finesse which is subjective. Stastistics will not account the ripping Shane Warne Leg Break which deceives even the most accomplished Batsman as well as the confidence and Gumption of a Sachin Tendulkar dancing down the pitch to the same Bowler and depositing the Ball into the Stands. Doesn't make warne less great or sachin any greater.

No Grandpa will tell his Grandson how much per wicket Shane Warne averaged but will always demonstrate somewhat exaggeratingly how much the Ball turned to a wide eyed little boy with a red ball in his hand.

Ananth: I think you started arguing on the basis of strike rates but never used it. Let's say a bowler A bowls 12 overs for 30 runs, and another bowler 'B' bowls 50 overs for 150 runs and 1 wicket, then you have ignored A but taken B into account. Probably you should chosen those spells as successfully where the strike rate was less than 66.

Ananth concedes upfront that this is not a 'perfect' assessment, but an attempt to measure the 'consistency' of the bowler. Besides strike rate, economy rate is equally important. So Ananth can perhaps identify benchmarks for economy rate as well. Once that's done, we can define a 'successful spell' as one in which BOTH the benchmarks are equalled or bettered (one can allow tolerances too). The definition of a 'relevant spell' could be modified to include all those which satisfy at least one of the following conditions:

1. The actual no. of deliveries bowled is equal to or more than the benchmark
2. The actual no. of runs given away is equal to or more than the benchmark no. of runs (based on economy rate benchmark) for the benchmark no. of deliveries

This could provide a more comprehensive measure of 'consistency'. While ensuring that the no. of wickets are taken into account implicitly, this method ensures that freak feats (like that of Michael Clarke vs. India) are excluded.

This doesn't tell you jack about the quality, or lackthereof of the bowler. The definition of consistency is misconstrued, so is that of spell. Spell and innings are different things in cricket.

This article is a total waste of time.

'How does one measure a test bowler's consistency?'

Well, Mr Narayanan, it turns out there are these things called averages and strike rates, which indicated how much wickets cost for bowlers in terms of balls and runs. This gives an indication of how often he takes wickets, and how efficiently he does so, which, to my mind, could be termed 'consistency'.

By the way. I didn't come up with these straightforward measures myself. They've actually been in use for some time.

Consistency is not taking at least 1 wicket for a huge cost in runs. Consistency is landing the ball where you want it, not going for 100 runs but beefing the average by virtue of 5 tail end wickets. Talking of wickets, most come through batsmen error anyway.

Interesting post, Mr Narayanan. It suggests a very new measure of consistency... I guess this is more "reliable" bowlers - that is, the bowler who is capable of consistently picking up a wicket or two (or more). However, does this mean a bowler is "consistently" GOOD? I don't think so. In fact, it just means that a bowler is unlikely to go wicketless in a spell. Thus, a bowler can, in fact, waste plenty of overs and take one wicket... this does mean they are "consistent" in a sense (the sense that they they can get a breakthrough), but it doesn't necessarily mean they are they are good bowlers. Perhaps to see who has been consistently GOOD, it would be useful to see in what percentage of spells a given bowler has picked up more than one wicket (e.g. in what percentage of spells has he taken 2 or 3 or more wickets?). A single wicket doesn't make a good bowler, but the term "consistent" is ambiguous in your context (it doesn't mean consistently GOOD).

It's difficult to see this analysis as anything other than fluff.

For example, the definition of 'consistent' here is worthless. You have settled on an arbitrary definition and have not provided any meaningful justification for it.

You could just as well have defined consistency as 'tending to concede more than four runs an over' and presented your findings in an equally sober fashion.

"Spell" is defined incorrectly. A spell is a uninterrupted series of overs and a bowler will usually bowl several of them in an innings. You lose credibility when you play around with accepted terms in this way.

This analysis could only have merit as part of a larger study of bowlers and bowling, taking into account various other obscure and arbitrary indices of achievement or otherwise.

Still, you've got people thinking, so well done on that score.

The simplest way to measure consistency is to see standard deviation on the average strike rate for the bowler, and batting average for the batsmen. All that is needed is a table of how many balls were bowled by a bowler (irrespective of whether this spanned over multiple inningns or test / ODI matches) between every wicket taken. Then average strike rate and standard deviation can be computed. Much simpler, cleaner, easy to understand. Also remember, a bowler with average strike rate of 120 balls but std deviation of 2 balls is more consistent than a bowler with strike rate of 30 balls and std deviation of 20 balls.

This is just one way of looking at bowlers and I think the trick is not to confuse consistency with the ability to win matches by taking a glut of wickets, total value to team or any other such measure. How would you measure these things? There are probably ways of assigning a value to all of them and the combination of these values would be interesting.

I'm not sure about the merit of discarding of short wicketless spells.

I really think you should do wicketkeepers, aiming for a rank on keeping alone and combined with batting. Start thinking about it in detail and I give it a couple of hours before you're calling the men in white coats*

---

*Best to specify in advance whether you want trained healthcare professionals or a group of umpires in a Ford Transit.

In 'The Cricket Statistician', issue 24, December 1978, you will find an article [slightly mutilated by the editor who deleted the E v RW series 1970] by Henk Volten RA: A Rating of Test Bowlers - the unused resources of Trueman and Snow.
The main indicator I used [derived from US baseball]was EWA [Earned Wicket Average: wickets divided by matches played].
In the group with over 50 Tests at that time Alec Bedser came first [EWA 4,62] ahead of Truemean [4,58] and Snow [4,09] but overall SF Barnes was way ahead with an EWA of 7.00 before Grimmett [5,83] and it pleased me to find them both in the top ten of the current table.

Volten, Sassenheim, The Netherlands [Feb. 10, 2008]

I'm interested in a statistic where the proportion of wickets a bowler has is broken down in terms of how many of those wickets were obtained in 5-wi vs not 5-wi. This would also be a kind of measure of consistency, as bowlers with several 5-wi but few non 5-wi wickets would be more dependent on favourable conditions to pick up wickets. Obviously someone like Murali would be an exception.

Yup the traditional definition od strike-rate-- that is the bottomline.
Also, it was interesting to see the average runs per wicket by places, such as Australia, India, Pakistan, and so on. The notion that wickets are harder to come by in the subcontinent doesn't seem true. However, lots of wickets in the subcontinent are bought thru spin-- it would have make sense how pace-bowlers have fared in the flat subcontinent tracks as well. It will make for a good analysis-- whether a bowler like Mcgrath without much swing would be as effective in the subcontinent and which sort of 'pace' bowlers as a whole have been most successfful?
It can also make an interesting comparison to how pace bowlers have fared in Australia, South Africa, and so on-- just a thought.

I also agree the term "spell" has been used wrongly here. But some really good insights have come out of Ananth's analysis. Amazing how some still see that "its all so easy for Murali with no threat from the other end". They always forget that unless you have good bowlers at the other end plugging away and putting the pressure, keeping the batsmen quiet, or making them jumpy it is very difficult to take wickets on your own consistently. Murali aside, I can only think of the great Richard Hadlee who was able to take wickets on a regular basis nearly all alone. But remember in Ewan Chatfield he had a run miser of a partner who would plug the runs at the other end leaving the masterful Hadlee to pluck the bounty. Murali on the other hand, except when Vaas is firing, is all alone.
As for the Bond case it merely reflects how good a bowler he is and the possibilities of becoming a great if not for his ever present injuries.

Congratulations. It seems people don't like the analysis because they don't like the conclusions. You have certainly set a challenge, as it is all very well to criticize, the test is come up with something better. While I may not agree with the "results" the analysis is excellent. Well done.
==================================================
Rob

Many thanks for the kind words. My purpose has always been to inform and also make people think. Bouquets without brickbats are somewhat hollow.

Ananth

Shouldn't Simon Jones be no.1 as he has played more tests than Shane Bond?

interesting stats...the first scientific proof of mc grath's method of stifling the batsman...that accounts for his long periods of going without a wicket, and hence the lower score on the consistency index!!

The term 'The most consistent in test cricket' should ensure no 'flash in the pans' get into the list, 17 test matches is barely a statistic these days. 50 Test matches is more of a yardstick to measure consistency.
==================================================
A very good suggestion. I might very well do a follow-up post with a higher cut-off level;. Thanks.
Ananth

This is very interesting and thank you for your work, but i was wondering if you could make a new list for those who bowled a lot more, ie at least 100 relevant spells, just to show the longevity of their consistency.

Interseting way of looking at it, Sir.

Having seen this brilliant wave of analysing a bowler, maybe next time you can take cue form what a guy was suggesting above that bowlers hunt in packs, meaning pressure from both ends.

Can there be made a statistic where it can be known that a bowler got more wickets because the pressure on the batsmen was created from both ends simulatneoulsy. For eg; A Shane Warne would have more chances of capturing wickets because Mc Grath or Gillespie would be on the other end with a tight leash, whereas Murali would have a Santha Jayasuriya on the other end bowling that is.

So can an index be made for that??

Ckt is all bot flow of runs. Fall of wkts is a event in the process of controlling runs. No statistics will be accepted if it doesn't include the runs conceded as a variable in analysis. Im happy to see such a simple commons sense statistics to analyse bowlers. I appreciate it. But it should include the variable "R-U-N-S".
Moreover, the "cut-off" margin is also a big concern for many readers, even i agree. Perhaps, the author should succumb the temptation to add "License-to-Kill" man!

Any definition of consistency is going to be subjective. You have defined consistency one way and the analysis is based solely on that definition, so criticism on that point is baseless.As a West Indian very interesting, as the main criticism of Mervyn Dillion and the reason for his downfall has been a lack of consistency in effort.

I agree with one or two of the previous writers that you should apply weights based on the number of wickets taken. Eg. (Success Spells x No. of wickets taken by bowler in spell/Total no. of wickes taken by all bowlers in Spell)/No. of relevant spells. This along with a higher cut off level as you have agreed with will improve the analysis.

Hey this is not strictly relevant to this post, but would you guys be able to do an analysis of who had the most 'Golden' arm amongst part-time bowlers - most wickets@best strike rate for players who've bowled in

ananth, i had posted a comment earlier, but it doesnt appear here.maybe i didnt express clearly what i was trying to say, which i will do now. i disagree with ur analysis bcoz i disagree totally with ur definition of consistency - it is too simplistic and does not sit well with established notions of consistency - accuracy of bowling, keeping the pressure on the batsmen, landing the ball in the right spot most times, creating opportunities for wickets whether for oneself or others, not giving the batsmen breathing space etc. where is all this taken into account. by this definition ajit agarkar wud have been a consistent bowler during the intial phase of his one-day career - but would anyone classify him as consistent? would any team have devoted a lot of time to analyzing his bowling? did his bowling ever create pressure for the batsmen regularly (which incidentally shud be one of the mandatory things a consistent bowler should do)?

What is interesting is that none of the bowlers that Bradman faced do well for consistency, putting the lie to the idea that Bradman was a great batsman. Fact is, the bowlers he faced were amateurs and inconsistent. Where is Larwood, Bradman's great nemesis? If Larwood doesn't rank so good, then Bradman would have been destroyed by the other bowlers on this list. Again Ananth, you need to be careful or else your statistical analyses could get you into trouble with all those ignorant Bradman fans.

I have not read through all the posts and at the risk of repeating what someone else might have mentioned already, here goes:

I like the analysis. I suggest embellishing it with a suitable weightage for certain variables like quality of opposition, home vs away, comparing the results with the overall average of all pace/spin bowlers as the case may be and skimming off the low perfs, total team score of the opposition in that innings (to show how effective the bowler was in batsman-friendly conditions), etc.

Thanks and best regards.

Without doubt Bond when fit was the best fast bowler of this era before he went to ICL.In the 16 tests he has played he has been amazing and for those skeptical of his ability should have a look at his S/R.It is a shame he didn't play more than 16 tests.

Chris Smith mentions that bowling is about partnerships. I've always thought it would be interesting to see an analysis of whoever is bowling at the other end - ie Nathan Astle is never going to rip through a batting line up but does the pressure he creates mean that wickets fall at the other end?

A similar analysis could be conducted on batsmen to look at who brings out the best in their batting partners.

hi,
i realy appreciate you for the hard work you have done, but to me its not acceptable because we cannt left out the greats like Glenn McGrath, Waqar Younis, Shane Warne Shaun Pollock, Imran Khan, so all i would like to say is may be you need to redefine your definition of consistant bowlers, or if it is possible then separate the terms " Consistant " and " Successful " bowlers and publish your stats again. i would love to read any such article by you as it seems tou are pretty much in to it.
Good work though.
Thanks & Regards
Faisal Ch Mehmood.
but