July 16, 2012

Developing a bowling value index

47

Muttiah Muralitharan bowled one third of the team balls and maintained an excellent strike rate
Muttiah Muralitharan bowled one third of the team balls and maintained an excellent strike rate © AFP

While looking back statistically over the career of the recently-retired Brett Lee, I mused over ways of assessing the value of bowlers to their team. Lee, of course, was a fast bowler best used in short spells, and was not one who one would use to plug up an end and bowl all day. He thus had quite a good strike rate (10th in the list of 35 Test bowlers with 250 wickets or more), but for limited times of the playing day only. He only bowled 21.25% of the balls bowled while his team was fielding, ranking him only 25th in the same list of 35 bowlers. Obviously, his value to his team would have been enhanced if could have bowled more while preserving his strike rate - but there is a risk, of course, that the strike rate diminishes if a bowler is asked to bowl more.

I am thus inching towards a statistical measure of the value of a bowler, V, that is a function of Strike Rate (balls per wicket), i.e., V = k(SR)

where k is a constant representing the amount of bowling a bowler does.

The simplest way to calculate this is to divide the proportion of overs that each bowler bowls for his team by the strike rate. Thus for Lee, V = 21.25/53.33 = 0.398

The table below displays the value of V for each of the 35 bowlers with 250 Test wickets or more:

* % tbb represents percentage of teams balls bowled by the bowler

Value of a bowler (V) - bowlers with 250-plus wickets
Player Wickets SR % tbb V
Muttiah Muralitharan 800 55.05 33.31 0.605
Dale Steyn 272 40.94 21.73 0.531
Richard Hadlee 431 50.85 24.86 0.489
Shane Warne 708 57.49 28.00 0.487
Waqar Younis 373 43.50 21.09 0.485
Dennis Lillee 355 52.02 25.19 0.484
Malcolm Marshall 376 46.77 22.62 0.484
Glenn McGrath 563 51.95 24.36 0.469
Joel Garner 259 50.87 23.41 0.460
Allan Donald 330 47.03 21.50 0.457
Anil Kumble 619 66.00 28.77 0.436
Wasim Akram 414 54.65 23.58 0.431
Danish Kaneria 261 67.80 29.06 0.429
Fred Trueman 307 49.44 21.11 0.427
Curtly Ambrose 405 54.58 23.17 0.425
Imran Khan 362 53.75 21.88 0.407
Courtney Walsh 519 57.84 23.18 0.401
Brett Lee 310 53.33 21.25 0.398
Harbhajan Singh 406 68.11 27.09 0.398
Craig McDermott 291 57.00 22.34 0.392
Shaun Pollock 421 57.85 22.47 0.388
James Anderson 267 57.29 22.02 0.384
Makhaya Ntini 390 53.42 20.50 0.384
Jason Gillespie 259 54.96 20.63 0.375
Ian Botham 383 56.96 21.29 0.374
Bob Willis 325 53.41 18.66 0.349
Bishen Singh Bedi 266 80.32 27.75 0.345
Zaheer Khan 288 58.05 19.09 0.329
Brian Statham 252 63.71 20.77 0.326
Daniel Vettori 359 78.98 25.59 0.324
Derek Underwood 297 73.61 23.85 0.324
Chaminda Vaas 355 66.02 20.71 0.314
Kapil Dev 434 63.92 20.05 0.314
Lance Gibbs 309 87.75 26.85 0.306
Jacques Kallis 276 68.67 12.64 0.184

Murali was clearly, under this measure, the most valuable bowling commodity to his team in Tests - not only does he possess a strike rate more in tune with that of strike fast bowlers (compare it with Kumble, Bedi and Vettori, for example), but he bowled an extraordinarily high percentage of Sri Lanka's overs in Tests in which he took part, bowling one over in three.

It would appear, too, that Dale Steyn is forging a career that will put his name up in lights alongside the great bowlers of yesteryear. No one can question Hadlee's high index, given his value to an otherwise unexceptional New Zealand team, while the high positions held by Warne and McGrath formed the basis of Australia's period of domination in the first decade of this century. Similarly, Waqar and Wasim formed a potent Pakistani partnership as did Marshall and Garner for West Indies. The recently-retired Lee is where you might expect him, about half-way down this list, just outside the indices of the really great bowlers. The index of Kallis is significantly below those of others in this table, just emphasising that there are limits to what even great all-rounders like the South African can do. An index of around 0.4 seems to separate the really great bowlers from those who were perceptibly less so.

The following table gives the same results for those Test bowlers with between 100 and 249 Test wickets (min V = 0.4):

Value of a bowler (V) - bowlers with 100 to 249 wickets
Player Wickets SR % tbb V
George Lohmann 112 34.20 27.00 0.789
Sydney Barnes 189 41.66 30.81 0.740
Charlie Turner 101 51.28 33.72 0.658
Charlie Blythe 100 45.46 27.25 0.599
Saeed Ajmal 122 61.32 31.98 0.522
Bobby Peel 101 51.64 26.14 0.506
Mohammad Asif 106 48.77 23.28 0.477
Colin Croft 125 49.32 23.23 0.471
Clarrie Grimmett 216 67.19 30.84 0.459
Hugh Trumble 141 57.44 25.65 0.447
Saqlain Musthaq 208 67.64 30.09 0.445
Stuart MacGill 208 54.02 23.98 0.444
Fazal Mahmood 139 70.75 30.66 0.433
Bill O'Reilly 144 69.61 30.07 0.432
Graeme Swann 188 57.73 24.47 0.424
Darren Gough 229 51.62 21.63 0.419
Michael Holding 249 50.92 21.30 0.418
Terry Alderman 170 59.89 24.79 0.414
Shoaib Akhtar 178 45.75 18.66 0.408
Andy Roberts 202 55.12 22.40 0.406
Alec Bedser 236 67.45 27.40 0.406
Andy Caddick 234 57.94 23.27 0.402
Ian Bishop 161 42.25 20.92 0.400

Lohmann and Turner bowled in conditions that are rarely replicated today, small innings totals allowing them to bowl a greater proportion of overs, and with a more potent strike rate. Barnes is confirmed as the genius he truly was, while the index for Saeed Ajmal is interesting, and may in time, if continued, allow him to be classified as an all-time great.

The bottom five in this list are also of interest:

Bottom five in the list (100-249 wickets)
Player Wickets SR % tbb V
Trevor Bailey 132 73.58 14.72 0.200
Garry Sobers 235 91.91 17.84 0.194
Ravi Shastri 151 104.31 18.60 0.178
Ray Illingworth 122 97.82 16.82 0.172
Carl Hooper 114 120.95 13.22 0.109

At the risk of destroying the reputation of an icon, the placing of Sobers at this end of the table really does suggest that history may be overrating his contribution as a bowler. The jury is still out on whether he or Kallis has been Test cricket's greatest allrounder - but that is a topic for another day.

The complete list of the statistical measure V for all bowlers can be downloaded here

Comments have now been closed for this article

  • John on July 25, 2012, 11:55 GMT

    Another really interesting article, thank you. Chatting about it to my dad yesterday, also a big cricket fan, he wondered about the impact of bowler at the other end. In his words; "but does it show what impact a very accurate bowler at one end has on the wickets taken at the other." We were not immediately sure how that might easily be measured (value of bowler X as a whole v value in a particular partnership in some form pehaps) but thought it worth a comment!

  • Noah on July 18, 2012, 17:06 GMT

    One factor that your equation does not take into consideration is the quality of the team that any particular bowler plays for, and therefore their bowling percentage. For example had Kallis played for Sri Lanka, a side without the fast bowling quality of Ntini, Stein or Morkel, he would have bowled far more that 11% of the total. The same is true for non all rounders as well, Jason Gillespie for example would have bowled more if he was born in New Zealand and didn't compete with Warne, Lee and Mcgrath. It is amplified by 5 man attacks with players like Steven Harmison not making the list as a result of sharing the workload by virtue of being part of a team that had 5 bowlers due to the presence of an all rounder.

    Otherwise very interesting

    Ric: I think you are just confirming what I formulated: Kallis in a hypothetical Sri Lankan team would have had a higher value of V, and hence would have been more valuable to them than being the fourth or fifth bowler in a South African outfit.

  • Waspsting on July 18, 2012, 12:35 GMT

    I rank Sobers the 2nd greatest batsman ever.

    Didn't mean to give the impression i was comparing Sobers the bowler to Sehwag the bowler. kind of like pointing out if a bowling all-rounder had the same batting average as Ajit Agarkar - a legit point made, slightly 'cutely' (i.e. ironic, not particularly accurate but still legit)

    I think of him more as "One of the greatest batsman ever, and he could be a useful bowler" rather than the usual "undisputed greatest all rounder ever"

    @Jeff - telling stats about Gibbs and Sobers. Kind of supports a thought i had about a spinner being unnecessary if you have 4 great pacemen on the previous article on this blog

  • Gerry_the_Merry on July 18, 2012, 11:50 GMT

    My top picks (only last 40 years, since 75% of test cricket has been played in last 40 years) 1. Ambrose 2. Marshall 3. Lillee 4. Holding 5. McGrath 6. Hadlee 7. Imran 8. Wasim 9. Roberts 10. Gillespie/Waqar/Steyn/Donald 10.

  • Harsh Thakor on July 18, 2012, 8:53 GMT

    My list in order in terms of performance of paceman(over 250 wkts) are 1,Hadlee 2.Imran 3.Lillee 4.Marshall 5.Mcgrath 6.Ambrose 7.Waqar 8.Wasim Akram 9.Dale Steyn 10.Fred Trueman

    Ambrose was arguably,the best match -winner amongst paceman particularly in the 4th innings as he showed against England in 1994 at Trinidad and South Africa at Barbados in 1992.Mcgrath.and Marshall were the best match-winners in terms of percentage of games won.

  • Harsh Thakor on July 18, 2012, 8:45 GMT

    Imran Khan,is to me too low in your ranking,who had the 2nd best match-winning average and best performance by any paceman in his hey day in average and strike rate.Above all Imran was an all-time great match-winner who was a better performer than Wasim Akram if you asses his role in shaping wins in test matches and series versus the best opposition,be it England,Australia or West Indies.Marshall and Mcgrath bowled for top quality attacks and teams while Wasim Akram was hardly supported by the batting side.In his peak years Wasim performed better than Mcgrath .Overall to me Lillee and Marshall were the greatest and vie for the top spot,with their great competitivenes.What goes against Hadlee was his inability to be outstanding on slow tracks like Wasim Akram or Imran Khan.Waqar Younus was not as effective as Wasim against the best opponents like Australia,S.Africa or England and capitalized against weaker teams

  • Harsh Thakor on July 18, 2012, 8:34 GMT

    A very good effort Ric.I praise your ranking of Sir Richard Hadlee ,who to me on pure statistics is the best pace bowler ever.He has the best match performance analyis by any pace bowler ,more 5 and 10 wicket hauls,the best average in matches won,and has the best relative performance by any pace bowler in his peak,capturing 330 wickets in 60 tests from 1978-1988.Above all he bore the brunt of a weak attack.In the statistical light Imran Khan would come 2nd,considering that in his peak years from 1980-88 he had the best strike rate and average of any pace bowler of the modern era and was the best performer against the mighty West Indies.Howevever he also missed years from 1983-1985.3rd,I place Lillee.Sadly ,Lillee's scalps in Packer cricket and against rest of the world are excluded,otherwise he would have had 458 wkts.in 89 tests.To me, Marshall and Mcgrath are almost equal at 4th and 5th place but I give Marshall the edge.They were assisted by great bowlers and teams.

  • anand on July 18, 2012, 8:01 GMT

    i think your formula is heavily skewed to the percentage of team balls bowled, since that dramatically increases or decreases the net value.

  • Ananth on July 18, 2012, 3:59 GMT

    Ric: Perfectly acceptable, Ananth, and probably more encompassing in that it explicitly includes runs conceded. As I explained to another commentator earlier, I have implicitly assumed runs conceded in the bowling proportion in an attempt to minimise for simplicity's sake the number of variables involved. Ric: What I have suggested does not add any complexity at all. In fact the Bowling Average is often the only measure available and is used in many an analysis. I have only suggested that the Bowling Averages be used instead of Strike Rates. If anything, it makes the calculations simpler. Let me add that I have no problem with using the Strike Rate. The key measure is the % of Team bowling load which remains intact. This important career aspect is part of all my Ratings work.

  • Sohail on July 18, 2012, 3:35 GMT

    Interesting work, Ric. But I believe an important measure of bowler's worth is the bowling average. That should also play some part in your formula; maybe another division by the bowling average.

  • John on July 25, 2012, 11:55 GMT

    Another really interesting article, thank you. Chatting about it to my dad yesterday, also a big cricket fan, he wondered about the impact of bowler at the other end. In his words; "but does it show what impact a very accurate bowler at one end has on the wickets taken at the other." We were not immediately sure how that might easily be measured (value of bowler X as a whole v value in a particular partnership in some form pehaps) but thought it worth a comment!

  • Noah on July 18, 2012, 17:06 GMT

    One factor that your equation does not take into consideration is the quality of the team that any particular bowler plays for, and therefore their bowling percentage. For example had Kallis played for Sri Lanka, a side without the fast bowling quality of Ntini, Stein or Morkel, he would have bowled far more that 11% of the total. The same is true for non all rounders as well, Jason Gillespie for example would have bowled more if he was born in New Zealand and didn't compete with Warne, Lee and Mcgrath. It is amplified by 5 man attacks with players like Steven Harmison not making the list as a result of sharing the workload by virtue of being part of a team that had 5 bowlers due to the presence of an all rounder.

    Otherwise very interesting

    Ric: I think you are just confirming what I formulated: Kallis in a hypothetical Sri Lankan team would have had a higher value of V, and hence would have been more valuable to them than being the fourth or fifth bowler in a South African outfit.

  • Waspsting on July 18, 2012, 12:35 GMT

    I rank Sobers the 2nd greatest batsman ever.

    Didn't mean to give the impression i was comparing Sobers the bowler to Sehwag the bowler. kind of like pointing out if a bowling all-rounder had the same batting average as Ajit Agarkar - a legit point made, slightly 'cutely' (i.e. ironic, not particularly accurate but still legit)

    I think of him more as "One of the greatest batsman ever, and he could be a useful bowler" rather than the usual "undisputed greatest all rounder ever"

    @Jeff - telling stats about Gibbs and Sobers. Kind of supports a thought i had about a spinner being unnecessary if you have 4 great pacemen on the previous article on this blog

  • Gerry_the_Merry on July 18, 2012, 11:50 GMT

    My top picks (only last 40 years, since 75% of test cricket has been played in last 40 years) 1. Ambrose 2. Marshall 3. Lillee 4. Holding 5. McGrath 6. Hadlee 7. Imran 8. Wasim 9. Roberts 10. Gillespie/Waqar/Steyn/Donald 10.

  • Harsh Thakor on July 18, 2012, 8:53 GMT

    My list in order in terms of performance of paceman(over 250 wkts) are 1,Hadlee 2.Imran 3.Lillee 4.Marshall 5.Mcgrath 6.Ambrose 7.Waqar 8.Wasim Akram 9.Dale Steyn 10.Fred Trueman

    Ambrose was arguably,the best match -winner amongst paceman particularly in the 4th innings as he showed against England in 1994 at Trinidad and South Africa at Barbados in 1992.Mcgrath.and Marshall were the best match-winners in terms of percentage of games won.

  • Harsh Thakor on July 18, 2012, 8:45 GMT

    Imran Khan,is to me too low in your ranking,who had the 2nd best match-winning average and best performance by any paceman in his hey day in average and strike rate.Above all Imran was an all-time great match-winner who was a better performer than Wasim Akram if you asses his role in shaping wins in test matches and series versus the best opposition,be it England,Australia or West Indies.Marshall and Mcgrath bowled for top quality attacks and teams while Wasim Akram was hardly supported by the batting side.In his peak years Wasim performed better than Mcgrath .Overall to me Lillee and Marshall were the greatest and vie for the top spot,with their great competitivenes.What goes against Hadlee was his inability to be outstanding on slow tracks like Wasim Akram or Imran Khan.Waqar Younus was not as effective as Wasim against the best opponents like Australia,S.Africa or England and capitalized against weaker teams

  • Harsh Thakor on July 18, 2012, 8:34 GMT

    A very good effort Ric.I praise your ranking of Sir Richard Hadlee ,who to me on pure statistics is the best pace bowler ever.He has the best match performance analyis by any pace bowler ,more 5 and 10 wicket hauls,the best average in matches won,and has the best relative performance by any pace bowler in his peak,capturing 330 wickets in 60 tests from 1978-1988.Above all he bore the brunt of a weak attack.In the statistical light Imran Khan would come 2nd,considering that in his peak years from 1980-88 he had the best strike rate and average of any pace bowler of the modern era and was the best performer against the mighty West Indies.Howevever he also missed years from 1983-1985.3rd,I place Lillee.Sadly ,Lillee's scalps in Packer cricket and against rest of the world are excluded,otherwise he would have had 458 wkts.in 89 tests.To me, Marshall and Mcgrath are almost equal at 4th and 5th place but I give Marshall the edge.They were assisted by great bowlers and teams.

  • anand on July 18, 2012, 8:01 GMT

    i think your formula is heavily skewed to the percentage of team balls bowled, since that dramatically increases or decreases the net value.

  • Ananth on July 18, 2012, 3:59 GMT

    Ric: Perfectly acceptable, Ananth, and probably more encompassing in that it explicitly includes runs conceded. As I explained to another commentator earlier, I have implicitly assumed runs conceded in the bowling proportion in an attempt to minimise for simplicity's sake the number of variables involved. Ric: What I have suggested does not add any complexity at all. In fact the Bowling Average is often the only measure available and is used in many an analysis. I have only suggested that the Bowling Averages be used instead of Strike Rates. If anything, it makes the calculations simpler. Let me add that I have no problem with using the Strike Rate. The key measure is the % of Team bowling load which remains intact. This important career aspect is part of all my Ratings work.

  • Sohail on July 18, 2012, 3:35 GMT

    Interesting work, Ric. But I believe an important measure of bowler's worth is the bowling average. That should also play some part in your formula; maybe another division by the bowling average.

  • ChrisB on July 18, 2012, 2:56 GMT

    Hopefully I've just read this wring but shouldn't the 'value' of a bowler be their proportion of bowling divided by their comparative strike rate when compared against the others who bowl for their team? Given you've taken a player's proportion of the teams overs shouldn't the strike rate used be the player's 'proportion' of the team's strike rate?

    This would of course make your analysis absurdly difficult as you'd have to pull out the other bowlers strike rates for every game they've played with the player you're analysing but I'd imagine it would remove a few of those players who seem to be 'valued' seemingly due to %tbb alone.

  • Pankaj Joshi on July 18, 2012, 2:19 GMT

    Hi Ananth, If you could just combine the batting order number of the wickets taken, it would separate the heavy duty guys from the rabbit killers. When the top order goes quickly, generally all bowlers then bowl fewer overs. A lot of things would flow from there. Else, excellent as always.

    Ric: I am flattered by the name you have given me!

  • Meety on July 18, 2012, 1:08 GMT

    @Ric re: Imran Khan. Plenty of comments about the fact he had injuries that limited his bowling in the last 1/3 of his career. I wonder if you could do a one off, where you exclude the series where he didn't bowl at all & post what his "adjusted" V would be?

    Ric: Taking Imran between 8/10/1980 and 25/11/1986, when he took 164 wickets in 31 Tests, his BVI climbs to 0.56!! Useful....

  • Tanny on July 17, 2012, 20:42 GMT

    "I think Imran is so low because his percentage of overs has been brought down by all the times he played solely as a batsman, and after 1990, when he hardly bowled. Or did you somehow factor that in? Ric: No, not factored in. The figures we use are over his whole career, so that if there were periods when he didn't bowl, this will have the effect of reducing his percentage of balls bowled. This, I think, is fair enough - there is no value to a team as a bowler if he is not bowling."

    This happened because even when he was injured, he was good enough to play as only a batsman. And when his bowling declined, again he was good enough as a batsman and all-rounder, so he played on, instead of retiring. The problem with your analysis, in this case at least, is that it dilutes his stats, so people would come away thinking he wasn't a good bowler. At his peak: http://www.espncricinfo.com/magazine/content/story/484478.html (see the top bowlers between 1980 and 1988, strike rate and average).

    Ric: I agree, it is a shortcoming of the index that it doesn't take the best periods of a bowler's career, This would be very difficult to calculate for everyone. I guess it is detrimental to allrounders, who often can be included in teams for their batting, sometimes when they are unfit to bowl.But there is an argument, too, that doing so would be unfair to those bowlers who performed consistently throughout their career.Thank you for your comments!

  • JeffG on July 17, 2012, 16:08 GMT

    @ Waspsting

    I agree that Sobers bowling may benefit from a bit of a halo effect from his batting/fielding greatness but to compare him to Sehwag is a bit disingenuous.

    Leaving aside the fact that Sehwag has only taken 40 wickets and the obvious question of whether he would have managed to hold his SR if he’d been required to bowl in all matches in any conditions (as opposed to being used as a 2nd/3rd spinner on turning pitches), you need to consider players in relation to the era they played in.

    Sobers SR was 92. It sounds woeful until you realise that in the matches that Sobers played in, the average SR by all the other bowlers was 83.

    And when compared to Lance Gibbs, who most people recognise as a very good (if not quite great) spinner, Sobers looks even better.

    In the 60 matches that they played together, Gibbs took 237 wkts at a SR of 88 and Sobers took 181 at a SR of 89.

    I think that all in all, Sobers can be considered a good bowler

  • Mallu_Don on July 17, 2012, 15:36 GMT

    Great work and nice answers to boot. Your simple analysis is better than many complex ones I have seen at diferent places. And the ranking makes great sense. Thanks for this.

  • Boll on July 17, 2012, 15:20 GMT

    @Ric, this is a really interesting comparison, which has already raised issues with many people. I appreciate your adherence to a simple calculation, and understand your arguments vis a vis SR over average. However I think, apart from the first 30 years of tests, bowling averages have remained pretty consistent. The greats average close to 20, whereas strike rates have varied, depending on over rates, conditions etc.

    By your measure we have Kaneria (ave 35) on par with Bill O`Reilly (ave 22) - simply because of similar SR, similar % of balls balled. `An index of around 0.4 seems to separate the really great bowlers from those who were perceptibly less so.` No, I disagree, and also think that O`Reilly was a far more important component of his team as well.

    Be that as it may, this is really interesting work, and I hope you keep contributing to the site.

    Cheers

  • Boll on July 17, 2012, 15:05 GMT

    @Wasp. I think Sobers` rep as the greatest all-rounder probably has more to do with the fact that he was, for almost 15 years, probably the best batsman in the world. (By ICC end-of-year rankings NO.1 12 times from 1959-1973)Yes, he was an extremely versatile bowler (if average had been used instead of SR here, he would be a lot further ahead of Kallis than he is on this measure), and a brilliant fieldsman, but for most people he`d be batting at 5/6 in a World XI anyway.

  • Amit Ranjan on July 17, 2012, 13:06 GMT

    Interesting and very believable numbers. I always wondered can we get the value index based on matches rather then whole career for a bowler and then average it out? For example in a game total of 1100 runs scored and 35 wickets taken by 13 bowlers. How would it sum up if we count value index on average strike rate per game for a bowler? The point is, to nullify venue advantage for batsman and bowlers while getting stats.

  • Waspsting on July 17, 2012, 12:41 GMT

    Sobers' rep as "greatest all rounder" I think is based mainly that he bowled all the styles of bowling that one can (as opposed to being a good bowler).

    The guy's strike rate is the same as Virender Sehwag! VERY HARD to justify him being a "good" let alone "great" bowler in that light.

    I believe he had small periods when his bowling was solid - particularly his pace. Orthodox left arm slows was always economical. But overall, he seems a pretty average, if very versatile bowler.

  • sudhir on July 17, 2012, 9:08 GMT

    Hi Ric really nice one actually..thanks for the same. If I may I just wanted to add the % of runs conceded which will really increase the value of some bowlers bowling long and not conceding more like mcgrath and murali.

  • Jay on July 17, 2012, 8:52 GMT

    I do not agree with Ananath and Jeffji fully. The article is about the Value of a player to his team and not the greatest bowler of all time, which Ananath has already done several times, in several ways. However, I would mention that you should not have used the term "Great Bowlers". A useful bowler may not be a great one always, e.g. Kaneria.

  • Aneil on July 17, 2012, 7:02 GMT

    Stupid analysis, where Kaneria is rated over the great Imran Khan and even Ambrose.

    Ric: Thanks for your positive contribution!

  • PM on July 17, 2012, 5:01 GMT

    "Obviously, his value to his team would have been enhanced if could have bowled more while preserving his strike rate – but there is a risk, of course, that the strike rate diminishes if a bowler is asked to bowl more."

    Some doubts about this: 1)"Preserving his strike rate"? This assumes that the bowlers initial strike rate was higher. 2)This may apply to fast bowlers. 3)Spinners and good old ball/reverse swing bowlers may improve their strike rates as they bowl. 4)The strike rates are calculated after a match (or career)- so again this whole thing doesn't quite make complete sense to me. 5)Murali on top is almost a given. Good bowler in weak bowling unit. 6)I prefer the use of strike rate over bowling average too. Bowling average gives equal importance to runs given and wickets taken.Am not sure if this is the case.

    Ric: On point 1, I contend that a bowler who is asked to bowl long spells is likely to bowl more defensively than one who is assured of shorter bowling spells. My thesis is that the strike rate of long-spell bowlers is potentially less than short-spell bowlers, as a consequence.

  • shrikanthk on July 17, 2012, 4:17 GMT

    1960s were such a terrible time for bowlers' strike rate

    Ravi : Agree. That's why I prefer bowling average - a more resilient metric across eras.

  • shrikanthk on July 17, 2012, 4:14 GMT

    No, not factored in. The figures we use are over his whole career, so that if there were periods when he didn't bowl, this will have the effect of reducing his percentage of balls bowled

    I think you should factor it in. Atleast consider those innings where the bowler has bowled atleast 1 ball!

    That way we won't penalise excellent bowlers who chose not to bowl for a few tests (say, on account of an injury).

    An alternative way of measuring bowling value would be to consider 1. Bowling average and 2. Proportion of team's wickets taken. Again this can be done by considering only those innings where the bowler has turned his arm over.

    So someone like a Dennis Lillee averaged 23 and took roughly 25% of all oppositions wickets. So his V will be a function of these two figures (post normalisation).

    This way, we don't favour spinners (who will naturally bowl more) nor do we penalize express pacemen (who bowl a small proportion of total overs)

  • Shafiq on July 17, 2012, 4:10 GMT

    Imran played as a batsman and captain, not as a bowler in his injury era! Kaneria is off course unluckiest of bowlers. IMran is always there in the cricket's best all rounder race., when you see his leadership along with batting and bowling.

  • shrikanthk on July 17, 2012, 3:51 GMT

    Neat analysis.

    As Ananth suggested, I would use the bowling average instead of the Strike rate in the denominator.

    Also though I don't believe it will influence the rankings by much, I would ideally normalize both the components.

    I would do it using percentile ranking.

    Component A : Bowler with highest proportion of %TBB : 100 75th %ile bowler ranked by descending order of %TBB : 75 So this component will vary from 0 to 100.

    Similarly we've Component B. Bowler with lowest SR (or Bowling avg) : 100 75th %ile bowler ranked by ascending order of SR (Avg) : 75 etc. Again this component varies from 0 to 100.

    Bowling Value index = Component A + Component B.

    Also while presenting the results, I would bucket bowlers based on number of overs bowled (instead of total career wickets).

    Ric: Thanks for these suggestions. Involving bowling average brings runs conceded into the equation, which is getting a little away from my original investigation of seeing how bowling strike rate varies with the amount of bowling done. By the way, this measure (and over 2000 others!) can be had by obtaining our CSW database, which is publically available from www.cricketworld.com.au for those who are interested.

  • Gerry_the_Merry on July 17, 2012, 3:09 GMT

    Hadlee kept NZ competitive throughout the 80s, and it was quite common to see an analysis like 36-7-100-5, with the opposition making 350. He had made his action very economical and despite advancing age, kept going like a vintage Rolls. It is good to see his name at the top (I dont think of Murali in the same league as Warne as his action was decidedly suspect, with the extra snap in the wrists coming from the "elbow room", so Hadlee is effective #1 in the way I look at this table. Good work.

  • Meety on July 17, 2012, 0:59 GMT

    @umair - not having a go at you, but your comment sort of validates an observation I have of Kaneria. Putting aside any talk of fixing etc, Kaneria is more highly regarded by non-Pakistanis than Pakistan fans (I assume you are a Pakistani). I think he is a top class leggie, vastly underrated & could of been way better if a) didn't have Akmal as a keeper, b) most of his Paki team mate as fielders. I believe him to be one of the more unluckier bowlers to have played in the modern era. @Elvis - I agree, simple but very informative. @Dale Brown - the stats are whatever you want to make of them, I wouldn't worry if they don't paint Sobers bowling in a great light, it doesn't diminish his overall greatness, just that it is a tough gig being a star batsmen AND bowler at the SAME time!

  • Tanny on July 16, 2012, 19:30 GMT

    I think Imran is so low because his percentage of overs has been brought down by all the times he played solely as a batsman, and after 1990, when he hardly bowled. Or did you somehow factor that in?

    Ric: No, not factored in. The figures we use are over his whole career, so that if there were periods when he didn't bowl, this will have the effect of reducing his percentage of balls bowled. This, I think, is fair enough - there is no value to a team as a bowler if he is not bowling.

  • Dale Brown on July 16, 2012, 18:43 GMT

    It is amazing that you have totally misunderstood the value of Sobers as a bowler. The West Indies were the best team in the world between 1963 and 1967 and his value as a bowler during this period was on par with Hall and slightly less than Gibbs and Griffith. At the end of his career Sobers was the fourth highest wicket taker (235) in the history of the game in addition to making the most runs (8032)and taking the 3rd most catches(109) other than a wicketkeeper. He averaged bowling 23 overs per inning and took a three wicket haul every 7 innings. That is on par or just below other top bowlers such as Imran,Davidson,Wasim Akram,Botham,Vinoo Mankad,Shaun Pollock,Richie Benaud,Keith Miller and Kapil Dev. The difference between Sobers and the top bowlers is that he was not the bowler to get 5 wickets in an inning but that should not diminish his value. I admire your work but please look beyond the naked stats.

    Ric: My measure takes a bowler over his whole career, not just his most productive period. Unfortunately, by the time he had played 15 Tests, his bowling average had blown out to over 50, and despite his period of dominance, his final strike rate was 91 balls per wicket, which compares poorly with the peers you mention. This, of course, is going to have a detrimental effect on his Bowling Value Index. Sobers was a fine bowler at times, but not consistently so over his whole career.

  • Ravi M on July 16, 2012, 16:38 GMT

    Am I the only one to notice that DK Lillee was the ONLY 250-club pace bowler to have bowled more than 25% of the team's overs!? Contrary to the popular belief, the great man carried so much load for his team and when you look at his overall career, he had far less support that most people often made it sound like.

    As for Sobers, he still has better "V" than Kallis; doesn't he?! ;)

    1960s were such a terrible time for bowlers' strike rate. Of those with 50 wickets for that decade, nobody has sub-50 strike rate in the 60s. Only one with sub-56!!!! Sobers in fact had superior strike rate to Lance Gibbs, Benaud (past his prime though). Still "nailed" Barrington & Boycott 7 times each! Nobody else managed that.

    Sobers was often handed the worst role in the team. Took the new ball on turners so WI could pick more specialist spinners. Played as spinner on pacy wickets so WI could play all pace attack.

  • shafaet on July 16, 2012, 15:19 GMT

    Nice analysis,many of us never thought of analysing a bowler this way. Is there any analysis on odi economy rate? Can anyone give me the link? Thanks.

  • Arjun on July 16, 2012, 13:59 GMT

    "Securing a high percentage of wickets is a function, at least partly, of being given an opportunity to bowl, which is already there in the percentage of team balls variable"

    Ric, do not agree completely.... B Lee(0.398) is more valuable than C Mcdermott(0.392) according to your method. What is % of team wkts captured by both of them. I am sure Mcdermot would have picked up higher share of team wkts than lee.

    Ric: Lee took 22.76% of his team's wickets off 21.25% of the overs. McDermott took 26.72% of his team's wickets off 22.34% of the overs. Under your measure,McDermott comes off better, but because Lee's strike rate (53.33) is better than McDermott's (57), he comes out better in my measure. The original purpose of my investigation was to quantify the trade-off between strike rate and the proportion of bowling done. Lee, just, has come out better in my analysis, because of his superior strike rate for a similar proportion of bowling.

  • Arjun on July 16, 2012, 12:59 GMT

    Ric,

    Couple of years back i had suggested Ananth a different method. Proportion of team wickets taken in proportion to team balls bowled. D Steyn was on Top with a ratio of above 1.50. He had taken more than 30 % of his team wkts and bowled only about 20 % of his team balls. Isn't this more appropriate method of knowing value of a bowler to his team ?

    Ric: Certainly a valid way of measuring this stat. I guess you are simply substituting % of team wickets for strike rate. I think, on reflection, I prefer strike rate, since it is independent of bowling opportunity. Securing a high percentage of wickets is a function, at least partly, of being given an opportunity to bowl, which is already there in the percentage of team balls variable.

  • Jake on July 16, 2012, 12:51 GMT

    I love what you have done here Ric, and I wonder if it is possible to do something similar for batsmen and bowlers across all forms?

    Say you might have 2 batsmen, one is a quick scorer but is prone to low scores, and another who puts a high price on their wicket (someone who bats for a long time), and you compare runs scored off their own bat as a % to the team (as opposed to average) and balls faced, also as a % to the team (similar to strike rate, but team-based).

    eg. Batsman 1 on average scored 12% of his team's total and faced 22% of deliveries, while Batsman 2 on average scored 18% of the total, but faced only 17% of deliveries.

    Just some food for thought...

    Ric: Thanks, Jake, next time it rains....:-)

  • Farido on July 16, 2012, 12:47 GMT

    At first glance, the list of >250 wickets largely has players that played in the last 20 years. The list of 100-249 wickets is largely players who played more than 20 years ago (never heard of Grimmet, Blythe, Turner). What can we deduce from that? More workload taken by the better bowlers as time progressed?

    Ric: I think it is simply a case of more Test cricket having been played in the last 30 or so years than before.

  • Som on July 16, 2012, 12:03 GMT

    Ric...good work. Things which could further refine this analysis are:

    1-Is there a constant which could be added to the function to make the V=k(SR) function more responsive and thus a better measure 2-Based on where a cricketer comes from, the pitches play a big role in deciding whether a spinner or a fast bowler gets majority of overs. A case in point is Kapil Dev, who had the reputation of carrying a lions share of India's pace burden, but as pointed out here, had bowled only 20%. And thus in the overall scheme of things per this analysis, falls short. It would be interesting to have the percentages by bowling type and not overall. 3-While it is quite intuitive to say that the teams were able to maximize on their available bowling resources and may have just used bowlers to the point where their SR reached their max, but it may not have been so. One, not every captain or bowler may have understood that equation, and/or, may have been constrained by the quality of other bowler

  • JeffG on July 16, 2012, 11:55 GMT

    Thanks for the reply Ric.

    One further comment :

    While I understand the aim of this was to create a simple measure (and that is also the beauty of it) – I can’t help thinking about trying to refine it (I guess that’s part of the DNA of us stats junkies…)

    One thing that sprung to mind, was the differing Strike Rates over time and how this could affect the value of individual bowlers.

    Let me take Steyn & Trueman has the case study – they have a similar %tbb – about 21/22, but Steyn’s SR is much better (41 vs 49), making Steyn’s V score much higher than Trueman’s.

    However, Steyn plays in an era where the average SR is about 66 (lower than it’s been at anytime since WWI) where as Trueman played in the 50s/60s when average SRs were about 80.

    So, a SR of 49 in the 50s/60s is more valuable than it would be today.

    I’m not saying Trueman should move above Steyn, but in terms of value relative to the era they played in, I think they are a bit closer than the table suggests

    Ric: You are absolutely right, the measure takes no account of moving SRs as time elapses - one of the flipsides of having a simple measure. The best you can do is compare Trueman to Statham, Bedser, Davidson etc, of his era, while Steyn thanks his lucky stars that he is playing in an era of more aggressive batting where risk-taking is more prevalent. You can see the weakness of my measure when you look at the indices of Lohmann, Turner and (to a lesser extent) Barnes - very different to the indices of the modern bowler.

  • usama on July 16, 2012, 10:56 GMT

    Loved that work,it shows the real value of a bowler,thanks a lot,i hope u keep doing such fantastic statics table,Best wishes :)

  • Ananth on July 16, 2012, 10:40 GMT

    Ric A simple and elegant index. However as you would have seen in my articles I am a strong proponent of the Bowling Average as a composite measure. It is wonderful in that it encompasses both strike rate and bowling accuracy. That measure allows the bowlers with inferior values in one to make up with the other one. Steyn's high RpB is offset by his low BpW. And so on. So why would you not consider dividing the % of bowling share by the Bowling average instead of Strike rate. A sinmple but effective change. Muralitharan gets 1.47 Hadlee 1.13 McGrath 1.13 ... Kallis 0.407 and so on. 1.0 seems to be a nice cut-off point. Barnes 1.94 Lohmann 2.50. All matching your numbers with the added benefit that all aspects of a bowler have been taken into account. Ananth

    Ric: Perfectly acceptable, Ananth, and probably more encompassing in that it explicitly includes runs conceded. As I explained to another commentator earlier, I have implicitly assumed runs conceded in the bowling proportion in an attempt to minimise for simplicity's sake the number of variables involved.

  • Abhishek Mukherjee on July 16, 2012, 10:30 GMT

    Ric, nice work.

    But is considering proportion of overs bowled by his team the right parameter? Will that not depend a lot on whether the person is a seamer or a spinner?

    Ric: I am not sure you have understood the point being made here, summarised as the balance between bowling load and wicket-taking capacity. I believe the stat has relevance regardless of the type of bowler.

  • JeffG on July 16, 2012, 10:05 GMT

    Interesting stuff - thanks.

    Completely understand what you are doing here, but, in trying to measure a bowlers value to the team, don't we also have to consider the runs he concedes as well? I mean, it's all very well bowling 30% of your teams overs and taking a wicket every 8 overs, but it's somewhat diminished if you go for 5 an over while doing it. Someone with a slightly lower strike rate but a markedly better economy rate may well hold more value for their team.

    Maybe tweaking the equation to factor in "% of a teams runs conceded" might produce a more complete measure of value?

    Ric: Thanks for bringing this up! To my way of thinking, what you are suggesting is already built into the stat - if a captain thinks his bowler is too expensive, he will be taken off. This will be reflected in the proportion of overs bowled .It would of course be possible to do what you suggest, but I wanted to keep the idea simple, without the use of too many variables. I decided that in doing this, that wicket-taking capacity was more important than run-saving capacity.

  • ASIF on July 16, 2012, 9:57 GMT

    gr8 statistics Ric! It means waqar yunis and and shoaib akhtar could be more devastating if they could bowl for longer spells esp shoaib (keeping in view other outcomes of longer spells as well lol!).But tbb for a fastbowler is difficult in proportions of above 25. Well done.

  • Elvis on July 16, 2012, 9:47 GMT

    A greatness of this analysis lies in the simple and easily understandable formula/method used. While fast bowlers tend to have better strike rates, spinners bowl a higher percentage of the overs. The formula could have been given as V=K(1/SR). Could we also have included the bowling average as one more parameter in the formula?

    The results are in line with expectations for a large part but also throw up some twists. As usual Lohmann, Barnes and Murali are on top but there is some gap between Barnes and Murali. On many parameters in earlier analyses they were inseparable or extremely close.

    Bedi has the worst strike rate in the first table (Underwood's and even Vettori's strike rates are better) and Sobers strike rate is comparable to his. Since Sobers was usually the fifth bowler his %tbb is less. Hence his low V value as computed.

    For the Indian Scenario can this comparison be made for Gupte, Patel, the spin quartet, Kumble and Harbhajan and also for Kapil, Srinath and Zaheer?

    Ric: Thanks for your comments. A complete listing of all Test bowlers who have taken at least one wicket will be made available as a download.

  • umair on July 16, 2012, 9:18 GMT

    man i simply love your work.. keep it going..

    btw index of .4 and above does seem to be of bowling greats.. but what is kaneria doing there? lol :)

    Ric: Thanks for your kind comment. On Kaneria, don't forget, we are measuring value to his team - he bowled 29% of his team's overs, which is very high,and he had a reasonable strike rate for a spinner. Value and bowling greatness, although closely correlated, aren't exactly the same. If you understand this, then you will understand why he lists so highly.

  • No featured comments at the moment.

  • umair on July 16, 2012, 9:18 GMT

    man i simply love your work.. keep it going..

    btw index of .4 and above does seem to be of bowling greats.. but what is kaneria doing there? lol :)

    Ric: Thanks for your kind comment. On Kaneria, don't forget, we are measuring value to his team - he bowled 29% of his team's overs, which is very high,and he had a reasonable strike rate for a spinner. Value and bowling greatness, although closely correlated, aren't exactly the same. If you understand this, then you will understand why he lists so highly.

  • Elvis on July 16, 2012, 9:47 GMT

    A greatness of this analysis lies in the simple and easily understandable formula/method used. While fast bowlers tend to have better strike rates, spinners bowl a higher percentage of the overs. The formula could have been given as V=K(1/SR). Could we also have included the bowling average as one more parameter in the formula?

    The results are in line with expectations for a large part but also throw up some twists. As usual Lohmann, Barnes and Murali are on top but there is some gap between Barnes and Murali. On many parameters in earlier analyses they were inseparable or extremely close.

    Bedi has the worst strike rate in the first table (Underwood's and even Vettori's strike rates are better) and Sobers strike rate is comparable to his. Since Sobers was usually the fifth bowler his %tbb is less. Hence his low V value as computed.

    For the Indian Scenario can this comparison be made for Gupte, Patel, the spin quartet, Kumble and Harbhajan and also for Kapil, Srinath and Zaheer?

    Ric: Thanks for your comments. A complete listing of all Test bowlers who have taken at least one wicket will be made available as a download.

  • ASIF on July 16, 2012, 9:57 GMT

    gr8 statistics Ric! It means waqar yunis and and shoaib akhtar could be more devastating if they could bowl for longer spells esp shoaib (keeping in view other outcomes of longer spells as well lol!).But tbb for a fastbowler is difficult in proportions of above 25. Well done.

  • JeffG on July 16, 2012, 10:05 GMT

    Interesting stuff - thanks.

    Completely understand what you are doing here, but, in trying to measure a bowlers value to the team, don't we also have to consider the runs he concedes as well? I mean, it's all very well bowling 30% of your teams overs and taking a wicket every 8 overs, but it's somewhat diminished if you go for 5 an over while doing it. Someone with a slightly lower strike rate but a markedly better economy rate may well hold more value for their team.

    Maybe tweaking the equation to factor in "% of a teams runs conceded" might produce a more complete measure of value?

    Ric: Thanks for bringing this up! To my way of thinking, what you are suggesting is already built into the stat - if a captain thinks his bowler is too expensive, he will be taken off. This will be reflected in the proportion of overs bowled .It would of course be possible to do what you suggest, but I wanted to keep the idea simple, without the use of too many variables. I decided that in doing this, that wicket-taking capacity was more important than run-saving capacity.

  • Abhishek Mukherjee on July 16, 2012, 10:30 GMT

    Ric, nice work.

    But is considering proportion of overs bowled by his team the right parameter? Will that not depend a lot on whether the person is a seamer or a spinner?

    Ric: I am not sure you have understood the point being made here, summarised as the balance between bowling load and wicket-taking capacity. I believe the stat has relevance regardless of the type of bowler.

  • Ananth on July 16, 2012, 10:40 GMT

    Ric A simple and elegant index. However as you would have seen in my articles I am a strong proponent of the Bowling Average as a composite measure. It is wonderful in that it encompasses both strike rate and bowling accuracy. That measure allows the bowlers with inferior values in one to make up with the other one. Steyn's high RpB is offset by his low BpW. And so on. So why would you not consider dividing the % of bowling share by the Bowling average instead of Strike rate. A sinmple but effective change. Muralitharan gets 1.47 Hadlee 1.13 McGrath 1.13 ... Kallis 0.407 and so on. 1.0 seems to be a nice cut-off point. Barnes 1.94 Lohmann 2.50. All matching your numbers with the added benefit that all aspects of a bowler have been taken into account. Ananth

    Ric: Perfectly acceptable, Ananth, and probably more encompassing in that it explicitly includes runs conceded. As I explained to another commentator earlier, I have implicitly assumed runs conceded in the bowling proportion in an attempt to minimise for simplicity's sake the number of variables involved.

  • usama on July 16, 2012, 10:56 GMT

    Loved that work,it shows the real value of a bowler,thanks a lot,i hope u keep doing such fantastic statics table,Best wishes :)

  • JeffG on July 16, 2012, 11:55 GMT

    Thanks for the reply Ric.

    One further comment :

    While I understand the aim of this was to create a simple measure (and that is also the beauty of it) – I can’t help thinking about trying to refine it (I guess that’s part of the DNA of us stats junkies…)

    One thing that sprung to mind, was the differing Strike Rates over time and how this could affect the value of individual bowlers.

    Let me take Steyn & Trueman has the case study – they have a similar %tbb – about 21/22, but Steyn’s SR is much better (41 vs 49), making Steyn’s V score much higher than Trueman’s.

    However, Steyn plays in an era where the average SR is about 66 (lower than it’s been at anytime since WWI) where as Trueman played in the 50s/60s when average SRs were about 80.

    So, a SR of 49 in the 50s/60s is more valuable than it would be today.

    I’m not saying Trueman should move above Steyn, but in terms of value relative to the era they played in, I think they are a bit closer than the table suggests

    Ric: You are absolutely right, the measure takes no account of moving SRs as time elapses - one of the flipsides of having a simple measure. The best you can do is compare Trueman to Statham, Bedser, Davidson etc, of his era, while Steyn thanks his lucky stars that he is playing in an era of more aggressive batting where risk-taking is more prevalent. You can see the weakness of my measure when you look at the indices of Lohmann, Turner and (to a lesser extent) Barnes - very different to the indices of the modern bowler.

  • Som on July 16, 2012, 12:03 GMT

    Ric...good work. Things which could further refine this analysis are:

    1-Is there a constant which could be added to the function to make the V=k(SR) function more responsive and thus a better measure 2-Based on where a cricketer comes from, the pitches play a big role in deciding whether a spinner or a fast bowler gets majority of overs. A case in point is Kapil Dev, who had the reputation of carrying a lions share of India's pace burden, but as pointed out here, had bowled only 20%. And thus in the overall scheme of things per this analysis, falls short. It would be interesting to have the percentages by bowling type and not overall. 3-While it is quite intuitive to say that the teams were able to maximize on their available bowling resources and may have just used bowlers to the point where their SR reached their max, but it may not have been so. One, not every captain or bowler may have understood that equation, and/or, may have been constrained by the quality of other bowler

  • Farido on July 16, 2012, 12:47 GMT

    At first glance, the list of >250 wickets largely has players that played in the last 20 years. The list of 100-249 wickets is largely players who played more than 20 years ago (never heard of Grimmet, Blythe, Turner). What can we deduce from that? More workload taken by the better bowlers as time progressed?

    Ric: I think it is simply a case of more Test cricket having been played in the last 30 or so years than before.