How do you rate all-rounders? For a start, there are different views on what an all-rounder is. If there isn’t agreement there, it’s hard to agree on the weight to be given to the different contributions an all-rounder makes.
Is it better to make runs or take wickets? Given they’re recorded in different units, how do you compare returns in each discipline?
It always helps when you’re rating anything to try to reduce the comparison to a single figure. A common one used for all-rounders is calculating the difference between their batting and bowling averages.
That’s fine so far as it goes, but does favour the batsman who bowls a bit, without giving enough weight to the amount of bowling the player contributes. It tends to downgrade the all-rounder who is a better bowler than he is batsman.
I therefore started by trying to come up with a way of expressing a player’s total contribution as a batsman and as a bowler in one figure, in a way where there was equal weight given to batting and bowling.
How to equalise batting and bowling? The concept I started with was that if you look at the number of runs a batsman who is regarded as ‘good’ will score per game, and at the number of wickets a bowler who is regarded as ‘good’ will take per game as benchmarks, you can calculate that a number of runs is equivalent in ‘goodness’ to a number of wickets.
You can then come up with an adjustment figure so runs and wickets are recorded in the same units.
That no doubt makes little sense. Explaining with the figures may be clearer. If you look through the list of high run scorers in Tests, a return of around 70 per Test marks out players who are on the cusp between good and very good.
That figure is inevitably a bit of a fudge because it depends on my concept of what is good, and runs per Test is affected by various factors, especially where in the order you bat. Nevertheless 70 per Test would roughly work out to an average between 40 and 45 for most players. For bowlers, four wickets per Test and you’re at the high end of good.
So does four wickets equal 70 runs? I’d actually say four wickets per Test is a little better than 70 runs per Test although it will be reasonably close, but remembering we’ll be taking runs scored into account, I think you need to adjust it a little further because while most specialist batsmen don’t or hardly ever bowl, specialist bowlers get to bat.
To adjust for the runs a specialist bowler will score I chose an arbitrary figure of ten runs per Test the bowler will probably score on average (which I don’t think would be too far wrong). So, a ‘good’ batsman will contribute 70 runs per Test and a ‘good’ bowler will contribute four wickets and ten runs.
For the purposes of measuring quality of contribution you can think of 70 runs as being equivalent to four wickets and 10 runs – and thus 60 runs equals four wickets or one wicket equals 15 runs.
From that, you can arrive at a way of assessing total contribution for all-rounders (be they ‘true’ all-rounders, batting all-rounders, bowling all-rounders or the much maligned bits and pieces all-rounder) – runs per Test plus (wickets per Test x 15). That gives you a single ‘contribution’ figure to allow a ranking to be made – and the arguing to start.
I acknowledge that there are numerous possible disputes regarding this method – including that runs per Test somewhat favours top order batsmen over middle order players, because they will get more batting opportunities than middle or lower order players (against that, you can argue it can be harder to make runs against the new ball and a fresh attack).
Measures such as averages and strikerates are not considered. Fielding is ignored. Subjective measures of the circumstances in which runs are made (where in the batting order, against what bowling, pitches) or wickets taken (quality of opposition, higher or lower order batsmen, pitches, competition for wickets, support from the other end) are ignored.
The calculation is also made over the whole of a player’s career – and players change what they do especially over a long career. Whole career figures also are prone to distortion if a player starts young and/or slowly, or keeps going when still good enough but past their best.
Most of the players mentioned in a recent Roar article on this subject which used the method of subtracting bowling average from batting average have been rated using my method in a table below.
I’ve included a lot of Australian players, particularly from the 70s on, because this is an Australian site and I’m more familiar with those players. I’ve made no effort to include non-Australians not mentioned in Dutski’s article and I’d be surprised if there aren’t a number who would rate highly.
For the purposes of setting some sort of qualification to get a rating, I’d say you need to have played 20 Tests, made 30 runs per Test and taken 0.8 wickets per Test. The 30 run requirement disqualifies Mitchell Johnson and Ray Lindwall (although they’re still in the table below) – that seems reasonable to me in that to me they were bowlers who could bat rather than bowling all-rounders.
I’d have said one wicket per Test should be the minimum qualification, but Stan McCabe and Andrew Symonds didn’t quite achieve that and one of them was clearly picked as an all-rounder, while the other was picked to bat and also to open the bowling more than once. If you’re picked as an all-rounder or expected to open the bowling, you should get a rating in my view.
| Name | Tests | Runs total | Wkts total | Runs/ Test | Wkts/ Test x 15 | “Contrib-ution” | Comment |
| Sobers | 93 | 8032 | 235 | 86.37 | 37.90 | 124.27 | No great surprise |
| Hadlee R | 86 | 3124 | 431 | 36.33 | 75.17 | 111.50 | Slight surprise? |
| Botham | 102 | 5200 | 383 | 50.98 | 56.32 | 107.30 | Ditto |
| Kallis | 165 | 13206 | 291 | 80.04 | 26.45 | 106.49 | Maybe a fraction lower than many would think |
| Cairns C | 62 | 3320 | 218 | 53.55 | 52.74 | 106.29 | Big surprise to me |
| Imran | 88 | 3807 | 362 | 43.26 | 61.70 | 103.96 | Lower than many think based on his batting av |
| Gregory J | 24 | 1146 | 85 | 47.75 | 53.13 | 100.88 | Will surprise many |
| Miller | 55 | 2958 | 170 | 53.78 | 46.36 | 100.14 | Slips a bit despite his bowling av |
| Stokes | 23 | 1383 | 57 | 60.13 | 37.17 | 97.30 | Producing a batsman’s runs. Early days |
| Simpson | 62 | 4869 | 71 | 78.53 | 17.18 | 95.71 | Some surprise – better bowler than is thought |
| Benaud | 63 | 2201 | 248 | 34.94 | 59.05 | 93.99 | This method puts him higher than others |
| Davidson | 44 | 1328 | 186 | 30.18 | 63.41 | 93.59 | 30 runs per test makes him an all-rounder? |
| Pollock S | 108 | 3781 | 421 | 35.01 | 58.47 | 93.48 | Underrated player |
| Johnson | 73 | 2065 | 313 | 28.29 | 64.32 | 92.61 | Doesn’t qualify (just) |
| Flintoff | 79 | 3845 | 226 | 48.67 | 42.91 | 91.58 | Underrated for mine |
| Noble | 42 | 1997 | 121 | 47.55 | 43.21 | 90.76 | Good player for a long time (runs harder to get then) |
| Chappell G | 87 | 7110 | 47 | 81.72 | 8.1 | 89.82 | Doesn’t qualify |
| Giffen | 31 | 1238 | 103 | 39.94 | 49.84 | 89.78 | Same comment as Noble |
| Dev | 131 | 5248 | 434 | 40.06 | 49.69 | 89.75 | This will be a big surprise |
| Gilmou | 15 | 483 | 54 | 32.20 | 54.00 | 86.20 | Doesn’t qualify. Included as the 70s’ biggest coodabeen |
| McCabe | 39 | 2748 | 36 | 70.46 | 13.85 | 84.31 | Qualifies as he opened the bowling a few times |
| Armstrong | 50 | 2863 | 87 | 57.26 | 26.10 | 83.36 | Maybe suffers from the length of his career |
| Moeen | 23 | 949 | 64 | 41.26 | 41.74 | 83.00 | Epitome (to me) of the bits/pieces all-rounder |
| Walters | 74 | 5357 | 49 | 72.39 | 9.93 | 82.32 | Batsmen who bowls a bit as was common in the 70s |
| Watson | 59 | 3731 | 75 | 63.24 | 19.07 | 82.31 | Possibly where you’d expect him to be |
| Lindwall | 61 | 1502 | 228 | 24.62 | 56.07 | 80.79 | Long career affects figures. Not enough runs |
| Macartney | 35 | 2131 | 45 | 60.89 | 19.29 | 80.18 | Started as a bowler who could bat, finished as a batsman who could bowl |
| Border | 156 | 11174 | 39 | 71.63 | 3.75 | 75.38 | Too few wickets. Validates the methodology ;)? |
| Waugh S | 168 | 10927 | 92 | 65.04 | 8.21 | 73.25 | Ditto |
| Symonds | 26 | 1462 | 24 | 56.23 | 13.85 | 70.08 | A bit worse than I would have thought |
| Waugh M | 128 | 8029 | 59 | 62.73 | 6.91 | 69.64 | Same comment as Steve |
Dilip
Guest
Make fast bowler
Joe
Guest
My system is the BEST...........what do you think.
Joe
Guest
It's better if you convert the bowling average to a batting average as such and then add it to the batting average to find out how good an allrounder is on quality . As far as test cricket is concerned top batsman aim to average between 40 to 60,while top bowlers aim to average between 20 to 30.........there's a ratio of 2 to 1 run's,a 20 run batting range difference to a 10 run bowling range difference. So if you have a batting average of 60 it equates to a bowling average of 20..........while if you have a batting average of 40 that equates to a bowling average of 30, conversely if you have a batting average of 50 it equates to a bowling average of 25 ect. MORE DETAILED A 60 batting average is a 20 bowling average. A 59 batting average is a 20.50 bowling average. A 58 batting average is a 21 bowling average. A 57 batting average is a 21.50 bowling average. A 56 batting average is a 22.00 bowling average. A 55 batting average is a 22.50 bowling average. Ect,Ect A 50 batting average is a 25.00 bowling average. Etc,Etc A 45 batting average is a 27.50bowling average. Etc,Etc A 40 batting average is a 30.00 bowling average. This continues after if the bowling average is over 30,but at this time the difference is a run for a run. 'Example' A 39 batting average is a 31.00 bowling average. A 38 batting average is a 32.00 bowling average. A 37 batting average is a 33.00 bowling average. A 36 batting average is a 34.00 bowling average. A 35 batting average is a 35.00 bowling average. Thereafter you could go to a maximum of.................... A 30 batting average is a 40 bowling average . If you use this system it's far more accurate in working the quality of the all rounder rather than using various questionable standards........one that is totally incorrect is if you take away the bowling average from the batting average.......as i noted earlier there's a 2 to 1 run ratio.....that aside in using this format the top 4 player's are as follows showing the batting average first then there bowling average converted to a batting average this then has a total average. GARY SOBERS---------57.78...35.96-------93.74 JACQUES KALLIS----55.37...37.34-------92.71 IMRAN KHAN-----------37.69...54.37-------92.06 KEITH MILLER----------36.98...54.04-------91.02 Remember this is a quality standard not a quantity standard,meaning that as with IMRAN KHAN and KEITH MILLER both have similar bowling averages but theres a 1 wicket difference per match in favour to IMRAN KHAN ......that's another system all together. Quickly on that 20 runs equates to 1 wicket.
My2cents
Guest
The main flaw with my ranking is that it's useless for test cricket. However I've decided to tweak it for limited overs. I will run it soon but if you'd like to have a go. The new formulae would be. For t20 for ODIs replace every 120 with 300 For test cricket it doesn't work because batting is defense and bowling is attack so the whole system would have to be reworked ORTG (batting average x (strike rate/100) x average balls faced) / 120 for T20 and 300 for ODIs DRTG ( bowling average x (strike rate / 100) x ( 25 - average balls bowled per match) ) / 120 Then ORTG - DRTG for net rating
Bearfax
Guest
So My2CentsWorth, I'm now interested in how your calculations look against the top all rounders we've been discussing. Are you able to supply those figures. I'm actually interested based on your outcomes with some recent Australian all rounders.
My2cents
Guest
Let's compare that measurement to Australia's current crop of t20 all rounders. Watto real batting average 43.14 real bowling average 5.46 net rating 37.68 Glenn Maxwell net rating 26.47 Mitch marsh 31.7 Steve smith real batting average 33.8 which is surprisingly low but his surprisingly impressive bowling numbers give him a net 30.9 I'd wager if you added all these numbers up for both teams you'd get pretty close to who was more likely to win a T20 game
My2cents
Guest
I'm not sure I'd agree with this system. But since I can't think of anything better good job. If I were to come up with a stat it would probably be runs scored as a batsman - ( wickets taken as a bowler x bowling average) If you have a positive number then you contributed positively to your teams chance of winning. If you didn't you actively made your team worse by engaging your secondary skillset. Let's apply this to Garry sobers 86.37 runs per test - 85.99 ( wickets per test x bowling average) Sobers was only a +0.44 net positive to his teams. This stat doesn't take into account his versatility or use as a holding bowler. this stat would be much more useful when evaluating limited overs cricketers in the T20 or ODI Format. Another would be (average x strike rate ) \ 100 then subtract that by ( bowling average x strike rate) wickets are obviously accounted for with the strike rate.
Aransan
Guest
Thanks for the explanation JGK, I suspected he had some physical issue. I wonder whether he was considered to have had some x-factor, being able to crack a game open when things just seemed to be meandering on. Miller may have also been over bowled in the 1948 first test at Nottingham where he had match figures of 7/163.
JGK
Roar Guru
@ Aransan - Miller had a bad back (a remnant of his service in the war if not "shaggers back") so often went a few games at a time with little or no bowling. Also, the 50s were one of the slowest times ever in test cricket with many bore draws on flat decks. Therefore there were rarely 20 wickets a test to go round but averages were still pretty low.
Bearfax
Guest
Aransan wisdom has nothing to do with it. I'm as thick as the next person. I just operate on figures and logic...question everything. I mean did Adolph Hitler actually live. Lots of evidence says he did and he caused a lot of mischief. But I never saw him. He supposedly lived before I was born, and that Aransan is the beginning of the World for me.
Aransan
Guest
Bearfax, when it suits you then you just use batting and bowling averages and when it doesn't you go to other arguments. Wickets per test does pick out very good bowlers, for example: Lillee 5.07, Hadlee 5.01, Warne 4.88, MacGill 4.73, Bedser 4.63, Trueman 4.58. Spinners are not favoured by bowling averages or wickets per test and the statistics are less relevant for them. Bowlers who take close to 4 wickets a match will still have respectable bowling averages. I believe it is very difficult to do meaningful comparisons with bowling averages between eras and countries and also between medium/fast bowlers and spinners.
Aransan
Guest
I bow to your greater wisdom, however you have ignored his wickets per test. Two of Miller's contemporaries were Bill Johnston who took 160 wickets in 40 tests, or 4.00 wickets per test and Alan Davidson who took 186 wickets in 44 tests or an average of 4.23, I have already referred to Lindwall's average of 3.74. However, I am always prepared to look beyond the statistics and accept your argument that it was a strong bowling lineup. It is interesting that in 3 tours of England there was one test in each tour when Miller didn't bowl at all -- the second test at Lords in 1948, the first test at Nottingham in 1953 and the third test at Leeds in 1956.
Bearfax
Guest
Thats false economy Aransan and you know it. You're valuing a player the same on the basis of number of wickets per innings without considering the number of overs bowled and runs scored against them. Miller averaged about 10-12 over per innings because he had a very strong fast bowling team supporting him. Botham bowled about 20-25 overs an innings because England were weak with fast bowlers and it was left to him to try to take the wickets. But you're virtually saying that someone who takes 3-20 in a match is the same as someone taking 3-100. Wickets are what its about but the economy in how you take those wickets is just as important. If Miller had not had Lindwall (average 23) and Johnson (average 24) in the side for most of his career he would have been the front line bowler, but being a batter as well he was saved. Miller actually had a better average than those two at 22 and was considered a very hard bowler to bat against. He was much more involved in his last number of tests when he bowled in tandem with the great Davidson (average 20). Botham only ever had Willis (average 25) as an effective fast bowler in the team, so he had to step up and bowl more frequently. Therefore he got more wickets per innings and the most wickets for England despite a 28 average. If there is no one else effectively taking wickets then its obvious Botham, bowling as much as Willis, would take wickets even if for a more inflated average.
JGK
Roar Guru
Miller would most certainly have been picked purely as a bowler for most of his career. He was lucky enough to play in an era when Aust had a strong bowling side so he wasn't needed to bowl 50 overs per test. No less a player than Len Hutton said he found Miller more difficult to bat against than Lindwall but of Miller's unpredictability. But his average and strike rates are comparable to anyone's of the era.
Aransan
Guest
We can get a very good idea of the quality of bowlers by looking at the average number of wickets they take per test. Miller took 170 wickets in 55 tests, at an average of 3.09 -- (55,170,3.09) For the other players you mentioned: Botham (102,383,3.75) Willis (90,325,3.61) Trueman (67,307,4.58) Broad (91,333,3.66) Botham's average compares very well against Willis and Broad, he hasn't just taken a lot of wickets because he played more tests. Trueman was a very good bowler. Averages for other well known players are: Statham 3.60, Bedser 4.63, Lindwall 3.74, Lillee 5.07, Hadlee 5.01, McGrath 4.54, Warne 4.88, Benaud 3.94, MacGill 4.73, Lyon 3.61. Miller took at least half a wicket less per test than any other bowler mentioned above, and I would argue would not have been selected just as a bowler. There is no doubt however that he more than justifed his selection as an all rounder.
Bearfax
Guest
Thing is with Botham was that he had a bigger than life persona and so spectators were enamoured by him. But he played for over 100 tests and 15 years. Little wonder his figures look good such as taking the most wickets, Mind you both Willis and Truman would have easily surpassed him if they had stuck around with their averages. And if you look closely Stuart Broad is not that far behind and, will probably surpass him with a comparable bowling average. Would you say Broad is one of the best bowlers England has had. He's good, but not that good. Point I was making is that your formula is good but needs treaking because, based on averages, it doesnt seem to represent the true worth of some of the all rounders in the group, and raises some who were good but not that good.
Bearfax
Guest
On the contrary Aransan, I have always said stats are not the be all and end all, but they are by far the best method of measuring a players worth. If not why do sports coaches all over the World in all sorts of sports retain stats on players. Thing is you look at the stats first and then you look at other factors. Of course age, experience, injury are factors. But most people seem to base their view of players on impressions, and style, on how they look on the field. They have their worth, but only after relevant stats to the sports are considered. Some offer stats of certain periods in a career and that's why I gave the Voges/Waugh example. Players form varies over the years. Sometime their hot, sometimes their not, but that is factored into long term stats. Bradman averaged 105.7 after the war. This from an almost crippled 40 year old. But one has to take into consideration that many he was playing against were novices or war torn vets. Each season has its own variables, but over the long term such things tend to even out
JohnB
Guest
Bearfax, 70 runs per test intuitively seems not that many - but I ask you to look at the list above and see who on it exceeded or approached that number. Certainly it's mostly a list of all-rounders - but it includes a number of blokes who were much more batsmen who bowled, and who in their day jobs were very, very good batsmen. Take out the absolute top flight batting all-rounders Sobers and Kallis, and it's only top order players or very good middle order players doing better than that 70 mark. I don't want to buy into the Botham argument - but I will point out that up until very recently he was England's all time highest wicket taker, and that his bowling figures lose little by comparison with those of Jimmy Anderson who overtook him. Australians tend to be pretty qualified in their rating of Anderson, but Englishmen don't, for what that may be worth. Botham's bowling figures also lose little compared with those of his contemporary Bob Willis. I think his career bowling is underrated.
Ronan O'Connell
Expert
Shakib is a decent cricketer but he hasn't proved himself against the top sides. He's never played Australia in Tests and has been ordinary in his 15 Tests against the other 3 best Test sides of his era - SA, India and England. Against those 3 strong teams he has averaged just 21 with the bat and 36 with the ball, which is nothing special. His ODI record is the same story...against the dominant 3 ODI sides of his era - Australia, SA and India - his stats are way worse than his career stats. Against those 3 elite ODI sides he's averaged 40 with the ball (compared to career average of 28) and 29 with the bat (compared to career average of 35).
Aransan
Guest
Bearfax, a statistician would tell you that with such small sample sizes the averages are subject to significant sampling error. I am not blind about using statistics but there is a lot more to cricket than batting and bowling averages. The more one understands numbers, the less confidence one has in measures. You are too confident about your statistics when it suits you.