SELECT BLOG
October 2, 2010

Test All-rounders: an alternative BCG view

Anantha Narayanan

Keith Miller: one of the finest all-rounders  © Getty Images
Enlarge

Just to recap, Bruce Henderson of BCG (Boston Consulting Group) had created these charts during 1968 to study the Growth-Share aspects of products/business units. This is an excellent way to study two related variables together. These are plotted on a graph which is split into four equal (or unequal) size quadrants. The placement of a particular player, gives excellent insight into the player's position in the galaxy of all-rounders. However please do not forget that this is clearly a two-dimensional graph between two related variables. Also these are all career figures.

I elected to do an analysis of all-rounders, to start with, for Tests since that offers the clearest two-dimensional look. The all-rounder, based on a traditional definition, is clearly a two-dimensional player, Batting and Bowling. We can derive a lot of insight into the position of all-rounders and their relative strengths by doing the BCG charts.

As usual the real test starts in the selection criteria. Unlike the ODI bowlers and batsmen where a straightforward runs/wkts cut-off was used. Here the situation is too complex for a simple cut-off. We have multiple tasks in front of us. We have to have a reasonable number of players, not too many nor too few. The all-rounder standard should not be diluted. After a lot of trial and error efforts, I have decided on the following criteria.

1. All players who have scored 2000 runs or more and captured 100 wkts or more will be automatically included. This gets 23 players in.

2. Out of the remaining, players who have scored 1500 runs or more and captured 75 wkts or more will be included if their Batting average is better than their Bowling average. The later condition ensures that very average all-rounders like Emburey, Prabhakar, Streak et al are excluded. This gets 9 players in. Some of the players who get in are Faulkner, Armstrong, Mushtaq Mohd et al.

3. Now to take away the bowlers who can bat, players who have scored below 25 runs per test will be removed. This means two players, Warne (21.9 rpt) and Kumble (19.0 rpt) go out. Very fair since these two are not really all-rounders.

4. Also to take away the occasional bowlers who are primarily batsmen, all players who captured below one wicket per test will go out. This is fair since this is an analysis of all-rounders. So Hammond (0.98 wpt), Jayasuriya (0.87 wpt) and Steve Waugh (0.55 wpt) go out. I have been quite hard-nosed about this definition and have not been influenced by the very loose definition of all-rounders. Even though Steve Waugh has been called an all-rounder, there is no way he can be classified as one in view of the fact that he has captured one wicket in two tests.

That leaves 27 all-rounders for analysis.

Now we go to the analysis. This time I will do two different BCG analyses. The first will be based on two qualitative measures, the Bowling average and Batting average. The second will be based on two quantitative measures, Wickets per test and Runs per test. The advantage with this method is that it is not longevity based and gives equal chances to players whether they scored 11126 runs or 1968 runs or captured 431 wickets or 75 wickets.

I have not made any adjustment for the period or home country. My very loose conclusion is that such adjustments are not needed in an all-rounder analysis. If a player played during a batting-centric period, he would have the opportunity to have better batting figures which should compensate for the expected lower bowling figures. If a player played during a bowling-centric period, he would have the opportunity to have poorer batting figures which should be compensated by the expected better bowling figures. Similarly if he played on batting-friendly pitches, his better batting figures should compensate for the lesser bowling figures and vice versa on bowler-friendly pitches.

a typical BCG all-rounder chart

The above represents a typical BCG chart. The players in the top-right quadrant, the red one, are the "Top all-rounders". They are to the right of the Batting average line and above the Bowling average line. The ones in the bottom right quadrant, the green one, are the "Batting centric all-rounders". They bat very well but can at best function as fourth/fifth bowler for the team. Similarly, the top left quadrant, the blue one, contains the "Bowling centric all-rounders". They are normally the leading bowlers for their teams but bat at 7/8. The bottom left quadrant, the black one, represents the "Average all-rounders". They play the supporting roles in both batting and bowling.

Now let us view the graphs. I experimented a lot with the sloping dividing lines, as suggested by Sriraman, but could not work out a clear formula. The basis for a proper slope could not be worked out. Hence I have stuck to the dividing lines parallel to the axes. However I have made two significant changes, as suggested by Murali. The lines are drawn now at the centre but the scaling on either side of the lines is different. This makes for very good viewing despite the lopsided data. Sobers and Kallis cause this lopsidedness on the batting front with their extraordinarily high batting averages. On the other side, Shastri and Hooper cause this lopsidedness with their 40+ bowling averages. The numbers are shown along with the player names. I have also shaded the quadrants with the appropriate colour.

First the qualitative one, based on averages. I have also made my comments on the positioning of players without drawing any conclusions.

qualitative graph based on averages

Imran Khan and Miller are the leading all-rounders in this analysis. Aubrey Faulkner's presence would please the followers of Test cricket across the ages. The under-rated Trevor Goddard of South Africa is a surprise, but well-deserved, presence in this top quadrant. Botham is comfortably in this top group.

The batting centric group of all-rounders is led by the incomparable Sobers and Kallis, both with 55+ batting averages. There is another clutch of four all-rounders led by Greig, Brian McMillan (a surprise entrant - he just about makes it) and two greats of the 1910-20s, Armstrong and Woolley. Cairns just about misses the top quadrant. Two very average all-rounders, with awful bowling averages, Hooper and Shastri just about make it to this quadrant.

The bowling centric is a well-populated quadrant. This group is led by Pollock and has two greats of yonder, Noble and Rhodes. then we have Hadlee, Benaud and Wasim Akram.

The last group has Flintoff, Mankad and Vettori as clear residents. Vaas also belongs here. Bailey and Kapil Dev are on the borderline.

```No Player            Runs   Avge  Wkts   Avge  ARIdx1

1.Kallis J.H       11126  55.08   266  31.59   1.744
2.Sobers G.St.A     8032  57.78   235  34.04   1.698
3.Imran Khan        3807  37.69   362  22.81   1.652
4.Miller K.R        2958  36.97   170  22.98   1.609
5.Faulkner G.A      1754  40.79    82  26.59   1.534
6.Pollock S.M       3781  32.32   421  23.12   1.398
7.Mushtaq Mohammad  3643  39.17    79  29.23   1.340
8.Goddard T.L       2516  34.47   123  26.23   1.314
9.Greig A.W         3599  40.44   141  32.21   1.256
10.Hadlee R.J        3124  27.17   431  22.30   1.218
11.Noble M.A         1997  30.26   121  25.00   1.210
12.Botham I.T        5200  33.55   383  28.40   1.181
13.McMillan B.M      1968  39.36    75  33.83   1.164
14.Armstrong W.W     2863  38.69    87  33.60   1.152
15.Cairns C.L        3320  33.54   218  29.40   1.141
16.Rhodes W          2325  30.19   127  26.97   1.120
17.Woolley F.E       3283  36.08    83  33.92   1.064
18.Kapil Dev N       5248  31.05   434  29.65   1.047
19.Bailey T.E        2290  29.74   132  29.21   1.018
20.Mankad M.H        2109  31.48   162  32.32   0.974
21.Flintoff A        3845  31.78   226  32.79   0.969
22.Wasim Akram       2898  22.64   414  23.62   0.959
23.Vettori D.L       3962  30.71   325  33.87   0.907
24.Benaud R          2201  24.46   248  27.03   0.905
25.Shastri R.J       3830  35.79   151  40.96   0.874
26.Vaas WPUJC        3087  24.31   355  29.58   0.822
27.Hooper C.L        5762  36.47   114  49.43   0.738
```

I have presented the table above. The only additional field is the ARIdx1 value which is the Batting average / Bowling average. This is a far better measure than Batting average - Bowling average. An example will explain this. 50 and 30 would give an index value of 1.67 and a difference of 20. 40 and 20 would give an index value of 2.00 and the same difference of 20. It is clear that 40 and 20 is much better than 50 and 30. The difference of 10 in bowling is far more important.

Now let us view the second graph, which is quantitative one, based on per test values. I have again made my comments on the positioning of players without drawing any conclusions.

quantitative graph based on per Test values

Faulkner is comfortably placed in the top group. Botham and Miller are in this top group. Cairns is a surprise resident of this quadrant and this is a reminder to the New Zealanders that there were two top quality all-rounders there.

The batting centric group of all-rounders is led by Sobers and Kallis. Their wickets per test value is quite low, either side of 2.0, to let them move to the top quadrant.

The bowling centric is again a well-populated quadrant. This group is led by Hadlee and Imran Khan. Mankad moves up into this quadrant.

The last group is led by Bailey and Rhodes.

```No Player            Runs   RpT  Wkts   WpT  ARIdx2

1.Sobers G.St.A     8032  86.4   235  2.53   136.9
2.Hadlee R.J        3124  36.3   431  5.01   136.6
3.Faulkner G.A      1754  70.2    82  3.28   135.8
4.Botham I.T        5200  51.0   383  3.75   126.1
5.Imran Khan        3807  43.3   362  4.11   125.5
6.Cairns C.L        3320  53.5   218  3.52   123.9
7.Mankad M.H        2109  47.9   162  3.68   121.6
8.Goddard T.L       2516  61.4   123  3.00   121.4
9.Kallis J.H       11126  79.5   266  1.90   117.5
10.Miller K.R        2958  53.8   170  3.09   115.6
11.Benaud R          2201  34.9   248  3.94   113.7
12.Pollock S.M       3781  35.0   421  3.90   113.0
13.Greig A.W         3599  62.1   141  2.43   110.7
14.Wasim Akram       2898  27.9   414  3.98   107.5
15.Kapil Dev N       5248  40.1   434  3.31   106.3
16.Flintoff A        3845  48.7   226  2.86   105.9
17.Noble M.A         1997  47.5   121  2.88   105.2
18.Vettori D.L       3962  39.6   325  3.25   104.6
19.Armstrong W.W     2863  57.3    87  1.74    92.1
20.Vaas WPUJC        3087  27.8   355  3.20    91.8
21.Mushtaq Mohammad  3643  63.9    79  1.39    91.6
22.McMillan B.M      1968  51.8    75  1.97    91.3
23.Shastri R.J       3830  47.9   151  1.89    85.6
24.Rhodes W          2325  40.1   127  2.19    83.9
25.Bailey T.E        2290  37.5   132  2.16    80.8
26.Hooper C.L        5762  56.5   114  1.12    78.8
27.Woolley F.E       3283  51.3    83  1.30    77.2
```

The second table is presented here. The ARIdx2 value is simply RpT + WpT x 20. The 20 has been derived based on these all-rounder figures rather than the all-tests figures.

We can take a batting average of over-40 and a bowling average of below-20 to be a Bradmanesque all-rounder. No one exists like this. Even if we change to 40-plus and below-22 we have no one. Looking at the two charts, we can conclude that Miller, Faulkner and Imran Khan are right there in the leading group. It is of interest that Faulkner played for a weak team and Miller for a strong team. Pollock and Botham also belong there. The summary figures for the four top all-rounders are presented below. It will be difficult to question the credentials of any of the five. Only Botham is slightly out of place in this group. Even then his bowling average is better than the best Indian bowler ever, Bedi at 28.71.

```Player            Runs@Avge   Wkts@Avge  Idx1    RpT  WpT   Idx2

Faulkner G.A      1754@40.79   82@26.59  1.534  70.2 3.28  135.8
Miller K.R        2958@36.97  170@22.98  1.609  53.8 3.09  115.6
Imran Khan        3807@37.69  362@22.81  1.652  43.3 4.11  125.5
Pollock S.M       3781@32.32  421@23.12  1.398  35.0 3.90  113.0
Botham I.T        5200@33.55  383@28.40  1.181  51.0 3.75  126.1
```

Kapil Dev (31.05 and 29.65) does not belong to this group. Also if his bowling figures are adjusted because of bowling in the sub-continent, then his batting figures will get adjusted the other way. Similarly Hadlee (27.17 and 22.30) has too low a batting average. The bowling averages of Sobers and Kallis are quite high (31.59 and 34.04).

Finally let me conclude with a request to send in constructive comments which add value to the article. Bouquets or brickbats, it does not matter. If this article makes a young cricket follower look up GA Faulkner and Keith Miller in the records and marvel at their achievements, I would have achieved something.

My next article is a fascinating one analyzing Test Series. The Test Series, with their myiad variations as compared to the single tests provides scope for some interesting insights.

An important announcement to the readers. I have created an open mailid to which the comments and suggestions, not meant for publication, can be submitted. The mail id is ananth.itfigures@gmail.com. Since the readers would have to use a mail route I give the readers my assurance that the mail id is safe and will never be used by me for anything other than communicating with the reader specifically. This will not be part of any group mail nor will mails be cc'd.

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

Feeds: Anantha Narayanan

Posted by Murray Archer on (February 2, 2013, 7:48 GMT)

@ Ananth :)

I like concept of Innspell ! however.... limited to when you have them on the run just hammer away. To me the greatest spells were the most giving of all. (maybe because I know how much that hurt even if no result ?)

I reckon Clarkey, Tendulkar, Ponting, et al could have been 1st class players as bowlers, if they hadn't concentrated on batting at net practise. Of course, once they don't practise it, it can't be relied on.

Nice to see all the bowlers lately improve their batting ! :)

Posted by Murray Archer on (February 2, 2013, 5:47 GMT)

lol at Mark - this post is after reading back and back and back !

Ananth you truly amaze me at how close to what I think you deduce! (you know I have my thoughts from a completely different path !) [[ Thanks for the nice words which go a long way to make my day or two. Ananth: ]] Today, like in about 1952 Australia has to revert to athletes who chose cricket. The well of batting/fielding talent had run dry. Try the super gifted kids ? (why Richie Benaud, Ron Archer, Alan Davidson all got picked). Steve Smith, Glenn Maxwell, Moises Henriques (so wish Mitchell Marsh were not hurt - Moises not in that athlete class ! )

Gifted kids tend to be allrounders ? If Bradman or Tendulkar or - pick how many others .... 100 ?........ had not been so important to their team batting, they would have bowled more. They were good bowlers !

If Michael Clarke did not have a long time back injury, AND wasn't Aust's only good batsman, he could be second spinner on upcoming tour to India ? [[ My feeling is that he would be more valuable in ODIs a la Yuvraj Singh. In Tests both lack the variety to consistently come out with good innspell of 25 overs. Incidentally "Innspell" is my creation. It represents the complete bowling effort of a bowler in an innings since "Spell" normally refers to the continuous unbroken bowling effort.An Innspell can consist of multiple spells. Ananth: ]]

Posted by Mark on (February 20, 2011, 14:10 GMT)

This post is written a little after the article was done, but I've been analyzing Players & Stats trying to get a better understanding of past great cricketers & your articles are excellent over many years! This analysis though discounts a few Batting All-rounders from the past C.Macartney, S.McCabe, T.Dexter & others, these Bowled nearly 100 Balls per Match played (similar to M.Mohammad). Looking at these & other very good Batsman it really does increase their 'Value'. A worth that I'm very impressed by (I'd have liked to have been an all-rounder) in players, now & in times past!

Posted by Arvind Nandan on (January 3, 2011, 7:02 GMT)

Great Analysis Ananth.. I, however, feel that such analyses do not do proper justice to players. Without taking into account the teams that they were playing for, or the quality of support players (batting or bowling) they had, such comparisons can cause grave misunderstandings or misinterpretations.

I am not sure how a Botham would have fared if he was playing with a very weak bowling around him (did Flintoff last in this era, despite being hailed as the 'new Botham'?). Would Imran have lasted so long if he didn't have the fantastic pace battery around him?

Posted by Harsh Thakor on (November 7, 2010, 8:03 GMT)

I agree that Hadlee was the best fast bowler amongst all-rounders and could well make the all-time test 11 as an ideal partner to Marshall, and Akram or Lillee.

However readers must be explained why Kallis is not in the same level as Sobers,despite of being the champion statistically.The point is how many matches can an allrounder win with ball and bat or perform outstandingly.At his best Botham was Sobers greatest challenger while amongst fastbowling alrounders Imran Khan was the king.The Ian Botham of 1977-82 and the Imran Khan of 1981-87 were the greatest challengers to Sobers with Keith Miller very close.In certain phases if you consider his performances against the mighty West Indies Kapil Dev was almost in that bracket.Had he played for astronger team on helpful pitches he may well have joined Botham and Imran.

Posted by Andrew G on (November 1, 2010, 7:47 GMT)

Hi Ananth, A few bloggers I noticed bagged Flintoff, whilst his career stats are not mind blowingly special, the fact is like Botham before him they totally bossed an individual Ashes Series (unfortunately). So perhaps to put some perspective an an analysis on a 3 Test or more series in isolation may show who had the greatest impact on a Test series ever. I suspect Bothams '81 Ashes series would be hard to beat. Great article & can't wait for the MVP analysis. Cheers AG

Posted by Harsh Thakor on (October 9, 2010, 10:53 GMT)

Deservingly Gary Sobers is at the top.Kallis may match him statistically but when placed in true light Sobers is incomparable.No all-rounder has so consistently swung the fortunes of a match with bat and ball.

At his peak Ian Botham was the best match winner of all if you study the period from 1977-1982.The manner he won the 1980 Jubilee test at Mumbai and the 1981 home Ashes is legendary.Imran was more consistent than Botham and the best from 1981-87 ,but could not at his best turn the complexion of a match with both ball and bat like Botham could.Imran was mainly a match-winning bowler and became agood batsman late in his career.Infact in the light of ball and bat together Kapil Dev had more prowess.Richard Hadlee,is rated too high as he again was mainly a fast bowler and became a good batsman late in his career.

Kallis is an all-time great but has not consistently been a match-winner with both ball and bat.Had Keth Miller played more he could well have reached the top.

Posted by Ananth on (October 9, 2010, 5:26 GMT)

[[ Z-Scores for Test all-rounders Anirban had mentioned the value of Z-scores and I had requested him to do the Z-scores work and send me the document. It is a very interesting one and the final graphs are more or less similar to the averages based one I have posted. I have uploaded the document for the readers to download and view. The link is given below. http://www.thirdslip.com/misc/all-rounders-z-scores.docx My thanks to Anirban. Ananth: ]]

Posted by Sanjeev Chandran on (October 8, 2010, 11:23 GMT)

Interesting analysis - though i feel the quadrant names, unintentionally give a slightly unfair sense. I understand it but not calling Kallis and Sobers top allrounders takes a bit of sheen off. A suggestion - an all rounder contributes with both ball and bat. Suppose you take all the tests of a player - see the runs he scored wrt runs his team scored(say he scored x% of the teams runs) and ditto for wickets taken(say he had y% of the wickets taken. Then a composite would be Avg (x,y) which would give the all rounder's value. One could also decide if a player is bowling/batting centric by cut offs on x and y independent of each other. The problem with the above analysis is that Sobers is made a batting centric allrounder because his batting is so brilliant. But he took 235 wickets in an era when the world record was 309. If he took, say 15% of WI's wickets, he's a bowling centric allrounder as well. What those cut offs should be, i don't know- perhaps the data would show that.

Posted by Abhi on (October 7, 2010, 6:46 GMT)

Ananth, 1) Kallis should be referred to as “Kallis the Kollosus”…Unfortunately he suffers from the “Dravid/Steve etc. Syndrome”…His “style” or lack of relative to some other contemporary greats tend to reduce the true appreciation of his gifts. If he had played in a bygone era his numbers would have spoken for themselves. For eg. I don’t recall too many ppl referring to Bradman’s “style”, “flamboyance”, “flair”, “artistry” etc etc….He was simply a Run Machine…and his sheer numbers drown everyone else.

2) Since you are using BCGs , a Business/management “tool” – Here’s and interesting analysis connecting the stockmarket to India’s ODI performances.

http://www.indianexpress.com/news/stock-markets-decline-when-india-loses-odis-says-study/693648/

I realise that this type of analysis would be slightly out of your jurisdiction but it still is a good indicator of what ODIs and Tendulkar mean to India.