Rashid Khan v Joel Garner v Mitchell Starc - who's the best ODI bowler?

In May, I wrote an article presenting the Boston Consulting Group (BCG) charts for the ODI batting measures: the batting strike rate and Runs per Weighted Innings. Now it's time analyse one-day bowlers similarly.

There is one fundamental change to the presentation. Up till now I used the standard method of BCG graphs presentation of four customised quadrants. However, I have recently seen many newspaper/web visuals in which only the top-right quadrant is used. The quadrant concept has its own uses in certain facets of analyses where player classifications are needed, but the standardised single-quadrant method of presentation gives us more flexibility in terms of showing a number of entries. And I have found that it's easier to view the relative positions of the players.

First, let me look at the base chart comparing the two contrasting measures: the strike rate (balls per wicket) v the runs per over. The selection criterion for all the graphs will be the powerful and simple bowling average. The other important criterion is that the career wickets captured by the bowlers should be 100 or more. My sincere apologies to Nathan Astle, the only retired bowler with 99 wickets, but the line has to be drawn somewhere. Bhuvneshwar Kumar will surely reach this mark in the next ODI he plays.

Through trials, I have found that a maximum of 40 bowlers can be shown comfortably. To the extent possible, the names are adjacent to the circle depicting the bowler. Pace bowlers are shown in kight blue and spinners in dark blue.

Readers should remember that in all these single-quadrant graphs, the high achievers are at the top and right. We are really looking for those whose placing is as high as possible and as much to the right as possible. Look at Rashid Khan's position. He is the undisputed king so far as strike rate is concerned. He is so far to the right that it is possible to conclude that his strike rate of 22.3 can only be bettered by him. This means that if Rashid Khan bowled his complete spell, he is likely to capture close to three wickets. No other bowler can even come close to this. His RpO is 3.91, keeping him down a little bit.

The next best strike rate is that of Mitchell Starc, with 25.5. However, Starc is extravagant, conceding nearly five runs per over. This pushes him down a lot. Ajantha Mendis, who bamboozled batsmen for a while before losing some of his sting, has a career strike rate of 27.4, which makes you wonder why he just disappeared. Even if batsmen had sorted him out, he was still striking at around 40 towards the end of his career. Maybe the Sri Lankan selectors were too tough, measuring him only against his own earlier standards and not against the standards of his peer bowlers. Mendis had an acceptable RpO of 4.80. In addition to these three bowlers, Shane Bond, Trent Boult and Brett Lee had strike rate values below 30.

Now let us look at the RpO measure. Just as Rashid Khan was the master striker, Joel Garner was the most accurate bowler ever. His RpO of 3.10 will never ever be bettered. I would even say that no modern bowler will even come within 25% of this figure. Since Garner needed nearly six overs to capture each wicket, his position is at the top but towards the middle as far as strike rate is concerned. Richard Hadlee and Michael Holding were close, with figures around 3.3. Both these bowlers also had strike-rate values either side of 39, keeping them closely bunched. However, let us not forget that all these bowlers bowled in the 1980s.

As can be seen, no fewer than 13 bowlers had career RpO figures of less than 4.0. Other than Rashid Khan, Glenn McGrath and Muttiah Muralitharan, the others played before the turn of the century. The last bowler to make the cut is Kyle Mills, who had a career bowling average of 27.02. Many top bowlers, like Kapil Dev, Lasith Malinga, Courtney Walsh, Anil Kumble, Daniel Vettori, and several others missed the cut because their bowling averages was higher.

I have drawn a line from the top-left corner to the bottom-right corner. This line of excellence can be used to separate the good bowlers from the very good, since anyone to the right of and above the line would have done very well in the combination of the two contrasting measures. In this chart, 14 bowlers achieve this distinction: Garner, Hadlee, Holding, Curtly Ambrose, Dennis Lillee, Shaun Pollock, Wasim Akram, McGrath, Murali, Saeed Ajmal, Allan Donald, Saqlain Mushtaq, Bond and Rashid Khan.

Since only spinners found their position in the selected 40, I have shown the same chart with the top 30 spinners only. The proportion has to be changed to accommodate the varying range of numbers.

Rashid Khan stands supreme, followed by Mendis. See how well-placed Saqlain and Imran Tahir are on the strike-rate front. The real surprise is Adil Rashid's strike rate. These four bowlers, along with Murali, are the only ones with strike-rate values below 35. However, Adil Rashid has conceded a whopping 5.5 runs per over. Now we see the advantage of a chart like this. The double-edged sword that Adil Rashid is can clearly be seen. It also shows the value of someone like Tahir, who combines a very good strike rate with an acceptable RpO value of 4.67, giving him an excellent bowling average of under 24.

A perusal of the top of the graph shows you that there are only four spinners with RpO values below 4.0: Rashid Khan, Murali, Roger Harper and Ray Price. Abdul Qadir is just above 4.0. The chart also makes it clear that the only spinner who has an RpO value exceeding 5.0 is Adil Rashid, who has a bowling average in excess of 30 despite the wonderful strike rate. It makes one certainly think what his real value is to England. The very good position of Mohammad Nabi indicates his value to Afghanistan. R Ashwin and Ravindra Jadeja are not that well placed, which may be a pointer to their recent fall out of favour.

In this graph there are 15 bowlers who are on the right side of the line of excellence: Rashid Khan, Mendis, Murali, Saqlain, Ajmal, Tahir, Graeme Swann, Abdur Razzak, Shakib Al Hasan, Anil Kumble, Nabi, Vettori, Rodney Hogg, Qadir and Shane Warne.

In the next chart, I have plotted the Wickets per Match (WpM) value against the Weighted Wicket Batting Quality (WBQ) value. These are two contrasting figures. The WpM value is self-explanatory and runs from 2.31 for Rashid Khan to 1.04 for Imran Khan. However, it is clear that Rashid Khan would not hit any high values in the WBQ value. Most of his opponents are the below-average teams. But, in order to be fair to the bowlers in matches between two below-average teams, I have adopted the following approach to determine the Weighted Batting Quality.

When Afghanistan play UAE, the average batting quality of UAE may not be very high, but it is important for Rashid Khan to pick up key top-order wickets to secure a win. He may not always get the opportunity to bowl against Virat Kohli or Babar Azam but he has to dismiss William Porterfield or Shaiman Anwar to give Afghanistan the chance to win these equally important matches. Hence I have adopted the following approach to determine the WBQ values.

Each wicket captured carries a maximum of ten points. These are split as follows.

Batting quality of batsman dismissed: three points. The maximum of three points are allotted when batsmen with batting averages exceeding 50, like Kohli, Babar Azam, AB de Villiers, Joe Root etc are dismissed, irrespective of when they were dismissed. Between 50 and 20, the points out of three are proportionately allotted. Dismissing batsmen with averages below 20 does not get the bowler any credit.

When the batsman was dismissed: four points. For batting averages between 50+ and 0, maximum credit is given for dismissing, say, Root for 0. Credit is given for dismissals up to the batting average value.

Batting order: three points. To take care of lower-level matches and matches where batsmen do not have great batting averages, the batting position of the batsman dismissed determines how many points are allocated out of three. Batting positions 1-4: three points; positions 5-6: two points; positions 7-8: one point; positions 9-11: 0.5 point.

So if Kohli is dismissed for 0, the bowler gets ten points - the maximum. If Kohli is dismissed for 70, the bowler gets six points.

In the selected group of 40 ODI bowlers, the WBQ values run from 4.02 (Rashid Khan) to 5.86 (Mills). Overall, within the 100-wicket bowlers, the range is from 4.02 (Rashid Khan) to 6.26 (Nuwan Zoysa). Now let us move on to the graph.

Rashid is all the way to the left of the graph but also way above others at the top because of his extraordinary 2.31 wickets per match. He is the only bowler in ODIs to have captured above two wickets per match. Starc comes in with 1.96 wickets per match but his average WBQ is a respectable 5.32. Bond has high figures in both measures: 1.79 and 5.62. Boult's figures are almost equally good: 1.85 and 5.65. As are Lee's - 1.72 and 5.58.

Saqlain's average WBQ is only 4.19, but this is made up by his capturing 1.7 wickets per match. Hogg is low in both; his numbers are 1.24 and 4.24 respectively. Many famous fast bowlers like Pollock, Malcolm Marshall and Ambrose are clocking in only at below 1.3 wickets per match but more than make up with WBQ values well above 5.2. Spinners like Murali, Tahir, Saqlain, Mendis and Warne are bunched together around the 1.6/4.5 mark.

This time there are only 11 bowlers to the right of the line of excellence: Starc, Boult, Bond, Lee, Mills, Shoaib Akhtar, McGrath, Mitchell Johnson, Makhaya Ntini, Nathan Bracken and Waqar Younis - all pace bowlers.

Finally, I have compared two contrasting parameters: the strike rate (BpW) and the maidens percentage. Let us first get some insights on maidens bowled in ODI matches. The quantum of maiden overs bowled has reduced from 6.5% during the 1970s to 2.4% now. There was a time when eight to ten maiden overs were bowled regularly in an ODI innings. Now, if we get three or four, we stand in awe. India bowled no fewer than 19 maidens in an innings against East Africa in their 1975 World Cup match. West Indies bowled 19 in a match against Pakistan in 1992. England bowled 18 against Canada in 1979.

The measure used is percentage of maidens bowled rather than the actual number of maidens, since that depends on the workload of the bowlers. The idea is to look at how the accurate bowlers, determined using the maidens per cent measure, did in the strike-rate measure. These two are rather contrasting measures. The maidens per cent varies from 18.0 for Hadlee to 4.0 for Rodney Hogg. Outside the selection, Gavin Larsen has only 3.8% maidens. The strike rate varies from 22.3 for Rashid Khan to 45.8 for Marshall.

Rashid Khan's excellent strike rate just about keeps him to the right of the line of excellence. Starc's equally good strike rate is not sufficiently high to give Rashid Khan company. There are very few bowlers who have done well enough to stay on the right side. It can be seen that barring Bond the bowlers with high maiden per cent values have strike-rate values above 30. Rashid Khan is the only spinner to make a mark. Otherwise, the pace bowlers excel. There is a heavy concentration of bowlers in the 8% and 30-40 grouping.

This is a tough chart to excel in. There are only seven bowlers above the line that matters: Hadlee, Ewen Chatfield, Garner, Lillee, McGrath, Bond and Rashid Khan. Starc, Boult, Bracken and Lee come quite close.

Just to conclude, Hadlee bowled 1.61 maidens per match, Garner 1.44 and Chatfield 1.36. Rashid bowled just short of 0.5 maiden per match and Swann 0.32.

In conclusion, let us give Rashid Khan the credit he deserves. His strike-rate figures are the best ever and are unlikely to be challenged. Similarly, his wickets-per-match figure. And let us not forget his bowling average. Garner comes close: his RpO figure will remain the best forever. And we have to give the next spot to Starc.

The table containing all these key measures is too long to be shown here. There are 143 entries and I thought that it would be a good idea to upload it to my Dropbox folder and allow interested readers to download it.

Over the past six months, it has been tables, graphs, more tables, more graphs, still more tables/graphs and so on for me. I am longing to do a more descriptive piece so for my next article, I will go back to an easier-to-read anecdotal piece. There won't be a single table or a single graph in it.