Wednesday, June 23, 2010

Rahul Dravid has the highest number of "significant" innings: A Cricinfo analysis

It is nice to be back after a valuable and recharging break. It is also wonderful to renew acquaintance with the valued readers. The break was necessary but I could not wait for the self-imposed sabbatical to be over.
In this article I have gone back to the reader's suggestions, specifically Xolile. He had suggested a few months back that I should look at separating the significant Test innings based on runs scored and balls faced, wherever such information is available, and rating batsmen using this information. I have taken that suggestion and completed the analysis after significantly improving the basis.

He had suggested that I take 80 runs and 160 balls as the basis. I have instead worked on a dynamic fixing of the cut-off points based on the specific match conditions. The idea is that I should achieve the following inclusions and exclusions through this analysis.

The analysis should be done so that the following innings (just a few examples) are included.

- Gillespie's 9 (off 51) out of Aus total of 93 a.o (30 overs) at Mumbai
- Guptill's 30 (off 122) out of Nzl total of 157 a.o (59.1 overs) at Wellington
- Srinath's 76 (off 159) out of Ind total of 416 a.o (128.3 overs) at Hamilton
- Hutton's 30 (balls n/a) out of Eng total of 52 a.o. (42.1 overs) at Oval
- A.H.Kardar's 69 (balls n.a) out of Pak total of 199 a.o (91.3 overs) at Karachi
and so on.

and the following innings (just a few examples) are not included.

- Collingwood's 60 out of Eng total of 569 for 6 at Chester
- Clarke's 83 out of Aus total of 674 for 6 at Cardiff
- Ranatunga's 86 out of Slk total of 952 for 6 at Colombo
- Walcott's 88* out of Win total of 790 for 3 at Kingston
- Rae's 63* and Stollymeyer's 76* out of Win total of 142 for 0 at Trinidad
and so on.
I have taken one decision, slightly reluctantly. Any 100 would be considered to be significant. Although I do not consider a 100 by itself to be anything special, I think this is a correct decision since out of the 68,879 innings played to date only 3370 hundreds have been scored and this constitutes around 5%. It is not a bad premise to start with, banking one in twenty innings.

As far as the often quoted instances of batsmen scoring 100s in dead match situations, the following example will show the pitfalls.

Take a match where two days have been washed out. The match scores are

Team 1: 300 for 5. Team 2: 300 for 6. Team 1: 300 for 7 (Xyz 100+).

If the first two days are lost due to rain, the third innings century is a totally irrelevant one scored on the last day. On the other hand if the last two days have been washed out, the third innings century is a very relevant one made in a live match situation on the third day. If the rain had occurred on other days, the value of the 100 would oscillate significantly. Hence pre-conceived notions of the significance or non-significance of innings should not be used to come to conclusions. Also incorporating rain factor, when it happened, on what day the runs were scored all are virtually impossible in any analysis because of the absence of dependable data.

Since 80 and 160 are arbitrary, I have worked on a dynamic determination of the cut-off for each match, separate for either team. This makes sense since I should include an innings of 9 and exclude a 88* innings. There cannot be common cut-off criteria.

The cut-off methodology is explained below. Based on the cut-off points 2 to 5, 12,529 innings below 100 have got selected.

An innings is considered to be significant if it satisfies any one of the following five conditions.

1. The runs scored is greater than or equal to 100 (already talked of).

2. The balls faced is greater than or equal to 200.

3. The runs scored is greater than or equal to the cut-off figure for the team, as explained below.
- For batsmen 1-7, 1.333 times the Runs per wkt value for the team for the two innings together.
- For batsmen 8-11, 1.167 times the Runs per wkt value for the team for the two innings together.

4. The balls faced is equal to or higher than the cut-off figure for the team, as explained below.
- For batsmen 1-7, 1.667 times the Balls per wkt value for the team for the two innings together.
- For batsmen 8-11, 1.333 times the Balls per wkt value for the team for the two innings together.

5. To take care of very low innings totals, see Hutton example above, the runs scored is greater than or equal to one third of the team total. The team should have lost 5 wickets or more. Otherwise Stollymeyer-type innings would get through.

Seems complicated but all conditions are logical once the above 5 conditions are understood properly, and the fact that an innings has to adhere to at least one of these in order to be seen as significant in this analysis. Of course, a cursory glance would be woefully inadequate. These cut-off numbers have also been determined after a lot of trial work during the past few days. A higher cut-off will mean missing out of some significant innings while a lower cut-off will mean inclusion of ordinary innings. Overall this method is slightly unfair to older batsmen since they have only the "Runs scored" criteria available to them. However nothing can be done about that.

I got a massive list of 15,899 innings, which is about 23% and this figure looks good. Then I posted these into the player database and got the player table. This table is sequenced on the % of significant innings since the number of innings played varies considerably. The cut-off for batsman selection is 3000 runs and above. 159 batsmen qualify.

SNo Batsman For Mats Runs Inns SI % SI

1.Bradman D.G Aus 52 6996 80 43 53.8
2.EdeC Weekes Win 48 4455 81 39 48.1
3.Hobbs J.B Eng 61 5410 102 49 48.0
4.Chanderpaul S Win 123 8669 210 98 46.7
5.Barrington K.F Eng 82 6806 131 61 46.6
6.Sutcliffe H Eng 54 4555 84 39 46.4
7.Lara B.C Win 131 11953 232 106 45.7
8.Dravid R Ind 139 11395 240 108 45.0
9.Hutton L Eng 79 6971 138 62 44.9
10.Flower A Zim 63 4794 112 50 44.6
11.May P.B.H Eng 66 4537 106 47 44.3
12.Viswanath G.R Ind 91 6080 155 68 43.9
13.Hammond W.R Eng 85 7249 140 61 43.6
14.Compton D.C.S Eng 78 5807 131 57 43.5
15.Umrigar P.R Ind 59 3631 94 40 42.6
16.Mitchell B Saf 42 3471 80 34 42.5
17.Sarwan R.R Win 83 5759 146 62 42.5
18.Manjrekar V.L Ind 55 3208 92 39 42.4
19.Javed Miandad Pak 124 8832 189 80 42.3
20.Gavaskar S.M Ind 125 10122 214 89 41.6

How often do we a table headed by Bradman. More than 1 out of 2 innings played by Bradman are significant. He is the only player to have exceeded 50%. Then come two giants, Weekes and Hobbs, who have figures around 48%, the one mitigating factor is that they are within 10% of Bradman.
Now the biggest surprise. The unheralded and unsung Chanderpaul clocks in at 46.7% ahead of his more illustrious contemporaries. It shows the solidity and quality Chanderpaul brought to position No. 6. He could very well improve in the years to come. Barrington and Sutcliffe come in next, both great defensive batsmen. Hutton chips in in the 10th position.

Now we have two modern greats, Lara and Dravid. Lara's playing in a weaker team has helped a bit in this regard, but there can be few detractors to the claims of his greatness. Same applies to Dravid. What he has achieved for India has not been acknowledged, especially on the Test front. It is very pleasing to see some of the Indian greats of the past eras, viz., Viswanath, Umrigar, Manjrekar and Gavaskar appear in the top-20. They played in tough times and this has been recognised. Rounding this table in the 9th position is Andy Flower, one of the greatest modern batsmen ever, slightly benefiting from playing for a weaker team.

I have also given below the top 10 batsmen in terms of number of significant innings.

Table of batsman by number of significant innings

SNo Batsman For Mats Runs Inns SI % SI

1.Dravid R Ind 139 11395 240 108 45.0
2.Lara B.C Win 131 11953 232 106 45.7
3.Border A.R Aus 156 11174 265 103 38.9
4.Tendulkar S.R Ind 166 13447 271 103 38.0
5.Chanderpaul S Win 123 8669 210 98 46.7
6.Kallis J.H Saf 137 10843 231 94 40.7
7.Waugh S.R Aus 168 10927 260 92 35.4
8.Stewart A.J Eng 133 8465 235 90 38.3
9.Gavaskar S.M Ind 125 10122 214 89 41.6
10.Inzamam-ul-Haq Pak 120 8830 200 82 41.0
This is a quantity table. Dravid is on top with 108 performances and is followed by Lara with 106. Both are placed in the top-10 of the main table. Then comes the great fighter, Border and the incomparable Tendulkar with 103 significant innings. These four are the only batsmen to exceed 100 significant innings. Chanderpaul and Kallis should soon breach this number.


  1. awesome study man!! u really seem to be a great crazy fan of dravid like me! ( :

  2. u r so crazy abt dravid like me....nice analysis

  3. top class study, dravid is the best test batsman of his genration

  4. I am happy 2 see this...i too want icc 2 follow this type of pattern...i m damn fan od dravid

  5. Is this analysis available on Cricinfo?