Batting June 27, 2009

Test Batsmen Analysis: a follow-up

After looking at all comments to my previous article on best Test batsmen, I have come up with a revised set of tables which are a great improvement and should satisfy most readers

The follow-up to a major article is always fraught with pitfalls. One has to make sure that the changes are not just cosmetic, nor be influenced by a point only because it is made by the majority, nor knee-jerk reactions and finally must significantly improve the original submission. Each change has to be carefully considered and implemented. Hence, I have taken the time required to peruse all comments (over 700 in all), sift amongst these, pick up the meaningful and valid ones and come out with a revised set of tables which are a great improvement and should satisfy most readers. Let me summarise the changes below.

These changes are given in order of importance and the impact on the original submission.

1. The Match Performance points are divided by the number of innings played rather than the matches played. This will impact the calculations significantly and benefit players such as Richards who have played the second innings infrequently.

2. The Scoring rate measure has been dropped. This is again a significant change since it gets all the batsmen on an equal keel and is fair to all.

3. Instead a new measure, the Consistency index has been added. This information is available across years and for all the batsmen. Details of the calculations for this measure have been given elsewhere.

4. The weight for %-Team Score has been reduced from 10 to 5. This is fair to players who have played in relatively stronger teams. To those who have questioned this measure, for flimsy reasons, let me say that the highest value in this measure is that of Bradman, batsman extraordinary, in very strong Australian teams.

5. In Match Performance calculations, the Bowling quality measure is now Career-to-date instead of final career figures. This is also quite significant since the early Test figures for many bowlers is quite different to their career-end figures. The other benefit is that the Ratings figures calculated do not vary during subsequent calculations.

6. The Bowling quality is determined by a combination of Bowling Average and Strike Rate. This is based on Arjun Hemnani's excellent suggestion. This is fair to bowlers such as Waqar Younis, Marshall, Donald et al who are great strike bowlers but concede runs.

7. The Pitch Index calculations have undergone a very significant change. Now I am determining the Pitch index, not from the team scores and wickets, but using only the scores of the top 7 (or applicable) batsmen of each innings. This ensures that both the teams make their contributions to the index value. Also that the late order wickets do not distort the picture. I have also used the RpI rather than RpW. Makes lot more sense.

8. I had taken into account the relative team strengths in the Result parameter. Now I have extended this to the Home/Away parameter also. It means that instead of giving the benefit to the Away team automatically, now I take into account the relative team strengths. In other words, if Australia or India tour Bangladesh they will not automatically get the Away bonus. For Bangladesh touring, say, Sri Lanka, the Away bonus will be suitably increased.

9. "The Runs added with late order batsmen" measure's weight has been reduced from 1.00-1.30 to 1.00-1.20. This has been done to ensure the correct weight for the more important measures such as Pitch type, Bowler's quality et al.

10. Finally I have introduced a new measure called R-Factor. More on this later.

Consistency Index:

The Consistency index has been calculated as follows.

The career of each batsman was split into 5-Test slices. His 5-Test performance (Runs per innings used rather than Batting average so that the impact of not outs is negated) was measured against the Career RpI figures and the number of below-average performance slices (below 75%) used to determine the more significant part of the Consistency Index. 5-Test slices have been used since these represent a reasonable number to determine consistency. There is sufficient slack within 5 Tests to recover from bad form.

The other part of the Consistency index is based on the % of single digit dismissals. Together these two determine the Consistency record of the batsman.

The most consistent batsman is Alistair Cook of England, who has had no below-average slice and only 17.9% of single-figure dismissals. He gets an Index value of 4.28. Ross Taylor of New Zealand is also very consistent as, surprisingly, is Afridi. Amongst top batsmen, Hobbs and Sutcliffe are right at the top.

At the other end are Karthik, with 1.79 points, Wishart with 1.88 points, Richie Benaud with 2.05 points et al.

Separate tables for different eras:

I have also separated the tables into two independent ones. The first is for batsmen who started their career before 31 December 1959 and the other for batsmen who started their career after 1 January 1960. These dates have been decided after a lot of deliberations, summarized below.

I had earlier decided on 1 January 1940 as the cut-off date. Unfortunately very few Tests had been played upto that point (274 out of 1920) and there are not enough batsmen. Even 1 January 1960 cut-off does not give us enough Tests. However 483 Tests out of 1920 is a far better share.

The other key factor is that the 1950s (and some might say, the 1960s) really belonged to the old fashioned method of playing Test cricket and a Hutton or Barrington or Hanif Mohommad or Vijay Hazare would very easily fit in with the first era. Anyhow whatever date I take for cut-off there would be objections and this is a good enough point. It is also 50 years back.

I have also followed the separation very strictly, with debut match as the only criterion, knowing fully well that some players might have made their debut in 1958-59 but played most of their matches after 1960. Jarman who made his debut in Test no 483 (started on 19 December 1959) is placed in the first era while Durrani who made his debut in Test no 484 (started on 1 January 1960) is placed in the second era, and so on. I have to work on certain guidelines and have to be true to those. The number of players in the first era is a healthy 1124. The second era contains 1435 players.

I have also implemented another one of Arjun's suggections. That is to give a simple ratio between 2.0 and 0.0 against each batsman, based on a suitable mean, so that their position can be determined instantly and comparisons become easier. For this a value of 35.0 has been used as the notional mean (it does not matter what this figure is). Readers will instantly note the value of this single figure when they peruse the tables.

Let us now look at the revised tables.

The best Test batsmen: 1960-2009

No. Cty Batsman        Ratio  Total    Match  Bat   Runs Cons %-TS   R-Factor
Pts      Perf  Avge   Pts  Idx  Pts

1. Win Lara B.C 1.44 50.26 (22.63 10.43 11.93 3.37 1.90) 2. Ind Tendulkar S.R 1.41 49.24 (20.44 10.69 12.85 3.70 1.55) 3. Aus Ponting R.T 1.38 48.24 (21.54 10.85 10.88 3.54 1.44) 4. Ind Dravid R 1.31 45.98 (19.93 10.11 10.92 3.50 1.51) 5. Ind Gavaskar S.M 1.31 45.83 (20.52 10.02 10.12 3.49 1.67) 6. Saf Kallis J.H 1.30 45.65 (19.92 10.56 10.23 3.43 1.51) 7. Win Richards I.V.A 1.28 44.97 (21.81 9.90 8.65 3.11 1.50) 8. Aus Border A.R 1.28 44.83 (18.38 10.07 11.16 3.79 1.44) 9. Aus Waugh S.R 1.27 44.52 (18.35 10.12 10.90 3.86 1.28) 10. Slk Sangakkara K.C 1.26 43.98 (22.20 10.33 6.73 3.12 1.61)

11. Slk Jayawardene D.P 1.25 43.81 (20.59 10.00 8.15 3.49 1.58) 12. Pak Javed Miandad 1.25 43.62 (19.53 10.42 8.83 3.24 1.61) 13. Aus Hayden M.L 1.24 43.49 (20.77 9.83 8.54 2.93 1.42) 14. Pak Mohammad Yousuf 1.24 43.35 (21.36 10.60 6.81 2.98 1.60) 15. Pak Inzamam-ul-Haq 1.23 43.05 (19.39 9.71 8.91 3.56 1.47) 16. Aus Chappell G.S 1.23 42.91 (20.21 10.54 7.01 3.57 1.58) 17. Saf Pollock R.G 1.18 41.37 (22.20 11.88 2.22 3.42 1.66) 18. Win Chanderpaul S 1.18 41.21 (18.59 9.55 8.56 3.04 1.48) 19. Eng Gooch G.A 1.17 41.02 (18.85 8.45 8.75 3.41 1.56) 20. Saf Smith G.C 1.17 40.78 (20.14 9.46 6.39 3.31 1.49)

Lara continues to stay in no.1 position but his lead over Tendulkar has been considerably reduced (only around 2%). Ponting is at third position at a similar distance from Tendulkar. In fourth and fifth place are Dravid and Gavaskar. Then we get Kallis, who can ever deny the contributions he has made without fuss. Now comes Richards, probably correctly placed in the Top-10. He could have been in the Top-5 with no complaints. Then we have the two great Australian batsmen, Border and Steve Waugh. The incomparable Sangakkara completes the top-10.

Jayawardene follows next and then the fighter-extraodinary, Javed Miandad. I am happy that three top-class Pakistani batsmen, Miandad, Mohd Yousuf and Inzamam occupy 3 of the next 4 places, Hayden occupying the 12th place. Greg Chappell, Greame Pollock and Chanderpaul are correctly placed in the Top-20 which is completed by Graham Gooch and Greame Smith.

Lara's ratio is 1.44, Sangakkara's 1.26 and Greame Smith's 1.17. It can be seen that the top-10 batsmen have a spread of only 12.5% and the spread between 11 and 20 is only 7%. The only way to treat these tables is to look at these players as "First 1/2/5/10/20 ... amongst equals".

To view the complete list, please click here

Given below is the support table. The data is self-explanatory. For the two Consistency index related columns, explanations have been given below.

The best Test batsmen ever: 1960-2009 - Support data

SNo. Cty Batsman        Inns Rating  Runs  Bat  ( Adj) Consistency  %-TS
Pts         Avge          1     2

1. Win Lara B.C 232 50.26 11953 52.15 (0.99) 26.9% 26.7% 19.0% 2. Ind Tendulkar S.R 261 49.24 12773 53.46 (0.98) 25.0% 23.8% 15.5% 3. Aus Ponting R.T 221 48.24 10956 54.26 (0.97) 26.9% 20.8% 14.4% 4. Ind Dravid R 233 45.98 10823 50.54 (0.96) 29.6% 21.9% 15.1% 5. Ind Gavaskar S.M 214 45.83 10122 50.10 (0.98) 20.0% 25.7% 16.7% 6. Saf Kallis J.H 221 45.65 10277 52.79 (0.97) 30.8% 20.4% 15.1% 7. Win Richards I.V.A 182 44.97 8540 49.52 (0.99) 29.2% 25.3% 15.0% 8. Aus Border A.R 265 44.83 11174 50.33 (1.00) 25.8% 24.2% 14.4% 9. Aus Waugh S.R 260 44.52 10927 50.58 (0.99) 21.9% 24.6% 12.8% 10. Slk Sangakkara K.C 132 43.98 6764 51.65 (0.94) 37.5% 18.9% 16.1%

11. Slk Jayawardene D.P 167 43.81 8254 50.02 (0.94) 20.0% 22.8% 15.8% 12. Pak Javed Miandad 189 43.62 8832 52.08 (0.99) 32.0% 20.1% 16.1% 13. Aus Hayden M.L 184 43.49 8626 49.17 (0.97) 42.9% 19.6% 14.2% 14. Pak Mohammad Yousuf 134 43.35 6770 53.00 (0.96) 37.5% 22.4% 16.0% 15. Pak Inzamam-ul-Haq 200 43.05 8830 48.56 (0.98) 16.7% 23.5% 14.7% 16. Aus Chappell G.S 151 42.91 7110 52.70 (0.98) 17.6% 22.5% 15.8% 17. Saf Pollock R.G 41 41.37 2256 59.38 (0.97) 20.0% 24.4% 16.6% 18. Win Chanderpaul S 206 41.21 8576 47.76 (0.97) 33.3% 25.2% 14.8% 19. Eng Gooch G.A 215 41.02 8900 42.27 (0.99) 25.0% 24.2% 15.6% 20. Saf Smith G.C 135 40.78 6343 47.28 (0.94) 26.7% 22.2% 14.9%

No of below-average 5-Test slices Consistency 1 % = --------------------------------- Total number of 5-Test slices

No of single digit dismissals Consistency 2 % = ----------------------------- Total number of innings

To view the complete list, please click here

The best Test batsmen: 1877-1959

SNo. Cty Batsman        Ratio Total     Match  Bat   Runs Cons %-TS  R-Factor
Pts       Perf  Avge   Pts  Idx  Pts

1. Aus Bradman D.G 1.97 69.08 (36.62 19.35 6.91 3.70 2.50) 2. Eng Hobbs J.B 1.36 47.57 (23.93 12.34 5.49 3.99 1.82) 3. Win Sobers G.St.A 1.29 45.03 (20.67 11.48 8.03 3.28 1.58) 4. Eng Hutton L 1.27 44.37 (20.72 11.35 6.93 3.55 1.83) 5. Eng Barrington K.F 1.26 44.27 (20.97 11.71 6.81 3.08 1.70) 6. Win Headley G.A 1.25 43.86 (24.07 12.00 2.18 3.45 2.16) 7. Eng Sutcliffe H 1.25 43.62 (21.88 11.61 4.52 3.88 1.72) 8. Eng Hammond W.R 1.24 43.49 (19.78 11.27 7.31 3.43 1.70) 9. Win EdeC Weekes 1.22 42.69 (21.16 12.21 4.44 3.11 1.77) 10. Win Walcott C.L 1.16 40.67 (20.29 11.75 3.73 3.30 1.61)

11. Aus Harvey R.N 1.16 40.50 (19.22 9.92 6.18 3.56 1.62) 12. Win Kanhai R.B 1.13 39.44 (18.86 9.37 6.23 3.55 1.43) 13. Eng May P.B.H 1.12 39.14 (19.65 9.63 4.48 3.75 1.63) 14. Eng Cowdrey M.C 1.12 39.05 (18.00 8.75 7.65 3.22 1.43) 15. Eng Compton D.C.S 1.10 38.64 (17.88 10.06 5.70 3.43 1.57) 16. Saf Nourse A.D 1.08 37.94 (19.28 10.61 2.92 3.30 1.82) 17. Eng Dexter E.R 1.07 37.32 (18.15 9.45 4.51 3.75 1.46) 18. Aus Simpson R.B 1.06 37.12 (18.25 9.21 4.87 3.28 1.52) 19. Win Worrell F.M.M 1.06 37.10 (18.10 10.14 3.86 3.50 1.49) 20. Aus Morris A.R 1.03 36.10 (18.45 9.68 3.53 2.99 1.45)

Bradman is on top with a Rating value of 69.08 (and ratio of 1.97). He is followed, at a distance, by Hobbs and Sobers. Hutton and Barrington complete the top-5. The next 5 positions are monopolized by the West Indians, Headley, Weekes and Walcott and two great English batsmen, Sutcliffe and Hammond.

If we take Bradman's numbers away, the spread between 2 and 10 is a managable 14%.

I would appreciate if readers digest the tables before making the usual "xyz is better than abc" or "how can pqr be so low (or high)" or "". I will again repeat that intangible and non-measurable factors have no place in this analysis. This analysis has the heart of a cricket lover but the mind of a cricket analyst are behind it.

To view the complete list, please click here

Given below is the support table. The data is self-explanatory.

The best Test batsmen ever: 1877-1959 - Support data

SNo. Cty Batsman        Inns Rating  Runs  Bat  ( Adj) Consistency  %-TS
Pts         Avge          1     2

1. Aus Bradman D.G 80 69.08 6996 96.75 (0.97) 20.0% 17.5% 25.0% 2. Eng Hobbs J.B 102 47.57 5410 61.68 (1.08) 16.7% 12.7% 18.2% 3. Win Sobers G.St.A 160 45.03 8032 57.40 (0.99) 31.6% 19.4% 15.8% 4. Eng Hutton L 138 44.37 6971 56.73 (1.00) 25.0% 17.4% 18.3% 5. Eng Barrington K.F 131 44.27 6806 58.55 (1.00) 37.5% 19.8% 17.0% 6. Win Headley G.A 40 43.86 2190 60.02 (0.99) 25.0% 20.0% 21.6% 7. Eng Sutcliffe H 84 43.62 4555 58.04 (0.96) 18.2% 14.3% 17.2% 8. Eng Hammond W.R 140 43.49 7249 56.35 (0.96) 29.4% 17.1% 17.0% 9. Win EdeC Weekes 81 42.69 4455 61.06 (1.04) 30.0% 24.7% 17.7% 10. Win Walcott C.L 74 40.67 3798 58.75 (1.04) 33.3% 17.6% 16.1%

11. Aus Harvey R.N 137 40.50 6149 49.61 (1.02) 18.8% 21.9% 16.2% 12. Win Kanhai R.B 137 39.44 6227 46.84 (0.99) 25.0% 17.5% 14.3% 13. Eng May P.B.H 106 39.14 4537 48.14 (1.03) 7.7% 25.5% 16.3% 14. Eng Cowdrey M.C 188 39.05 7624 43.74 (0.99) 26.1% 25.0% 14.3% 15. Eng Compton D.C.S 131 38.64 5807 50.30 (1.00) 25.0% 20.6% 15.7% 16. Saf Nourse A.D 62 37.94 2960 53.07 (0.99) 28.6% 21.0% 18.2% 17. Eng Dexter E.R 102 37.32 4502 47.23 (0.99) 16.7% 18.6% 14.6% 18. Aus Simpson R.B 111 37.12 4869 46.04 (0.98) 25.0% 24.3% 15.2% 19. Win Worrell F.M.M 87 37.10 3860 50.71 (1.02) 10.0% 29.9% 14.9% 20. Aus Morris A.R 79 36.10 3533 48.42 (1.04) 33.3% 25.3% 14.5%

No of below-average 5-Test slices Consistency 1 % = --------------------------------- Total number of 5-Test slices

No of single digit dismissals Consistency 2 % = ----------------------------- Total number of innings

To view the complete list, please click here

The significant changes to the tables are summarized below. Most of these should make the tables more acceptable to many readers.

1. All batsmen are treated across years uniformly with the same set of parameters.
2. Consistency amongst batsmen has been recognized well. Note the high consistency figures of Tendulkar, Border, Steve Waugh et al.
3. Performances of lower ranked teams have been recognized more.
4. The quality of bowling faced has a much sharper definition. I may very well do a separate article on this fascinating aspect.
5. The gap between Lara and Tendulkar has narrowed to 2%.
6. Richards has moved up significantly.
7. Steve Waugh and Alan Border have moved up.
8. The three top Pakistani batsmen are reasonably well placed.
9. There are no major changes in the first era other than the revised set of batsmen included in this set.

R-Factor:

The points for all the measures add up to 90. The balance of 10 points has been reserved for R-Factor, expanding to Reader-Factor. The readers have complained that many points have not been taken into account. These points range from ridiculous, silly, absurd, obscure to relevant, sensible, valid and crystal-clear. Of course no analysis can take care of all such factors, especially as these are mostly intangible and non-measurable. Hence I have invented the R-Factor. It is your tool to be used the way you want. Convert the tables to Excel sheets, plug in your own R-Factor values and do what you want. Frame your results, circulate amongst yourselves and in general, have a ball. My only request to all is, whatever you do, do not send anything you have done on this to me.

You may, of course, ignore it completely.

Some of the factors I have been informed as not having considered are outlined below.

- Playing in a good team.
- Playing in a poor team.
- Expectations of a billion people.
- Coming from an island of population of 7500.
- Lack of support.
- Short pitched bowling.
- Lack of helmets, thigh guards, chest support etc.
- Injuries.
- Selectors' foibles.
- Terrorizing bowlers.
- Too much cricket.
- Too little cricket.
et al.

Only comments which add value to the article and derived conclusions will be published. Comments which are repetitive, say the same things ad nauseum, which are with the theme of "abc is the best, not pqr" type, which say, in different forms, "if you take away this measure, xyz will be on top", "abc is better than pqr because his average against ... is higher" type of comments, "there is no change from the earlier table" after a cursory 2-minute perusal etc will not be published. I gave a lot of leeway last time in publishing of comments. This time I will weed out such comments from the beginning. They are coming in the way of serious readers from appreciating the article and the user responses.

I want to emphasize once again, whether your comment is published or not is in your hands. Another important point. Anonymous comments will not be published.

My sincere thanks to Arjun Hemnani's whose quality ideas were the foundation for a number of these changes. My thanks to others like Jack (Jagdeep Singh), Ashik, Shankar et al.

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

Comments