|Photos||Video & Audio||Blogs||Statistics||Archive||Games||Mobile|
This is one of the most awaited analyses and has been in the pipeline for long. Many readers have asked for this analysis and finally I have been able to complete the same. I have given below a summary of how I have gone about this complex analysis process. This is based on my own observations and comments raised by a number of readers over the past few months.
1. There has to be equal weight given to Match performances and Career achievements. The match analysis should deal with the specific match-related measurable situations only.
2. The batsman's longevity related measure such as career runs scored has to be recognised, but at a lower weighting level, no more than 15-20% of the total.
3. There has to be a clear recognition of the quality of bowling faced by the batsman in every match. A hundred against Australia has to be valued at a much higher level than a hundred against Bangladesh or New Zealand in 1933.
4. The batting average should be considered the most significant of the career-related measures. However this has to be adjusted based on the par batting average(s) of the period(s) spanned by the batsman's career. A pre-WW1 batsman's average has to be adjusted upwards for the low batting averages prevalent during these period while a current batsman's average has to be adjusted downwards in view of the high batting averages prevalent. If a batsman's career spans multiple periods, there has to be proportionate adjustment.
5. The batsman's career strike rate has become an important measure. This should be recognized, if available. For those batsmen whose strike rate information is not available, this parameter will not be included. If the strike rate is available for part of the batsman's career, it will be considered for that part only. This is explained in detail later. There is a case for the innings strike rate to be incorporated in the Match Performance calculations. However this revised methodology necessarily requires the strike rate to be a career-based calculation rather than match innings based.
6. Non-measurable cricketing factors such as bouncers, helmets, uncovered pitches et al cannot be incorporated. Nice for healthy discussions, but not beyond.
A lengthy introduction, however the weighty topic required this. Now let us look at the details.
First a minimum criteria is to be established. I have decided to keep the minimum runs required at 2000 runs so that great batsmen such as George Headley, Greame Pollock, Ponsford, Macartney, Hazare, O'Neill, McCabe et al would come in for consideration. If I had raised the bar to 3000 runs, all these wonderful players would have missed out. It is also true that in a different era and political situation they would have played more Tests.
Now for the Ratings methodology.
The Match Performances would carry a weight of 50 points, based on the methodology explained below. Career Achievements will carry a weight of 50 points, based on methodology explained later.
The following factors are considered.
Base for calculation will be Runs scored. Other factors are explained below. Each of these is a multiplicative parameter, ranging either side of 1.00. For certain parameters such as result, home/away, runs added with late order et al, there would not be a below 1.00 value.
1. Pitch type.
2. Quality of bowling - weighted by actual balls bowled by each bowler.
3. Position at batsman entry (5 for 1, 100 for 2, 24 for 3, 325 for 4 et al).
4. Runs added with late order batsmen (no. 8 onwards).
5. Innings type (1/2/3/4, Score faced, Target et al).
6. Match result, taking into consideration relative team strengths.
7. Match location (Home/Away).
8. Match importance (Series status).
The points for each innings are computed, summed and divided by the number of matches played to arrive at the Match Performance Ratings value. The highest Match Performance Ratings value is 40.03 achieved by Bradman. George Headley is the next best in this category with 28.48 points followed by Lara with 27.31 points. Thus the limit of 50 we set has worked out well.
The points allocation, totalling to 50 points, is explained below.
1. Batting average: 20 points (for an adjusted average of 100.00).
Adjusted by the years spanned by the batsman career and the average Batting Average during the period. The adjustment is done proportionately. Most of the adjustment has been downward, between 1% and 6% (for the 2000s batsmen). The highest adjustment has been 20%, for batsmen such as Hill, Trumper whose entire career has been before Pre-WW1.
The highest adjusted average is that of Bradman, whose average of 99.96 has been adjusted down to 96.75. He gets 19.35 points and is followed by Hobbs (61.68) with 12.34 points and Weekes (61.06) with 12.21 points.
2. Runs scored: 15 points (for scoring 15,000 Test runs).
Away runs are given a slightly higher weight as explained. Scoring 1050 home runs gets one point for the batsman while 950 away runs will be enough to get one point. The differential of 10% seems very reasonable. This is the only longevity-based measure and carries a weight of only 15%.
This is a straight-forward calculation. Tendulkar is the highest and gets 12.85 points followed by Lara with 11.93 and Border with 11.16 points.
3. Scoring rate: 10 points (for a scoring rate of 100).
The full 10 points (and a total of 100 points) will be the base for batsmen like Sehwag, Hayden et al, whose career strike rate is available in complete. For batsmen such as Bradman, Hobbs et al, there will be no points taken, consequently the total points for consideration of the Ratings points % will be 90. For batsmen such as Lara, Tendulkar et al, proportionate points, out of 10, will be considered, consequently the total points for consideration of the Ratings points % will be between 90 and 100. This delicate tweak was suggested by Shyam (Ananthanarayan).
Among those whose complete data is available, Gilchrist leads in this measure with 8.20 points followed by Sehwag with 7.87 points and Pieterson with 6.32 points.
4. % of Team runs scored: 5 points (for scoring 25% of team runs).
Bradman is the leader in this measure, having scored 24.98% of the team runs. He gets 4.99 points, followed by Headley with 4.32 points and Lara with 3.79 points.
Now for the Top-20 table.
The best Test batsmen of all time
No.Cty Batsman Mat Total Rating Match BatAvg Runs BatSR %-TS Max Pts Pts Perf Pts Pts Pts Pts Pts
1.Aus Bradman D.G 52 (71.27) 79.19 40.02 19.35 6.91 .... 4.99 90.00 2.Win Lara B.C 131 (58.44) 59.40 27.31 10.43 11.93 4.98 3.79 98.38 3.Ind Tendulkar S.R 159 (53.68) 54.59 22.43 10.69 12.85 4.60 3.11 98.33 4.Aus Ponting R.T 131 (54.46) 54.46 23.95 10.85 10.88 5.91 2.87 100.00 5.Eng Hobbs J.B 61 (48.53) 53.93 27.07 12.34 5.49 .... 3.64 90.00 6.Win Sobers G.St.A 93 (48.14) 53.49 25.48 11.48 8.03 .... 3.16 90.00 7.Eng Hutton L 79 (47.78) 53.09 25.85 11.35 6.93 .... 3.66 90.00 8.Ind Gavaskar S.M 125 (47.40) 52.51 23.81 10.02 10.12 0.11 3.35 90.28 9.Win Headley G.A 22 (46.98) 52.20 28.48 12.00 2.18 .... 4.32 90.00 10.Ind Dravid R 134 (51.80) 51.80 23.58 10.11 10.92 4.17 3.02 100.00
11.Aus Hayden M.L 103 (51.71) 51.71 24.47 9.85 8.54 6.01 2.84 100.00 12.Win EdeC Weekes 48 (45.83) 50.92 25.65 12.21 4.44 .... 3.53 90.00 13.Saf Kallis J.H 131 (50.73) 50.73 22.51 10.56 10.23 4.42 3.01 100.00 14.Eng Barrington K.F 82 (45.63) 50.70 23.72 11.71 6.81 .... 3.39 90.00 15.Aus Border A.R 156 (46.62) 50.40 21.50 10.07 11.16 1.01 2.89 92.50 16.Pak Mohammad Yousuf 79 (50.37) 50.37 24.50 10.60 6.81 5.26 3.20 100.00 17.Slk Sangakkara K.C 80 (50.09) 50.09 24.23 10.33 6.73 5.59 3.21 100.00 18.Aus Chappell G.S 87 (45.27) 50.04 24.31 10.54 7.01 0.24 3.17 90.48 19.Ind Sehwag V 69 (49.69) 49.69 23.59 9.40 5.77 7.87 3.06 100.00 20.Win Richards I.V.A 121 (44.67) 49.39 22.81 9.90 8.65 0.31 2.99 90.45
Any doubts as to the position of Bradman at the top should disappear after this analysis. Bradman is the best in three measures (Match Performance, Batting Average and % of Team Score). He is ahead of the next best batsman by a whopping 25%. A series average of 57 by Bradman was considered to be a failure. It is necessary to agree, once and for all, that Bradman is and was the best Test batsman who ever played. Note the qualification, "Test", however. Who can forget the 334, 304, 270, 173* that Bradman scored.
Lara's second position again should not surprise any one. Playing for a weak team, mostly with inconsistent support, he essayed some of the best Test innings ever. 153*, 213, 277, 375/400 should figure in anybody's list of Top-20 innings. He also scored at a fair pace. Finally has there ever been a better batsman's Test series than the one, away, against Sri Lanka when he scored 688 runs, all in a losing cause. Lara is way behind Bradman but is very comfortably ahead of the next batsman.
The third and fourth positions should be looked together. Tendulkar and Ponting are separated only in the second decimal, that too because of Ponting's below-average series aginst South Africa. Two wonderful batsmen, capable of saving or winning Tests, fully deserving of their high place in this table. If Tendulkar is a more all-round batsman with an outstanding technique and no weakness, Ponting is the more attacking batsman but with a known weakness against top class spinners.
Hobbs is in fifth position, deservedly so, no doubt aided by the upwards revision of his batting average. He played on a number of bowler-friendly pitches and it is difficult to think of a better opener other than the one who appears slightly below him.
The top-10 is completed by Sobers, Hutton, Gavaskar, Headley and Dravid. Every one of these batsmen deserves his position. Headley is in the Top-10 despite playing only 22 Tests and scoring 2190 runs. This is a vindication of the Ratings methodology in that a batsman who has scored only a fifth of the runs the others have scored can still come into the Top-10.
Richards' position at no.20 might be questioned by some. The problem is that Richards, with his carefree attitude alternated great performances with very average performances in his Test career. This is clearly shown in the Match performances points, quite low at 22.81. Also his adjusted average is below 50. Finally a simple indicator is the fact that he has taken 121 Tests to score 8540 runs (71 runs per Test) as compared to Lara, 131 Tests and 11953 runs (91 runs per Test) or Hayden, 103 Tests and 8625 runs (84 runs per Test). It is also true that he never faced the most fearsome bowling attack at that time.
Similarly Hammond's 22nd position (Pieterson is at no.21) must also be looked at with surprise. This is explained by the fact that a fair proportion of his runs were scored against very weak New Zealand attacks and average South African and West Indian attacks.
To view the complete list, please click here.
The support information for the Top-20 batsmen is given below.
The best Test batsmen of all time: Support data
SNo. Cty Batsman Mat Rating Runs Batting (Adj) BatSR %-TS Career Pts Average (%) BowQtyThe values are self-explanatory. The Batting Average shown is the adjusted figure and the figure in brackets shows the adjustment done. To view the complete list, please click here
1. Aus Bradman D.G 52 79.19 6996 96.75 (0.97) .... 25.0% 36.1 2. Win Lara B.C 131 59.40 11953 52.15 (0.99) 59.4* 19.0% 35.4 3. Ind Tendulkar S.R 159 54.59 12773 53.46 (0.98) 55.2* 15.5% 37.1 4. Aus Ponting R.T 131 54.46 10956 54.26 (0.97) 59.1 14.4% 37.0 5. Eng Hobbs J.B 61 53.93 5410 61.68 (1.08) .... 18.2% 35.7 6. Win Sobers G.St.A 93 53.49 8032 57.40 (0.99) .... 15.8% 33.4 7. Eng Hutton L 79 53.09 6971 56.73 (1.00) .... 18.3% 38.4 8. Ind Gavaskar S.M 125 52.51 10122 50.10 (0.98) 38.5* 16.7% 35.6 9. Win Headley G.A 22 52.20 2190 60.02 (0.99) .... 21.6% 32.6 10. Ind Dravid R 134 51.80 10823 50.54 (0.96) 41.7 15.1% 37.0
11. Aus Hayden M.L 103 51.71 8626 49.27 (0.97) 60.1 14.2% 37.3 12. Win EdeC Weekes 48 50.92 4455 61.06 (1.04) .... 17.7% 36.1 13. Saf Kallis J.H 131 50.73 10277 52.79 (0.97) 44.2 15.1% 37.1 14. Eng Barrington K.F 82 50.70 6806 58.55 (1.00) .... 17.0% 36.0 15. Aus Border A.R 156 50.40 11174 50.33 (1.00) 40.5* 14.4% 34.0 16. Pak Mohammad Yousuf 79 50.37 6770 53.00 (0.96) 52.6 16.0% 38.7 17. Slk Sangakkara K.C 80 50.09 6764 51.65 (0.94) 55.9 16.1% 39.4 18. Aus Chappell G.S 87 50.04 7110 52.70 (0.98) 50.8* 15.8% 33.4 19. Ind Sehwag V 69 49.69 5757 47.02 (0.94) 78.7 15.3% 36.0 20. Win Richards I.V.A 121 49.39 8540 49.52 (0.99) 68.1* 15.0% 33.8
'*' means only part strike rate information is available.
One final note on the quality of bowling faced. For correct cricketing reasons I had incorporated this measure in the "Match Performance" segment so that a 100 scored against a strong Australian attack would carry a lot more weight than a similar innings against a weak Bangladeshi attack. However I have also done another complicated exercise, to provide an analytical answer to those who had mentioned that Bradman faced weak bowling attacks.
I considered every innings played by a batsman, multiplied this by the weighted bowling quality for the particular innings, summed this value and divided by the total number of runs scored. There cannot be a better indicator of the quality of bowling faced than this. The last column above indicates this weighted bowling quality indicator for the top-20 batsmen.
It is clear that Bradman faced as good a bowling attack, on an average, as any one else in the Top-10. In fact he faced a better bowling attack than Tendulkar, Ponting, Hayden and Dravid. So that argument should be put to rest.
Amongst the top batsman the batsman who faced the most powerful bowling attack was Graham Gooch, Alec Stewart, Atherton et al, around 30. This is quite understandable, considering the quality of West indian bowlers Gooch faced.
The least powerful attack was faced by Ames, Hammond et al, with around 45. If one looks at the 1932-33 New Zealand attack (sum total of 29 career Test wickets) we can understand this. Then come Sangakkara, Atapattu and Jayasuriya. Understandable considering the tons of runs scored by them against Bangladesh and Zimbabwe.
To view the complete bowling quality list, please click here
If an analysis is made combining both Tests and ODIs, which I would do later, it is almost certain that Tendulkar would be no.1, although I expect that Richards, Lara and Ponting would run him close. We obviously have to exclude great batsmen such as Bradman, Hobbs, Sobers et al from this exercise.
Based on the requests from readers an illustration of the Match Performance Ratings calculation will be posted at the earliest.
Match Performance Ratings - Illustrative example
This innings is one of the top-3 innings ever played (might even be the best ever).
Laxman V.V.S Ind vs Aus, Kolkatta, 2001 Runs scored: 281
Base points: 70.25 - Runs multiplied by 0.25. Really does not matter, can be anything. 0.25 has been selected since there is no need to do any scaling up/down later.
Bowling quality: 1.109 (Range: 0.75 to 1.25) - Weighted bowling average determined by summing product of bowler's average and balls bowled in the match by each bowler and dividing by team balls. - Strong Australian attack. Well above average.Match Performance Rating points: 110.88 (70.25 x 1.5784).
Pitch type: 0.899 (Range: 0.80 to 1.20) - The pitch characteristics for the concerned year. - Very good batting pitch.
Entry: 1.075 (Range: 1.00 to 1.15) - Complex formula based on the innings position at batsman entry. 1/2 innings dealt different to 3/4 innings. Openers dealt with in a special manner. - 3rd innings, 222 in arrears at 52 for 1.
Match situation: 1.30 (Range: 1.00 to 1.30) - Various conditions such as which innings, arrears faced, target in front of team used. Notional target in case of 3rd innings and Actual target in case of 4th innings. - Follow on 274 behind - Still in arrears.
Runs with Late order: 1.00 (Range: 1.00 to 1.30) - Runs added with batsmen 7-11. - No runs with late order. No change.
Win bonus: 1.09 (Range: 1.05 to 1.10) - Result after considering the relative team strengths. - Win against very strong team - 21% better.
Home/Away: 1.00 (1.00 to 1.05) - Away bonus only if win is against good teams, not weak teams (e-g), Ban/Zim/Nzl(1920s) et al. - Home. No change.
Series status: 1.04 (Range: 0.98 to 1.06) - Complex analysis based on series status. Maximum for deciding tests and minimum for dead rubbers. - 3 test series, trailing 0-1.
For each batsman, the MPR points are added and divided by the number of innings played to arrive the Match performance Ratings Index.
If anyone thought that I sit and do these individually (and so can modify figures as I wanted), think again. There have been 67438 innings played until now. I have complex computer algorithms for each of these parameters and these are automatically worked out with no manual intervention.
Just to give the readers an indication of the validity of method let me compare two innings by Lara (same batsman chosen to avoid 20 comments). The 400* gets a MPR value of 65.20 points while the 213 carries a MPR value of 73.83 points.
If you are interested in perusing the "C" program code for just one index, just mail me. I will send you the code. It is not made available to all in view of the highly technical nature of the same.
Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systemsFeeds: Anantha Narayanan
© ESPN EMEA Ltd.
|Comments have now been closed for this article
Anantha spent the first half of his four-decade working career with corporates like IBM, Shaw Wallace, NCR, Sime Darby and the Spinneys group in IT-related positions. In the second half, he has worked on cricket simulation, ratings, data mining, analysis and writing, amongst other things. He was the creator of the Wisden 100 lists, released in 2001. He has written for ESPNcricinfo and CastrolCricket, and worked extensively with Maruti Motors, Idea Cellular and Castrol on their performance ratings-related systems. He is an armchair connoisseur of most sports. His other passion is tennis, and he thinks Roger Federer is the greatest sportsman to have walked on earth.