First, on the terminology used in these pieces: GW25 (Golden Willow 25) and RC25 (Red Cherry 25) have been established in the earlier articles in the series. I will use GW25-100 and RC25-100 to represent the Top 100 tables. I will refer to the 3200-odd entries (that have rating points greater than 400) as GW25-Selection or RC25-Selection.
I have created an Excel sheet containing the GW25 performances over 500 points and this can be downloaded here.
This particular article - on the batting performances - contains follow-up material to the GW25 features. It will also include additional analyses tables using the GW25 rating points. The second part will cover RC25 and additional analyses for the bowlers and allrounders.
Possible surprises in GW25
Sanath Jayasuriya's No. 2 position is a surprise. This innings came into the top three years back when I made my last major changes to the process that generates the ratings, and then it just stayed there, refusing to move out of the top three. How often have I checked the values? It is a clear case of the sum being less than its parts. I finally developed a grudging admiration for the stupendous innings, but although I now feel it certainly deserves a GW25 inclusion and a top-ten place, No. 2 is probably way too high. But let us give the process the due recognition and the innings that came with the process its due respect. Let us take this as atonement for the 340 monstrosity. A declaration at lunch on the last day in that match could have given India a few uncomfortable hours. However, Sri Lanka batted on, on and on.
I was pleasantly surprised by the inclusion of Dennis Amiss' 262 not out (No. 21), which arguably is the greatest match-saving innings ever played. The bowling attack wasn't great and the pitch was a belter, but to carry the bat for ten hours and add so many runs for each of the late-order wickets is something special.
The dropping of Clem Hill (to No. 15) by about ten places surprised me. I had always thought that this innings would never go out of the top ten.
Then the last three entries in the top 25 were surprises for different reasons: Brendon McCullum's double-hundred in Auckland, because of the home win and an average Indian attack, Rahul Dravid's 270 in Rawalpindi, because of Pakistan's bowling quality, and Gordon Greenidge's 214 not out at Lord's, because of the ease of the win. But we should not forget that the 25th-placed innings has 744 points and the 40th-placed innings 730 points. A mere 14 points separates 15 places.
1. There are 17 Tests with five GW25-Selection performances apiece. The Kolkata Test between India and South Africa in 1996 had the highest tally of points from these five performances - 2743 points. Also, only the 2014 Centurion Test between South Africa and Australia contained four performances each exceeding 500 points: by Shaun Marsh, Steven Smith, AB de Villiers and David Warner. And in the 1997-98 Durban Test between South Africa and Pakistan, there were two performances totalling 1468 points: Azhar Mahmood's 132 (809 points) and Saeed Anwar's 118, with 659 points.
2. The Bodyline series created a record for the maximum number of GW25-Selection performances in a series by one team, with 13 by England; Wally Hammond contributed four himself. The Australia v West Indies series in 1975-76 set a record for most performances in a Test series: 11 for Australia and nine for West Indies.
3. The best performance by a batsman in a series was by Clyde Walcott, who had five GW25-Selection innings, all hundreds, in the home series against Australia in 1955. Three batsmen - Don Bradman in 1930, Hammond in 1932-33, and Len Hutton in 1950 - had four performances each. As I write this article, Virat Kohli has three such innings in the current series in England, in the first four Tests, and no one can really bet against him securing at least one more such performance, if not two, at The Oval.
4. The highest placed triple-century was Inzamam-ul-Haq's 329 in Lahore against New Zealand, which got 730 points. The Pitch Quality Index (PQI) was 47.5, mainly because of New Zealand's miserable total of 73. However, there is no doubting the quality of Inzamam's innings. He came in at 57 for 2 and added 133 for the last three wickets. At the other end of the spectrum, Jayasuriya's 340 in the run fest in Colombo in 1997 against India fetched only 424 points and is the lowest-ranked triple-hundred. The PQI was 89, the bowling was pedestrian, and the declaration meant that only two innings were completed.
5. The lowest-placed double-centuries came in the Faisalabad Test between Pakistan and Sri Lanka in 1985-86: Javed Miandad's unbeaten 203 (287 points) and Qasim Omar's 206 (283 points). Roshan Mahanama's 225 in the Colombo Test against India gathered 321 points. In the 2010 Colombo Test against India, Kumar Sangakkara's 219 earned him 335 points and Sachin Tendulkar's 203 fetched him 347 points.
6. Centuries amounting to little: Aravinda de Silva's 126 in that Colombo Test against India earned him 211 points. Dinesh Chandimal's unbeaten 116 against Bangladesh in Galle in 2012-13 secured only 194 points. In 2005-06 in Lahore against India, Kamran Akmal earned 195 points for his 102 not out and Shahid Afridi 196 for his 103. These two were pretty quick innings but balls-faced data is not used in GW25 data.
7. The highest-placed 20: Australia's John Ferris walked in at 18 for 7 against England at Lord's in 1888 and made 20 to take Australia to a total of 60. In a match in which the innings scores were 116, 53, 60 and 62, this seemingly small innings was like the Cullinan diamond. It secured 278 points, more than 112 hundreds (including a score of 174).
8. In the GW25-100 table, there are 77 winning and nine losing innings, and 14 performances from drawn matches. The fact that one in four of this elite list appears in matches not won is ample evidence that the influence of the result on the final points tally is not as high as it is made out to be.
9. Points per run: The 82,083 innings analysed had an average of 100.0 points per innings. The average points per run was 3.79.
The most points earned per run was by England's Johnny Tyldesley, who scored 62 in Melbourne in 1903-04 and secured 533 points (8.60 points per run). This was out of a team total of 102. Against England at Trent Bridge in 1981, Allan Border walked in at 33 for 4 and scored a match-winning 63 at 8.49 points per run (535 points). At The Oval in 1999, Chris Cairns came in at 38 for 6, scored 80 at 8.27 (661 points) and took his team to a match-winning total of 162. Hutton got 509 points for his unbeaten 62, which he made after coming in at 30 for 6 to take England to 122 in Brisbane in 1950-51.
The lowest points-per-run figure was in that 1997 Colombo Test. Jayasuriya amassed 340 runs at 1.25 points per run (424 points). In this Test and in the 1985 Faisalabad one, also featuring Sri Lanka, there were quite a few other innings at a points per run value below 1.50.
Readers have suggested quite a number of tweaks to the GW25 process. I considered them all and have identified one, suggested by Giri, a long-time reader. This is to have two PQI values: one for the first two innings and another for the last two innings of a match. This makes a lot of sense considering there is a clear difference in the averages for the four innings. The overall average RpW for the four innings are respectively 32.8, 31.2, 28.0 and 26.8. There is a clear 10 to 15% difference between the first two innings and the next two innings. So I will compute PQI-1 and PQI-2 for the match and allot the PQI parameter based on these values in the next iteration of the ratings.
One other tweak that I might consider is to factor in the margin of loss in the Location-Result-Team-Strength parameter. That means that a defeat by seven runs will carry more points for allocation than, say, a loss by 200 runs. I might make the loss calculation a more continuous curve and be more sensitive to the final margin. However, let me clarify that these will be done only from the Lord's Test of the ongoing England-India series. I have already released the GW25 tables and these are set in stone, barring a major release, maybe a few years down the line.
Additional GW25 tables
The following two tables could not be accommodated in part one of the article.
It would have been a huge surprise to see anyone other than Bradman at the head of the frequency table. One performance every 3.2 innings is approximately one every other Test. Readers should remember that we are talking about a 400-point-level performance: maybe a good hundred. However, what is surprising is that George Headley is so close to Bradman despite a significant gap in batting averages between the two. Headley made a top-level performance once every four innings, which could very well be every other Test considering that he played for an average team.
Then comes a most pleasant surprise: Kohli is in third position with a GW25-Selection performance of once every 5.6 innings. He has 12 performances in the last 34 innings, which is positively Bradmanesque. If he continues in similar vein he could comfortably move into second place.
Everton Weekes is next with one every 6.2 innings. Those who followed cricket in the 1950s, or even read about that period now, will know how dominant Weekes was. The top five is rounded off by another modern giant, Steven Smith, with one performance every 6.5 innings. Let us all hope that he comes back from his suspension period as good as he was before. Finally, Kane Williamson makes it to the top 12, which includes the two great West Indian batsmen Brian Lara and Viv Richards.
This next table covers the GW25-Selection performances by players.
This looks at the total number of GW25-Selection performances by players, rather than the frequency. It is no surprise that Sachin Tendulkar, with 200 Tests, leads this table, with 34 such performances. The frequency is quite high for most of these batsmen. Only Lara, Richards and Bradman have sub-8.0 frequency values. Ricky Ponting shares Lara's collection of 32 performances, albeit in 55 innings more.
Jacques Kallis follows with 30 GW25-Selection performances, but with a high frequency of 9.33. Kumar Sangakkara is in fifth position with 30 such efforts. Richards leads a quartet of batsmen with 26 performances, and has the lowest frequency among this group.
Now we come to the three main add-on analyses. The first table is an extension of the GW25 table and orders the performances by percentage of the Team Rating points.
|Team||Inns||Year||Res||Vs||Runs||Pts||Team score||Team pts|
John Reid tops this table with a stirring performance in a losing Test, in which his hundred secured over 75% of the team rating points. The second-placed innings, by Moin Khan, is another one in a losing cause. It is not surprising that Charles Bannerman's first-day hundred in the first-ever Test is in third position, with just over 72% of the points. Then comes another losing innings, this time by Inzamam. In fifth place is Michael Slater's magnificent match-winning 123 out of 184 in Sydney. Graham Gooch's 154 finds its place in this table as the only other innings from the GW25 collection.
No fewer than 13 of the 20 innings have been played in losing causes, indicating that it is easier to shine when the chips are down. However, I must inform readers that I have excluded innings in which the team lost five or fewer wickets. I didn't want a score of 80 in a match-winning fourth-innings total of 110 for 2 to sneak in.
Incidentally, Kohli's 149 at Edgbaston secured 64.8% of the team points, and he just missed making the cut here. However, if the entire match is considered, Kohli's two performances secured 915 points, which worked out 52.3% of the team total. He is fifth on the all-time list.
The next table is an analysis of the match performance of batsmen from the point of view of GW25 performance points.
There are surprises galore in this list. Jack Russell, not the wicketkeeper who went on to be an artist but the English opener who played nearly 100 years ago, tops the table. In Durban in 1923, he scored 140 out of England's first-innings score of 281 and 111 out of the second innings of 241. This included a last-wicket partnership of 92. These two innings totalled 1401 points and is a GW25-100 innings.
The second-placed Test performance is a well-known one to all followers, and especially Indian ones - VVS Laxman's 281 in Kolkata against Australia in 2000-01 will probably never fall off the top-ten lists, and here it gathers 788 rating points. However, let no one forget the importance of Laxman's first-innings minor masterpiece of 59, which fetched 390.1 points. These dual performances get him into second position.
At Old Trafford in 1976, West Indies made 211 in their first innings. Greenidge scored over 60% of this total, with a magnificent 134, which fetched him a GW25-50 placement and secured 725 rating points. In the second innings, he scored 101 runs in much easier circumstances and this fetched him 442 rating points.
Steve Waugh's two hundreds at the same venue in 1997 fetched upwards of 500 points each and got him the fourth place. Here also, the first-innings effort of 108 was an invaluable innings. Next up is Ian Botham's 1981 Headingley performance, similar to Laxman's in that the second-innings effort was bolstered by the invaluable 50 in the first innings.
A few other innings in the top 20 are worthy of mention. Lara's innings of 221 and 130 in Colombo in 2001-02 were made in a losing cause and is the only such entry in the table. Dimuth Karunaratne's performances came in the recent Test in Galle. Virender Sehwag makes it to the top 20 for his epochal 201 not out and a useful 50 in Galle. Gooch's 333 in the first innings at Lord's against India in 1990 helped him get into this table.
Now, for a career analysis. Let me make it clear that this is not an analysis on who is the best batsman in Test history. (But whenever I do that analysis, this will be one of the parameters.)
I have divided the total GW25 Rating points secured by the batsman by the Weighted Innings - much fairer than the innings or dismissed innings. As readers will know, the Weighted innings is determined by assigning an innings weight of between 0 and 1.0 for each not-out innings, based on the basic Runs per Innings (RpI) value. If the RpI is 40, an innings of 12 not out will get a weighted value of 0.3, innings of 27 not out - 0.675 and 75 not out - 1.0. These are then added and the Weighted innings arrived at. This is a fair and accurate determination of the denominator for all computations. The normal method of using dismissed innings will include the points but ignore the innings count for innings such as 400 not out, 248 not out and so on.
Who else but Bradman at the top. He averages 271 points per innings. Some distance behind comes Jack Hobbs, 85 points behind. A few decimal points after comes Steven Smith. Since he is still an active cricketer, there is a good chance that he could overtake Hobbs.
Headley's short but magnificent career proves that his fourth place is well earned - though it is true that it would be easy to maintain high levels across a short career. He is followed by Herbert Sutcliffe.
The top-ten table is followed by a quintet of top batsmen: Walcott, Weekes, Ken Barrington, Hutton and Lara. It is interesting to note that Kohli is knocking at the doors of the top ten and might very well go past Lara in the next few Tests.
Before I conclude, I would like to dwell on some comments from readers on the previous articles in this series.
1. My system rewards the team that wins (and indirectly the players who are part of the team), just as it happens everywhere - football World Cup football, cricket World Cup, IPL, tennis, golf, Olympics etc. But for the rewarding of wins, Azhar Mahmood's 132, Neil Johnson's 107, Gundappa Viswanath's 114, Graham Thorpe's 119, Kim Hughes' 100, John Graham's 105 etc would all slide down the table. It will be a cascading effect.
2. It should not be forgotten that Kohli's innings, in a losing cause, is still placed just outside the top 100, a few hundred places above Lara's quadruple. Kohli himself has said that his innings lost a lot of value because the team did not win.
3. Lionel Messi could score three goals and convert his penalty, but if his team-mates concede more goals than he scores, Argentina will still take the early flight back to Buenos Aires. One man cannot win a match. Cricket is also like that.
4. Someone said that in the 1999 Chennai Test against Pakistan, the Indian tail were "walking wickets". That's completely off the mark. Sunil Joshi, Anil Kumble, Venkatesh Prasad and Javagal Srinath were, on paper on that day and on any other day, better than Nehemiah Perry, Curtly Ambrose and Courtney Walsh, the tail that Lara guided West Indies to that famous triumph in Barbados against Australia. What happened was that the Indian late order failed and the West Indian late order succeeded. Maybe fortune favoured the West Indians. That is part of life and cricket. Maybe Lara gets the credit for inspiring his less-talented late-order batsmen to walk on water.
5. Readers should stop comparing two innings. The circumstances, context and the result will be totally different, and subjective comparisons do not hold water. This analysis has been worked on for years and has involved many cricketing specialists. It is not an overnight creation. I will not change things just because a few readers disagree.
6. Please keep your favourite innings in your heart but do not expect the computer analysis to recognise all these feelings within you. I can admire many such efforts with my heart but will not let these feelings interfere with the mind-based work I do. While admiring Tendulkar's 136 in Chennai, which is among the top five Tendulkar innings across formats, I also know why the 248 rates much higher.
7. I will never have an X-factor in my analyses. Leaving aside the subjective nature of this approach, who will populate this parameter?
8. Imagine this hypothetical scenario: July 2019, the World Cup final. South Africa play New Zealand. South Africa scores 300 for 5, with AB de Villiers getting 110 in 95 balls. The next best score is 57. New Zealand respond with 300 for 5, with Martin Guptill getting 110 in 95. The next best score is again 57. Both get a life and a DRS reprieve each. It so happens that the bowling numbers are identical. In the tie-breaker, South Africa win without these two batsmen since both are nursing a sprain. De Villiers will get a share of the bigger prize money, the bigger medal, a bottle of champagne for the Man of the Match, a cheque, a $10,000 watch, and three-fourths of the accolades. Guptill will get a share of the smaller prize money, the smaller medal, and one-fourth of the accolades.
Is it unfair? Of course not. De Villiers is rewarded because his team won. No one in their right mind would say that Guptill gets punished. Of course, Guptill will get a lot of sympathetic comments and that is fine. There is no difference between this scenario and my analysis. In my case, it is a few places up or down the table. In real life, it may be a million dollars.
Readers can comment on this article here. Note that this link takes you outside ESPNcricinfo, and ESPNcricinfo does not monitor, publish, or otherwise associate itself with comments posted there.
Anantha Narayanan has written for ESPNcricinfo and CastrolCricket and worked with a number of companies on their cricket performance ratings-related systems