This article is in three parts. The first is the main article extending the scope of player support to all the disciplines: batting, bowling and wicketkeeping. The second is a primer into the CTX methodology I have been using in various analyses I have done recently. The third is an analysis-based all-time best Pakistan XI, as asked for my long-time reader, Abdullah Khalid.
A. Total Support Analysis
In my last article I had done an analysis of the support received by the other batsmen in the team. There was an excellent response from the readers. The most important comments were posted by Mohanlal and couple of other readers who mentioned that the support should be looked at in its totality, including bowlers. This seemed to make a lot of sense. What is the point of having an excellent batting unit if the team has a weak bowling attack, as India has had often. Or the futility of a good bowling attack let down by the batting unit not putting enough runs on board. So I decided to extend the concept to its full extent: including batsmen, bowlers and the wicketkeeper. This will complete the fascinating exercise.
The initial cut-off turned out to be simple. I decided on 25 Tests since that represented a very decent Test career: maybe nearing 100 wickets or 2000 runs, or a combination of 1000 runs and 70-80 wickets; 609 players qualified. Now came the actual work. Easier said than done. This turned out to be one tricky process as I started work on it. Instead of explaining all the obstacles on the road, I will explain the selection process first.
I did the Total Support work for the first and second innings of the Test.
1. Determined the top six batsmen based on CTX_Batting_average, irrespective of the batting position the batsman batted in. This ensured that the nightwatchmen do not throw spanners into the works.
2. Determined the top four bowlers based on CTX_Bowling_average. Extra care was taken to ensure that non-regular bowlers do not take precedence over regular bowlers.
3. Identified the wicketkeeper: the one who kept in the first innings. Both his keeping prowess, represented through the only dependable measure through the years: "dismissals per match", and batting abilities were considered. I am not going to dilute this important measure with the mundane "byes conceded" measure. I believe strongly that byes are conceded due to mistakes of both bowler and keeper, or pitch defects. I have assigned 66.67% weight to the wicketkeeping skills and 33.33% to the batting ability. His value to the team is evaluated as a specialist.
4. Multiple alternatives
4.1.If the player is one of the six batsmen
Include data of the other five batsmen, four bowlers, keeper and divide by 10.
4.2.If the player is one of the four bowlers
Include data of the six batsmen, other three bowlers, keeper and divide by 10.
4.3.If the player is in both groups (Top A/R-Imran/Kapil/Botham)
Include data of the other five batsmen, other three bowlers, keeper and the seventh-best batsman and divide by 10.
4.4.If the player is not in either group (Average A/R-Yuvraj/Symonds)
Include data of the six batsmen, four bowlers and keeper & divide by 11.
4.5.If the player is the keeper
Include data of the six batsmen and four bowlers & divide by 10.
For batsmen, I take the CTX_Bat_Avge. I sum this across the six batsmen. As explained later, there are 34 instances in which the CTX_Bat_Avge values are greater than 100. For this exercise, I limit these values to 100.00. Let's face it, there is no great difference in the support factor whether a batsman is taken at 125 or 100. Terrific support is guaranteed, that is all.
For bowlers, I take the CTX_Bow_Avge and calculate the index value as 2.0*(60.0-CTX_Bow_Avge). I limit the CTX_Bow_Avge values to 55.0 if they have captured 10 or more CTX wickets and the average is high. If there are insufficient number of bowlers, dummy bowlers are created with CTX_Avge of 57.5. All these are done so that one Fred Spofforth or Richard Hadlee will not dilute the process. I sum this across the four bowlers. Again, it is of interest that if ever the CTX_Bow_Avge goes below 5.0, the bowler index value will cross 10.0. But then, there is no single instance so far, and will be taken care of, if ever it happens.
For wicketkeepers I take their "average dismissals per match" and multiply by 20 and assign the 66.67% weight. This is the only certain performance measure available. The highest figure so far is that of Adam Gilchrist, who snaffled 4.72 dismissals per match and garnered 62.94 points. The wicketkeeper is also evaluated as an allrounder. Accordingly, to this fixed value, the variable value of the batting, in the form of CTX values, is added.
I have explained these calculations using Test #1824: Australia v England, Boxing Day, 2006 at MCG.
The top six batsmen were Michael Hussey (1055 CTH runs at 95.91), Ricky Ponting (5431 at 65.43), Matthew Hayden (4350 at 61.27), Michael Clarke (672 at 51.69), Justin Langer (4355 at 48.93) and Brett Lee (647 at 29.41). The top four bowlers were Glenn McGrath (281 CTH wkts at 22.47), Stuart Clark (16 at 23.86 (actual 18.43, increased to career average)), Warne (310 at 26.59) and Lee (121 at 31.50). The index points for Adam Gilchrist is 79.03 (62.94 + 0.3333*48.29), 48.29 being Gilchrist's CTH_Bat_Avge.
The total index points for the six batsmen is 352.6 and for the four bowlers is 271.2.
For Langer, the TSI is 65.39 = (Batsmen (352.6-48.93) + Bowlers (271.2) + WK(79.0))/10.
For the non-batting-non-bowling Andrew Symonds, the TSI is 63.89 = (Batsmen (352.6) + Bowlers(271.2) + WK(79.0))/11.
For Gilchrist, the TSI is 62.38 = (Batsmen (352.6) + Bowlers(271.2))/10.
For McGrath, the TSI is 62.70 = (Batsmen (352.6) + Bowlers(271.2-75.1) + WK(79.0))/10.
We should not forget that having Hussey, with over 1000 home runs at a near-100 average was some support, even for the experienced batsmen. That tremendous support element is reflected in these numbers. Hussey's CTA average came down to 61.19, still a good figure, by the time he retired. Incidentally, Lee batted at No. 3 as a nightwatchman. But the way the process works, that fact is accounted for smoothly.
Now for the tables. First, the players who have played a minimum of 25 Tests with very high TSI values.
The first cut of this table had over 30 Australian players topping the table with no other team represented. Then I applied a tweak to give more weight to the bowlers' contributions. This created a more balanced calculation process. The table is topped by Symonds, the below-par allrounder who almost always had the support of six top batsmen and four bowlers. But his contribution, in the form of his fielding and the energy he brought to the team, has to be appreciated. Then come a slew of Australians between 1990 and 2010.
After six Australians comes Alviro Peterson, with the support of Graeme Smith, Amla, AB de Villiers, Dale Steyn, Morne Morkel, Vernon Philander et al. Then a few Australians come in, a sequence broken by another average South African player, Paul Harris. The table is completed by a brace of the Australians from the fag-end of the Bradman era. Ashwell Prince, of South Africa, completes the table. After the top 20, the South African players find their way in comfortably. This Australian-African formation is broken by Seymour Nurse.
Let us see what the average of just over 56 for Symonds in 26 Tests mean. It means that Symonds had the following collection of players in each of the Tests he played.
A keeper like Gilchrist, six batsmen with average batting average of 50 and four bowlers with average bowling average of 28. That is one privileged career. Maybe also a problem since Symonds was also the perennial second rung all-rounder. But Hayden had only slightly lesser support for over 100 Tests.
In general, a ball-park estimate is that an average TSI of over 50 is exceptional, above 45 is very good and around 40 is average. Anything below 35 should be considered very poor support.
Since two teams monopolised the career TSI table, I have created a separate table to show the players from each team with the highest career TSI values. We have already seen the top players for Australia and South Africa: Symonds and Peterson. Nurse is the West Indian with the highest TSI value. His contemporaries were Garry Sobers, Rohan Kanhai, Clive Lloyd, Conrad Hunte, Wes Hall, Charlie Griffith and Lance Gibbs. Jonathan Trott, of England, is the most recent player in this table. Everyone knows about the level of support he received. Qasim Umar is a surprise topper for Pakistan. Javed Miandad, Imran Khan and Wasim Akram would be the main reasons.
Yuvraj Singh leads the table for India. Sachin Tendulkar, Rahul Dravid, Sourav Ganguly, VVS Laxman, MS Dhoni, Anil Kumble and Harbhajan Singh were the main supporting cast for Yuvraj. Tharanga Paranavitana had a distinguished set of players in Kumar Sangakkara, Mahela Jayawardene, Muttiah Muralitharan and Chaminda Vaas playing with him. Tom Latham, with the support of Kane Williamson, Ross Taylor, Brendon McCullum, Tim Southee and Trent Boult is the top Kiwi player in this measure. Bryan Strang and Imrul Kayes complete the table for Zimbabwe and Bangladesh respectively.
These are the players whose career TSI values are very low. It is not a surprise that the table is dominated by players from Bangladesh. Khaled Mashud's and Habibul Basher's TSI values are around 35% of what Symonds and Hayden had.
These are the Tests in which the players clocked very high TSI values. It is no surprise that Australia has occupied the top 25 places. The first non-Australian player is Frank Worrell, in the 26th place.
Symonds had the very strong Australian team for company and support. Lindsay Hassett had august company in the Adelaide Test during 1948 against India: Sid Barnes, Arthur Morris, Don Bradman, Keith Miller, Neil Harvey, Don Tallon, Ian Johnson, Ernie Toshack and Ray Lindwall. This was Harvey's debut Test. Otherwise Hassett's TSI would have crossed 70. Clarke's support cast was Langer, Hayden, Ponting, Hussey, Warne, Lee, Clark and McGrath. Langer's was in the same match. Langer's next performance has already been explained.
In order to complete the article I have created with a table listing the career TSI figures for some important players, without any specific comments. I refer the readers to the TSI of George Headley and Hadlee. The better quality of bowling support for a lone batting warrior like Brian Lara is reflected in the slightly better number for him.
B. A primer on CTX values
Over the past two years or so I have used the CTX values a lot more than the career average values. I use the career averages to get an idea of the player's overall status, achievement of landmarks, creating "what-if" scenarios and so on. I tend to say "Smith has come below 60", "Voges is not going back to 90 again ever", "Williamson has just crossed 50" etc. For the older players I tend to use facts such as "Out of 291 qualifying batsmen, 37 have averages exceeding 50", "32 of the bowlers who have taken fifty or more wickets have bowling averages exceeding 40" etc.
However, it is clear that I have used the CTX averages more and more for different types of analyses I do. First, let me define CTX. It is a collective reference to CTD (Career To Date), CTH (Career To Date Home) and CTA (Career To Date Away). Whether it was the quality of batsmen dismissed, WQAI for bowlers, quality of bowling faced, players reaching pinnacle, support level for batsmen/players et al, I have used CTX values.
The reason is quite obvious. Most of the players have had careers which are characterised by ups and downs. Many players have started very well and have fallen off as the career progressed. Some examples are Viv Richards and Hussey. Some players have had very average starts to their careers and have really blossomed as they went along. Couple of examples are Murali and Graham Gooch. There is no comparison in numbers between the 20th and 50th Tests for Hussey and Murali. And within the Career To Date domain, home and away performances vary considerably. Miandad was a lion at home and not so fierce away.
A couple of years back I set out to create a comprehensive CTX database within my main database. Now I have available, for each of the 2226 Tests (as I write), 22 data segments (one for each player) containing the CTD/CTH/CTA values relating to batting/bowling/fielding. All normalsation required is done at the time of data segment creation so that I can use the values as these are presented.
There is no problem in the data values once a reasonable number of CTX Tests have been played, say ten. If a batsman has scored 800 runs in his first ten Tests and has an average of 65, I have absolutely no problem in accepting that. But I have to do something if a batsman averages 130 at the end of the third Test. Or a bowler averaging 6.5 for 15 wickets because of a terrific start. I have to do some clear tweaking and have explained these below.
If the number of CTD innings is below 15, the CTD batting average is capped at the career batting average, if CTD is higher.
If the number of CTH innings is below 10, the CTH batting average is capped at the career batting average, if CTH is higher.
If the number of CTA innings is below 10, the CTA batting average is capped at the career batting average, if CTA is higher.
Only one batsman, not surprisingly Don Bradman, is exempt from this tweak. Of course nightmare starts like Marvan Atapattu's will remain as they are. These tweaks will take care of outrageous starts such as the ones Mohammad Azharuddin, Doug Walters, Lawrence Rowe, Rohit Sharma et al had. These batsmen were nowhere near as good as their first two to three Tests indicated.
If the number of CTD wickets is fewer than 30, the CTD bowling average is increased to the career bowling average, if CTD is lower.
If the number of CTH wickets is fewer than 20, I increase the CTH bowling average to the career bowling average, if CTH is lower.
If the number of CTA wickets is fewer than 20, the CTA bowling average is increased to the career bowling average, if CTA is lower.
Of course nightmare starts like Shane Warne's or Aaqib Javed's will remain as they are.
These tweaks will take care of outrageous starts such as the ones Narendra Hirwani, Bob Massie and John Lever had. Not one of these bowlers was anywhere near as good as his first two to three Tests indicated.
I have given below some interesting facts on the distribution of averages to bring out the subtle difference between the career averages and CTX averages.
Each of the 2226 Tests had 12 top batting performances (siz for each team). This works out to 26,712. Out of these, 4108 CTX averages were 50 or above, making it 15.38%. Contrast this with the career averages. A total of 518 batsmen scored 1000 or more runs in their career. Out of these, only 44 (8.5%) had averages exceeding 50.0. This clearly shows that it is much easier to reach a 50-level CTX performance than at the career level. I have recognised this fact when I scaled the average into an index. Out of these 518, only two batsmen (Bradman and Voges (for the next few days)) had career averages exceeding 70. On the other hand, there were over 500 CTX averages of over 70.
In these Tests there were 15,529 qualifying bowling CTX performances. Out of these, 3121 (20.1%) had CTX bowling averages below 25.0. Contrastingly, out of the 376 bowlers who captured 50 or more wickets in their careers, only 64 (17%) had career averages below 25. The variation here is not as much as in the batting but is still significant.
Bradman reached a CTX average of over 100 in 17 of the 52 Tests he played in. Most of these were in away Tests. Surprisingly, Miandad reached a CTA average of over 100 in eight Tests. And Miandad has the honour of reaching the highest CTX average ever: 154.33 before Test #832 against India in Lahore during 1978. Bradman's highest CTX average is 145.60, before Test #197 against England in Manchester during 1930. In addition to these two batsmen, Harvey reached 100 in two away Tests, Sunil Gavaskar reached 100 in two away Tests, Charlie Davis of West Indies crossed 100 in three home Tests and Jimmy Adams reached 100 in two away Tests. This makes a total of 34 such 100+ achievements. So it is a very rare feat.
In contrast, 14 bowlers reached a CTX bowling average below 10.0. All these were pre-WW1 bowlers. Johnny Briggs achieved this feat five times, George Lohmann three times and Bobby Peel twice.
Recently, Hirwani chipped in with 24 wickets at 10.71, Bedser had a haul of 22 wickets at 10.82, Bill Johnston's 29 wickets cost him 10.38 and Massie's 21 away wicket cost 10.90 each.
C. The all-time best Pakistani Test team
Recently I did an article for Sportstar in which I selected an all-time best Indian XI, solely derived through analytical means. A reader wanted me to do a similar selection for Pakistan, who played their 400th Test recently. I have done that and presented the team below. The rating points are shown in parentheses.
Hanif Mohammad (65.57)
Saeed Anwar (62.89)
Younis Khan (73.16)
Javed Miandad (70.72)
Mohammad Yousuf (69.04)
Imran Khan (C) (66.08)
Kamran Akmal (75.91)
Wasim Akram (79.97)
Waqar Younis (75.18)
Saqlain Mushtaq (61.29)
Zaheer Abbas (61.71)
Saeed Ahmed (60.47)
Fazal Mahmood (67.49)
Shoaib Akhtar (65.64)
Saeed Ajmal (66.20)
I will only give the broad outline of the work done here. To get the full details, please refer to the article. For batsmen, the parameters used are runs scored, batting average, away batting average, average quality of bowling faced, batsman support index and average of hundred scores through career. For bowlers, it was wickets captured, bowling average, bowling strike rate, bowling accuracy, away wickets captured, away bowling average and average quality of batsmen dismissed.
Saeed Ajmal's figures are better than those for Saqlain Mushtaq. However, this is mainly due to the fact that all his wickets have been deemed to be captured away. I consider UAE as "away" for Pakistan. Hence I downgraded the points secured by Ajmal and selected Saqlain Mushtaq. But Ajmal deserves his place as a reserve spinner. Abdul Qadir's numbers just do not come anywhere near these two spinners. On a subjective selection, he might very well get the nod. Kamran Akmal's dismissals per Test figure of 4.3 is second only to Adam Gilchrist's. The others would have selected themselves even in a subjective selection.
Finally, any team that can afford to keep Zaheer Abbas, Fazal Mahmood and Shoaib Akhtar in the reserves bench better be a bloody good team.