December 20, 2010

Test innings: a different peer-view

An analysis of how batsmen have fared when compared to their peers in Tests.
83

Recently Unnikrishnan had suggested a way of measuring individual Test innings in a different manner. His suggestion was that the innings should be evaluated against the average score of the other batsmen who batted in that particular innings. He also wanted the individual innings values summed across all innings for each batsmen and averaged across their career, similar to the way the Batting Averages are calculated. For instance, to compute Don Bradman's career Innings Index value, the Innings Index values for all his 80 innings would be added and divided by 80. These are excellent suggestions in view of the following plus factors.

- This is an out-and-out peer comparison, that too within the same team.
- The comparison is within the same innings: Hence the conditions would be almost identical.
- The bowling quality faced would be almost identical, barring innings-level variations.
- This effectively takes take care of the oft-repeated complaints by readers regarding batsmen playing in weak or strong teams.

In some ways this is similar to the simpler % of Team score measure. However the one major difference is that in the % TS measure the number of batsmen who batted is not taken into account. "For no loss" and "for 7 wkts" will produce the same % TS, as explained in the examples. However the Innings Index takes care of this very well and is a true peer comparison. Team score, as given below, is sans extras.

```Team-Score Batsman-Score  % TS   Inns Index

100 for 0      50          50        1.0
100 for 5      50          50        6.0
200 for 3     100          50        4.0
200 for 9     100          50       10.0
300 for 2     150          50        3.0
300 for 10    150          50       10.0

100 for 10     60          60       15.0
100 for 1      25          25       0.33
200 for 7     120          60       12.0
200 for 1      40          20        0.5
300 for 10    200          66.7     20.0

```
The formula for determining the Inns Index is quite simple and outlined below.

```Runs scored by batsman
Innings Index  = -------------------------------
Average score of other batsmen

where
Total runs made by other batsmen
Average score of other batsmen = --------------------------------
No of other batsman who batted

```

I had to do limiting of the Innings Index values for innings in which fewer than 5 wickets fell as otherwise the following silly situation emerges. Aamir Sohail's 46 out of 61/2 will get 15.33 ??? Such cases have been limited to a reasonable number below 5 since these do not really reflect batsman contributions in demanding situations.

I have shown two tables. The first is a table of the top innings based on the Innings Index value. The second is a table of batsmen ordered by the average Innings Index value over the career.

Now for the first table. Readers will note a clear correlation between this and the % Team Score. However this is a far more robust and well-thought out measure which stands any test. Let me repeat, for the sake of readers itching to put in their tuppenny-worth on the innings they think should be placed high. This is not a list of the best innings. It is a table of innings whose Innings Index values, as described in this article are high. That is all. Do not draw unnecessary inferences from either of the tables.

Now for the first table. I have listed here the top-25 innings ordered by the Inns Index.

```Table of Innings ordered by Innings Index
(Batsman score > 199 or Inns Index > 5.0)

MtId Year Batsman            Bat Team (Ext) Batsman Oth  Inns
Pos Score      Score   Avge Index

1156 1990 Gurusinha A.P       3  82/10 ( 8)   52*   2.2  23.64
0001 1877 Bannerman C         1 245/10 ( 8)  165*   7.2  22.92
1439 1999 Slater M.J          1 184/10 ( 4)  123    5.7  21.58
1481 2000 Laxman V.V.S        1 261/10 (21)  167    8.1  20.59
0732 1974 Amiss D.L           1 432/ 9 (41)  262*  12.9  20.31
0779 1976 Greenidge C.G       1 211/10 (11)  134    6.6  20.30
0542 1963 Reid J.R            4 159/10 ( 9)  100    5.0  20.00
1171 1991 Gooch G.A           1 252/10 (21)  154*   7.7  20.00
1306 1995 Moin Khan           7 212/10 (34)  117*   6.1  19.18
0401 1955 Sutcliffe B         3 125/10 (12)   74    3.9  18.97
0303 1948 Hutton L            1  52/10 ( 6)   30    1.6  18.75
1283 1995 Inzamam-ul-Haq      5 165/10 (19)   95    5.1  18.63
0164 1926 Macartney C.G       3 194/ 5 (17)  133*   7.3  18.14
0652 1969 Nurse S.M           3 417/10 (14)  258   14.5  17.79
1913 2009 Duminy J.P          6 138/10 (23)   73*   4.2  17.38
1884 2008 Sehwag V            1 329/10 (12)  201*  11.6  17.33
0665 1969 Burgess M.G         6 200/10 (12)  119*   6.9  17.25
1716 2004 Jayasuriya S.T      1 438/10 (37)  253   14.8  17.09
0130 1913 Taylor H.W          1 182/10 ( 9)  109    6.4  17.03
0079 1904 Tyldesley J.T       3 103/10 ( 8)   62    3.7  16.91
0846 1979 Yallop G.N          4 198/10 ( 5)  121    7.2  16.81
1206 1992 Kapil Dev N         7 215/10 ( 8)  129    7.8  16.54
0330 1951 Hutton L            1 272/10 (21)  156*   9.5  16.42
1820 2006 Sangakkara K.C      3 170/10 ( 9)  100*   6.1  16.39
0059 1899 Sinclair J.H        4 177/10 ( 6)  106    6.5  16.31
1747 2005 Sarwan R.R          3 194/10 (21)  107*   6.6  16.21
1444 1999 Saeed Anwar         1 316/10 (12)  188*  11.6  16.21
1749 2005 Lara B.C            4 296/10 (10)  176   11.0  16.00
0841 1979 Gomes H.A           3 151/10 ( 3)   91    5.7  15.96
0587 1965 Saeed Ahmed         3 307/ 8 (38)  172   10.8  15.96

```

A real surprise at the top. Asanka Gurusinha, a competent performer for Sri Lanka (with an average of 38.9), with his innings of 52 out of 84 all out (8 extras). This leads to an Inns Index value of 23.64. Then come three classics spread across 123 years. Charles Bannerman's 165 has a value of 22.9, Slater's 123 has a value of 21.5 and Laxman's career-defining 167 leads to an Inns Index value of 20.5. Then comes one of the greatest match-saving innings of all time by Amiss of 262, with an Inns Index value of 20.3.

There is a case for keeping a minimum team score as 100 to ensure that the index may have more validity. However I feel that in cases like Gurusinha's or Hutton's innings, the important factor is that the team was all out, in other words, 11 batsmen batted. Hence I have decided to retain these values.

Note the presence of some modern classics such as Sehwag's 201, Inzamam's 95, Jayasuriya's 253 and Sangakkara's 100. These are wonderful innings and fully deserve to be in this special list.

To view/down-load the complete Innings Index table of innings of 200 runs or more or an Inns Index value greater than 5.0, please click/right-click here.

Now for the batsmen table. To do this table I have added the Inns Index values for all the innings played by the batsman and divided the sum by the number of innings played. This leads to an average Inns Index value. An innings is what it says. When a batsman takes strike and whether he finishes at 400* or 0*, it is an innings. I am sure readers would come out with their own suggestions on excluding certain types of not outs, such as single digit ones. Let me wait for such suggestions and I am prepared to do the tweak and show the alternate table. As of now it is one straight forward calculation. As usual the batsmen who have scored over 2000 runs are shown. There is only one batsman of significance in the below-2000 group, Eddie Paynter who had scored 1540 runs at 59.23.

```Batsman              Cty Inns   Runs   Bat   Avge  Inns Index
Total   Avge  IIdx   <1.0 >5.0

Bradman D.G          Aus   80   6996  99.94  3.348   22   19
Headley G.A          Win   40   2190  60.83  3.226   17   11
Lara B.C             Win  232  11953  52.89  2.701   99   42
Taylor H.W           Saf   76   2936  40.78  2.560   38   10
Nourse A.D           Saf   62   2960  53.82  2.551   28    9
Hutton L             Eng  138   6971  56.67  2.504   55   16
Hobbs J.B            Eng  102   5410  56.95  2.463   43   14
Turner G.M           Nzl   73   2991  44.64  2.438   36    9
Flower A             Zim  112   4794  51.55  2.387   58   17
EdeC Weekes          Win   81   4455  58.62  2.360   34   12
Hazare V.S           Ind   52   2192  47.65  2.283   24    5
Sutcliffe B          Nzl   76   2727  40.10  2.255   29    8
Hanif Mohammad       Pak   97   3915  43.99  2.254   49   16
Pollock R.G          Saf   41   2256  60.97  2.246   22    5
Taylor R.L           Nzl   51   2077  41.54  2.240   30    7
Habibul Bashar       Bng   99   3026  30.88  2.216   47   11
Sangakkara K.C       Slk  156   8244  57.25  2.190   74   21
Gavaskar S.M         Ind  214  10122  51.12  2.185  106   27
Mitchell B           Saf   80   3471  48.89  2.178   31    9
Walcott C.L          Win   74   3798  56.69  2.160   42   11
Hammond W.R          Eng  140   7249  58.46  2.158   73   15
Gooch G.A            Eng  215   8900  42.58  2.156  103   25
Sutcliffe H          Eng   84   4555  60.73  2.147   34    9
Saeed Anwar          Pak   91   4052  45.53  2.135   37   10
Mohammad Yousuf      Pak  156   7530  52.29  2.130   63   17
Sehwag V             Ind  146   7613  54.38  2.117   63   17
Chanderpaul S        Win  219   9063  48.99  2.116  106   24
Cowper R.M           Aus   46   2061  46.84  2.104   23    3
Saeed Ahmed          Pak   78   2991  40.42  2.093   34    6
May P.B.H            Eng  106   4537  46.77  2.080   41   11
Tendulkar S.R        Ind  286  14513  56.91  2.075  131   29

```
Bradman leads the table with an average Inns Index value of 3.348. What does this mean. Across his career he has scored 3.3 times the average of his compatriots, taken innings by innings. And in Bradman's case, because of the strength of Australia, the proportion of late order batting performances which are compared with Bradman would be fewer. That says something. Imagine, each time he performed at the level of his fellow players, he has to make up by notching up an innings with an Inns Index value of nearly 6 !!! He has performed below his compatriots only just over 25%, while the other batsmen have done between 40 and 50%. Similarly he has batted at a level above 5 times the average of his fellow batsmen just below 25%, while the rest of the batsmen do this between 10 and 15%. These are unbelievable numbers to read, digest and marvel.

The "Black Bradman" is second with the only other 3+ value. The fact that he is quite close to Bradman speaks volumes. Lara is third with a score of 2.701. There is no doubt that he would have benefited from playing in a weaker teams. However it is still necessary to outscore them consistently. A surprise next. Herb Taylor of South Africa is next with 2.560 and the classical batsman, Nourse next with 2.551.

Look at the next six batsmen. Hutton, Hobbs, Glenn Turner, Andy Flower, Everton Weekes and Vijay Hazare. This is an eclectic mix of batsmen playing for stronger and weaker batting line-ups. So playing for a stronger batting line-up does not necessarily prevent a player from getting a reasonable high average value. Hutton, Hobbs and Weekes played in strong batting line-ups. I get the feeling that this might be true where there were 3 top batsmen in the side, not 5 was the case with the Australian team of the 2000s and recent Indian teams. Andy Flower virtually carried his team for most of his career as did Glenn turner. Hazare played in a reasonably strong batting line-up. Not a surprise that Gavaskar and Hanif Mohammad are the leading batsmen of their respective countries.

I would conclude that the top batsmen in this analysis would have an average Inns Index value of over 2.00.

Many thanks to Unnikrishnan for the suggestion. This will become part of my ratings work replacing the % of Team Score measure.

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

• shrikanthk on January 31, 2011, 4:20 GMT

Wasp: Those were some pretty perceptive comments on Hammond. Also, when you discuss English cricket with people who ought to know better, Hammond is often remembered as a flamboyant strokemaker whereas Hobbs is regarded as a somewhat stodgy opener, inhibited in his strokeplay.

I examined the strike rates of these two batsmen to see if this mental image of the two has any foundation. It has none! Hammond's career strike rate (in the games where it was recorded) is most definitely below that of Hobbs! On the other hand, pre-War Hobbs appears to have been a very attractive strokemaker with a strike rate almost comparable to that of a Bradman or a Lara. Even after the war, he appear to have played an aggressive brand of cricket in '20-21. He was perhaps a little sedate from the mid 20s onwards. But still he scored hundreds at an SR over 40 which is not bad for an opener approaching 50.

• Waspsting on January 29, 2011, 16:08 GMT

re - Steyn and Mcgrath.

Mcgrath is the Boycott of bowling. dull but effective. Bowling wide of off-stump over after over after over. Steyn is dynamic - fast, rushing the batsman, sharp bouncer, wonderful movement (old or new ball).

McGrath was the smartest bowler I've seen. He had accuracy, and worked players out.

Ambrose was just as accurate, but not as smart - he'd bowl the same to every batsman, while McGrath knew who to bowl closer to, who bowl wider to, when to throw one up a little further up. He also knew that a bouncer or yorker, even at his pace, was still effective if everything else is dead tight. Ambrose, with more bounce, more nip of the pitch and more seam movement - didn't grasp these nuances.

Wasim could have used more outswingers, yorkers and gone around the wicket more to be more effective. Waqar could have used the outswinger more and aimed to hit the stumps with it. They didn't.

Mcgrath did everything he possibly could with what he did.

• Waspsting on January 29, 2011, 16:02 GMT

@shrikanthk - I agree with you about Hammond being overrated, and would like to point out his poor record against fast bowling.

- The only team that had any at the time was West Indies (Martindale, Francis and Constantine) - and they were neither express fast, pinpoint accurate or masters of movement. Hammond's record against them was very ordinary

- Failed badly against Lindwall and Miller in '46. He was 44 at the time, but was scoring heavily in FC cricket. I think the failure came because of the fast bowling, not his age

- Well known dislike of bouncers. Nobody likes them much, but some handle it with grace anyway (e.g. Hutton). "not liking bouncers" is basically the same as weakness against fast bowling.

Hammond could butcher medium pacers, played wonderfully on bad pitches, and could grind out runs against quality non-fast men... but I think he had a SERIOUS weakness against fast bowling. Would have had a hard time in the 70's and 80's (like Amiss, maybe?)

• Alex on January 11, 2011, 10:46 GMT

Ananth - I regret the digression and never wish to have the last word. However, it is fun to debate with Abhi on anything to do with SRT, especially after a cracker of a series in SA. Would love such pitches in India!

• Alex on January 11, 2011, 7:47 GMT

Abhi - SRT does not have a single innings (in tests) that can be rated as "great" vs McGrath. Vs McGrath, SRT had trouble converting starts into big 100's ... he crossed 40 fully 8 times in 9 tests but had only 2 100's. Vs Steyn, he tends to perish early or score big ... he averages 50+ vs Steyn in 9 tests with 4 classic 100's.

To be fair, the 1999 series featured one of the best ever bowling attacks from Aussies and SRT did win MoS award in it (which should have gone to McGrath, IMO). I think SA had a terrific attack in this series --- stats of Morkel and Harris don't do justice to them. [[ I suggest you guys move on.. to the next article. Lot of water has flown under the bridge. Certainly no last word is needed !!! Ananth: ]]

• Abhi on January 10, 2011, 9:13 GMT

Alex, It is difficult for me to find a reason for your feelings.

Actually looking at Steyn recently I would rate him as absolutely the best fast bowler I've seen in the last 20 odd years (Mcgrath being just one among the pack)

Actually with both Tendulkar and Mcgrath going well Tendulkar has handled him just fine ('99 Aus series and '01 home series)- both of which included around 5 "wrong" decisions(or borderline decisions ,whichever camp you may be in)including a freak dismissal at Mumbai.

At his best Tendulkar didn't seem to have a real issue with Mcgrath. But Steyn seems to be a different ball game altogther...Even with Tendulkar playing almost at the top of his game ,Steyn more or less has his measure.

"Averages" of Tendulkar vis a vis Mcgrath are distorted due to several reasons- But a "Tendulkar at his best" vs "Mcgrath at his best" would go to Tendulkar most times....Not so with Steyn.

• Alex on January 7, 2011, 9:45 GMT

@Abhi - why do I get the feeling that the real reason why you put Steyn above McGrath is that SRT has played some truly great knocks (and averaged better) vs Steyn as opposed to vs McGrath ... much like what BCL did vs McGrath?

At any rate, this series underlined my assessment that Kallis is probably the greatest player (or the greatest failure-safe stat investment) to play the game of cricket. He lacks Sobers' natural flair and grace but has put up similar numbers across the board. I would still opt for Sobers but won't feel bad/worried at all if my sworn enemy lands him and I get Kallis instead.

• Abhi on January 4, 2011, 3:47 GMT

As rgds. ODIs that , ofcourse, is where RPOs are King. Mcgrath and co. were more effective because batsmen simply did not have the luxury of "seeing them off". They HAD to take risks and go for some shots- instead of simply defending or leaving balls a fraction outside off.

So,In ODIS economy rate should be given a greater weightage. But, i feel, that in Tests strike rate deserves greater weightage. Something Ananth is already applying to his ratings.

• Abhi on January 4, 2011, 3:42 GMT

Alex, Very valid points: 1) Steyn is perhaps having his best years.And we surely have to wait another 4/5 years before we can place him in the all time pantheon. However, I doubt that Mcgrath ever had a 200 wicket spell where he had a strike rate of 39.

2) RPO is surely important. However, the best batsmen would probably figure that if they see Mcgrath off (which given his relatively much poorer strike rate they would have a higher probability of doing)...they could then score off the other bowlers. Unless, ofcourse, you had 5 Mcgraths in a team! Then things would get tricky.

• Alex on January 3, 2011, 10:53 GMT

@shrikanthk & Abhi: In tests, low RPO tends to improve the SR of your bowling partner. So, it is an important team-centric metric. It is quite amazing that Holding had an RPO=2.8 at pace exceeding 140kmh (and even 90mph+) for most of his career.

1. Steyn is more in the Lillee & Marshall mould. Much less in common with McGrath who was in the Ambrose mould.

2. Just like Marshall, Steyn is not as effective in ODIs --- Lillee was quite good in ODIs.

Also, we are probably comparing the stats of Steyn over his best years to those of these legends over their entire careers. No doubt he is an all-time great, but best to give him 4 more years before debating his place in all-time Top 10 etc. I personally like him a lot. URL of my Imran citation on Lillee & Holding: http://pakstop.com/pmforums/showthread.php?t=15063 .

• shrikanthk on January 31, 2011, 4:20 GMT

Wasp: Those were some pretty perceptive comments on Hammond. Also, when you discuss English cricket with people who ought to know better, Hammond is often remembered as a flamboyant strokemaker whereas Hobbs is regarded as a somewhat stodgy opener, inhibited in his strokeplay.

I examined the strike rates of these two batsmen to see if this mental image of the two has any foundation. It has none! Hammond's career strike rate (in the games where it was recorded) is most definitely below that of Hobbs! On the other hand, pre-War Hobbs appears to have been a very attractive strokemaker with a strike rate almost comparable to that of a Bradman or a Lara. Even after the war, he appear to have played an aggressive brand of cricket in '20-21. He was perhaps a little sedate from the mid 20s onwards. But still he scored hundreds at an SR over 40 which is not bad for an opener approaching 50.

• Waspsting on January 29, 2011, 16:08 GMT

re - Steyn and Mcgrath.

Mcgrath is the Boycott of bowling. dull but effective. Bowling wide of off-stump over after over after over. Steyn is dynamic - fast, rushing the batsman, sharp bouncer, wonderful movement (old or new ball).

McGrath was the smartest bowler I've seen. He had accuracy, and worked players out.

Ambrose was just as accurate, but not as smart - he'd bowl the same to every batsman, while McGrath knew who to bowl closer to, who bowl wider to, when to throw one up a little further up. He also knew that a bouncer or yorker, even at his pace, was still effective if everything else is dead tight. Ambrose, with more bounce, more nip of the pitch and more seam movement - didn't grasp these nuances.

Wasim could have used more outswingers, yorkers and gone around the wicket more to be more effective. Waqar could have used the outswinger more and aimed to hit the stumps with it. They didn't.

Mcgrath did everything he possibly could with what he did.

• Waspsting on January 29, 2011, 16:02 GMT

@shrikanthk - I agree with you about Hammond being overrated, and would like to point out his poor record against fast bowling.

- The only team that had any at the time was West Indies (Martindale, Francis and Constantine) - and they were neither express fast, pinpoint accurate or masters of movement. Hammond's record against them was very ordinary

- Failed badly against Lindwall and Miller in '46. He was 44 at the time, but was scoring heavily in FC cricket. I think the failure came because of the fast bowling, not his age

- Well known dislike of bouncers. Nobody likes them much, but some handle it with grace anyway (e.g. Hutton). "not liking bouncers" is basically the same as weakness against fast bowling.

Hammond could butcher medium pacers, played wonderfully on bad pitches, and could grind out runs against quality non-fast men... but I think he had a SERIOUS weakness against fast bowling. Would have had a hard time in the 70's and 80's (like Amiss, maybe?)

• Alex on January 11, 2011, 10:46 GMT

Ananth - I regret the digression and never wish to have the last word. However, it is fun to debate with Abhi on anything to do with SRT, especially after a cracker of a series in SA. Would love such pitches in India!

• Alex on January 11, 2011, 7:47 GMT

Abhi - SRT does not have a single innings (in tests) that can be rated as "great" vs McGrath. Vs McGrath, SRT had trouble converting starts into big 100's ... he crossed 40 fully 8 times in 9 tests but had only 2 100's. Vs Steyn, he tends to perish early or score big ... he averages 50+ vs Steyn in 9 tests with 4 classic 100's.

To be fair, the 1999 series featured one of the best ever bowling attacks from Aussies and SRT did win MoS award in it (which should have gone to McGrath, IMO). I think SA had a terrific attack in this series --- stats of Morkel and Harris don't do justice to them. [[ I suggest you guys move on.. to the next article. Lot of water has flown under the bridge. Certainly no last word is needed !!! Ananth: ]]

• Abhi on January 10, 2011, 9:13 GMT

Alex, It is difficult for me to find a reason for your feelings.

Actually looking at Steyn recently I would rate him as absolutely the best fast bowler I've seen in the last 20 odd years (Mcgrath being just one among the pack)

Actually with both Tendulkar and Mcgrath going well Tendulkar has handled him just fine ('99 Aus series and '01 home series)- both of which included around 5 "wrong" decisions(or borderline decisions ,whichever camp you may be in)including a freak dismissal at Mumbai.

At his best Tendulkar didn't seem to have a real issue with Mcgrath. But Steyn seems to be a different ball game altogther...Even with Tendulkar playing almost at the top of his game ,Steyn more or less has his measure.

"Averages" of Tendulkar vis a vis Mcgrath are distorted due to several reasons- But a "Tendulkar at his best" vs "Mcgrath at his best" would go to Tendulkar most times....Not so with Steyn.

• Alex on January 7, 2011, 9:45 GMT

@Abhi - why do I get the feeling that the real reason why you put Steyn above McGrath is that SRT has played some truly great knocks (and averaged better) vs Steyn as opposed to vs McGrath ... much like what BCL did vs McGrath?

At any rate, this series underlined my assessment that Kallis is probably the greatest player (or the greatest failure-safe stat investment) to play the game of cricket. He lacks Sobers' natural flair and grace but has put up similar numbers across the board. I would still opt for Sobers but won't feel bad/worried at all if my sworn enemy lands him and I get Kallis instead.

• Abhi on January 4, 2011, 3:47 GMT

As rgds. ODIs that , ofcourse, is where RPOs are King. Mcgrath and co. were more effective because batsmen simply did not have the luxury of "seeing them off". They HAD to take risks and go for some shots- instead of simply defending or leaving balls a fraction outside off.

So,In ODIS economy rate should be given a greater weightage. But, i feel, that in Tests strike rate deserves greater weightage. Something Ananth is already applying to his ratings.

• Abhi on January 4, 2011, 3:42 GMT

Alex, Very valid points: 1) Steyn is perhaps having his best years.And we surely have to wait another 4/5 years before we can place him in the all time pantheon. However, I doubt that Mcgrath ever had a 200 wicket spell where he had a strike rate of 39.

2) RPO is surely important. However, the best batsmen would probably figure that if they see Mcgrath off (which given his relatively much poorer strike rate they would have a higher probability of doing)...they could then score off the other bowlers. Unless, ofcourse, you had 5 Mcgraths in a team! Then things would get tricky.

• Alex on January 3, 2011, 10:53 GMT

@shrikanthk & Abhi: In tests, low RPO tends to improve the SR of your bowling partner. So, it is an important team-centric metric. It is quite amazing that Holding had an RPO=2.8 at pace exceeding 140kmh (and even 90mph+) for most of his career.

1. Steyn is more in the Lillee & Marshall mould. Much less in common with McGrath who was in the Ambrose mould.

2. Just like Marshall, Steyn is not as effective in ODIs --- Lillee was quite good in ODIs.

Also, we are probably comparing the stats of Steyn over his best years to those of these legends over their entire careers. No doubt he is an all-time great, but best to give him 4 more years before debating his place in all-time Top 10 etc. I personally like him a lot. URL of my Imran citation on Lillee & Holding: http://pakstop.com/pmforums/showthread.php?t=15063 .

• Abhi on January 2, 2011, 12:00 GMT

Srikanth, As rgds. Mcgrath and Steyn I didn't mean that they were similar type of bowlers. What I meant is that if you probably give the top batsmen a choice of facing Mcgrath or Steyn- they would probably opt to face Mcgrath.

The best batsmen when playing well would more often than not be able to counter the chinese torture drip method of Mcgrath where sometimes he almost bores a batsman into commiting a mistake. However, against someone like Steyn- he doesn’t need to play on any batsman’s patience to get him out.

• Abhi on January 2, 2011, 11:58 GMT

Srikanth,

There will always be situations where a good strike rate may not be as beneficial as a good economy rate. I suppose the best generalisation that can be made is that a higher strike rate is the more "matchwinning" attribute. Even with Test match batting in poorer teams a lower strike rate and time at the crease may help in staving off defeat. Similarly, a better economy rate at the expense of a better strike rate is a more defensive attribute.

So, we can somewhat equate it with batting strike rates. In a poor team, where often winning is not even on a cards, a batsman may save a few games by blocking his way. In a good team, we would probably be looking for higher strike rates in both batting and bowling - to go for the "kill".

• Alex on January 1, 2011, 9:45 GMT

1. @Abhi: it possibly has more to do with VVS's mindset in crisis rather than his technique. As Ananth points, MSD does possess similar mental strength. When SRT hit 117* and 91 in the CBS finals and later the 103*, I thought he had reached that level ... however, maybe, that treat is still in store for us --- he displayed in often in his first two years ('89-'91) after all.

2. @shrikanthk: IMO, it is a mistake to compare Holding and Steyn since Holding has a far better ODI record. The likes of Waqar, Shoaib, Bond, Bishop (prior to his 1st surgery), Thommo, and Lee come closest to Holding as out-and-out fast bowlers. Steyn's strategy is similar to that of Marshall & Roberts, i.e., it betters that of Holding. However, let's not forget that Imran still maintains that Holding was the most talented fast bowler he ever saw. I agree with his evaluation. For the record, I might rate Steyn over Holding in tests.

• shrikanthk on January 1, 2011, 4:54 GMT

there must be something in Lax’s technique. I think it is that he rarely ever comes fully forward to pace…so there is not much momentum on the stroke/bat

We all have a warped idea of what constitutes the "ideal" technique. Forward play is an ancient, dated theory that perhaps originated with Fuller Pilch in the mid 19th cen. And he had to overcome round-arm bowlers.

Against modern overarm attack, I'd anyday prefer Ranji's batting maxim - "Play back or drive". When in doubt, always play back. Move forward only when you are confident of reaching the pitch of the ball!

A lot of great batsmen of the past century have stuck to Ranji's maxim - be it Bradman, Headley, Weekes, Tendulkar, Sobers, GRV, M.Waugh....even Laxman!

One of the reasons I don't particularly like Ponting's batting is the instinctive forward lunge against all bowlers. The forward lunge is wrong-headed since it commits you to a stroke and keeps you from playing late.

Lax epitomises Ranji's theory of back-play. [[ I would advocate a judicial mix of forward and backfoot play. No point in being rigid. Ananth: ]]

• shrikanthk on December 31, 2010, 15:29 GMT

By the way, it was interesting that you sought to compare Steyn with Holding. They are, in fact, very similar bowlers. Both bowl at great pace and attack middle-and-off relentlessly.

They are very different from the typical "modern" fast bowler like Lillee of McGrath who bowls a little outside the off-stump tempting batsmen in the corridor of uncertainty.

The reason why Holding has a much better economy rate is because batsmen in his day weren't as good at clipping straight balls on the on side! I'd like to think that Holding's stats would've been very similar to Steyn's if he were playing today.

Steyn vs Lillee/McGrath is perhaps a better comparison, to make your point. Because I think Lillee and McGrath would've been very economical in most eras since their line of attack is less straight and their methods decidedly more conservative. [[ No, Shri, I did not compare Steyn and Holding. rather I picked out two great bowlers with contrasting numbers and identical average. The comparison was number-centric rather than by type of bowler. Ananth: ]]

• shrikanthk on December 31, 2010, 14:51 GMT

Steyn is Mcgrath on steroids- Relentless contol PLUS considerably faster. Amazing combination- Mcgrath Redux if you will.

Though I haven't seen as much of Steyn as I have of McGrath, I feel they are very different bowlers.

Steyn is a more old-fashioned bowler in the great classic fast bowling tradition of Richardson, Larwood and Lindwall. He bowls straight with great pace on off stump. But unlike those great names, he is more expensive because modern batsmen are, I think, more adept at working straight balls on the on side (thanks to odi cricket).

McGrath is quite different. He hits the deck harder than Steyn. Bowls a slightly shorter length. And bowls more for catches than hitting the stumps. Technically, I think McGrath is closer to the likes of modern greats like Lillee and Ambrose as opposed to classic giants like Lindwall/Larwood.

• shrikanthk on December 31, 2010, 14:36 GMT

Abhi: I don't quite agree. Suppose the "Steyn" attack and the "Holding" attack are bowling in the 3rd innings of the Test Match. The batting side in the fourth innings will have to negotiate more number of overs if the Steyn attack bags 10 wickets in the 3rd innings.

This need not necessarily be a good thing. Suppose you're chasing 200 in the 4th innings - the Steyn attack is preferable as it gives you more time to chase the total of 200.

But suppose you're chasing 600 in the 4th innings, I'd rather choose the "Holding" attack as it leaves you with fewer overs to bat out for a draw!

The takeaway: The debate between strike rate and economy rate is essentially philosophical and not objective. The only objective means of evaluating a bowler is to look at his bowling average.

I made this point even in the earlier post. Nevertheless, it looks like I'm not persuasive enough in my arguments :(

• Abhi on December 31, 2010, 2:31 GMT

Ananth, Ha. Thank you for humoring my 5th standard math. I guess the total runs would have more easily have been got by multiplying the average into the number of wickets. But anyway....happy new year to you and all commentators. [[ That is true. However the fact of the earlier cobclusion to the innings can only be understood by your method. Ananth: ]]

• Abhi on December 30, 2010, 6:20 GMT

Actually, One of the very few situations where the “Holding” attack may be preferable would be when protecting a score of LESS than 237 runs with more than 67 overs of play remaining. In practically every other situation in TEST cricket- the “Steyn” attack would be preferable. [[ Abhi, that is some nifty arithmetic. However it is not correct to conclude that, IRRESPECTIVE OF SITUATION, capturing wickets at the cost of concedinbg runs is the better option. I can always work out some situation or other where it is necessary to keep it tight. Howver I myself feel that Strike rate should have a higher priority over Bowling accuracy and have always given higher weight for that, but for Ratings work. Ananth: ]]

• Abhi on December 30, 2010, 6:09 GMT

Ananth, Using your eg. Of “Steyn and Holding have approximately the same average (23.65 & 23.69). Steyn's lower level components are 40.2 (s/r) and 3.53 (rpo). Holding's components are 50.9(s/r) and 2.79 (rpo).”

Here’s some stat play. Assuming the entire bowling attack “totalled” to the individual figures of Steyn and Holding as above.

Then to get 10 wickets : The “Steyn” attack would take 10 X 40.2 balls = 402 balls. = 67 overs . In 67 overs they would give away 67 x 3.53 = 237 runs.

The “Holding” attack would take 10 x 50.9 =509 balls = 84.8 overs. In 84.8 overs they would give away 84.8 x 2.79 = 237 runs.

Clearly the “Steyn” attack is far superior?

• Rupert on December 29, 2010, 20:41 GMT

I absolutely love your analyses! Do keep up the great work!

I was wondering, would it be at all helpful to weight the batting positions !?! I mean, if you consider the openers' wickets and those of the tailenders to be equally valuable, you will end up depressing the Index values of batsmen in strong teams. Maybe weight the openers' wickets at 2, Nos. 3 and 4 at 1.5, Nos. 5 and 6 at 1, Nos. 7 and 8 at 0.5 and the tailenders at 0.

Another way would be to weight each wicket with the batting average for that position.

Have a great new year, and thanks once again for all the great work !

• unni on December 29, 2010, 14:28 GMT

I couldn't resist typing this comment eventhough, it is a batting article, eventhough it is beating a dead horse and eventhough I'm favouring strike rate along with Abhi and lovegoel....

It is about a beautiful symmetry between bowling average and the batting average (especially with the not-out considered as usual).

Consider the runs scored as a long ruler like line bisecting the two opposing landscapes; batting and bowling. Wickets are like the markers on this ruler. When a batsman gets out a marker is added on this line on the batting side. When a bowler gets a wicket a marker is added on the bowling side. Then both the 'average' are simply the reflections of same idea on both the landscapes.

• Abhi on December 29, 2010, 8:23 GMT

Ananth, I meant that for the Bowling quality index we may be better served using the modified bowling averages with different weightages for economy and strike rates (which apparently you do) . Not for general use in calculating bowling averages. [[ I can assure you that in any of my ratings work, Strike rate would get a higher weight than bowling accuracy. As one who has been bored to tears seeing Nadkarni bowl maliden after maioden., I know how important strike rate is. Ananth: ]]

As rgds. Laxman, it was a direct compliment. The inescapable fact is that he has done this time and again. It is not a one off thing- So, after a while it cannot be luck. Like the other crisis man ,Chanderpaul, Lax too has this type of - "play the ball late/soft hands/don't commit completely to the shot until absolutely sure" type of approach.

So, what may seem like an endless succession of luck/near misses may actually be par for the course.

I don't think temperament is the critical factor here. All top sportsman are fiercely determined competitors. The "trick" has got to be something intrinsic to Laxman's technique.

• Abhi on December 29, 2010, 5:36 GMT

Ananth, As always one comment leads to another- and I go on and on. Made me think about an analysis about "wicket-spells" i.e something along the lines of 1)wickets per spell 2)% of successfull spells etc. [[ I had done a Bowler consistency analysis earlier. Probably time to re-visit the same. Ananth: ]]

• Abhi on December 29, 2010, 5:31 GMT

Alex, Love goel Of course, there will always be a debate as to who is the best fast bowler...But I have rarely felt the same buzz as when Steyn starts a spell. You get the feeling he can get a wicket of two every spell- every time. You can count on your fingers the number of times the batsmen have hit him with total confidence off the middle of the bat for a boundary. The rest of the time it seems to be all edges,plays and misses, squirts through slips or behind square etc. Hardly any "V" shots.

To me Steyn is the complete package as a fast bowler- fantastic accuracy, swing/seam both ways in total control at incredible pace, intimidation etc.etc [[ And with UDRS in place a couple of more wickets !!! Ananth: ]]

• Abhi on December 29, 2010, 4:29 GMT

Uh Oh, I got it the other way round. The "lower" the average the "better" of course. So,perhaps for Tests: Bowling Average = Bowling Accuracy(0.6) x Strike Rate(0.4)…… [[ You would be suggesting a method which would shake the foundations of cricket numbers. Bowling Average = Runs/Wickets. No one would ever accept this simple and wonderful measure to be changed. However let me say that I do it differently. Whenever I do Bowler analysis I separate the two componenets and give more weight for Strike rate. Ananth: ]]

• Abhi on December 29, 2010, 4:24 GMT

Ananth, I cannot argue with the math ofcourse -You are absolutely right when you state that both economy and strike rate are the main components when concerning bowling quality. However, I guess the problem then lies in “weightages”. Would I be correct in assuming that when using averages we are effectively giving a 50% weightage to both strike rate and economy? .I think this is the major bone of contention- i.e how much weigtage to give to each. Perhaps for Tests it should be : i.e in the formula - Bowling Average = Bowling Accuracy(0.4) x Strike Rate(0.6)……and vice versa for ODIS….Just venturing a thought. [[ have already been covered in your next comment. Ananth: ]]

Alex, I am running out of words for Laxman now. (First this phenomenon was restricted to Tendulkar)…Watching the innings I was for half the time admiring his tenacity and for the other half his seemingly endless luck.Some 20 plays and misses, 10 balls just missing the edge, 5 balls inside edging a millimeter off the stumps, edges just falling short of the fielders. But then Lax has some Ten 2nd inn. Innings like that ! So, the matter has gone beyond Luck. Lara and Tendulkar have the odd 3rd or 4th great innings , so you can definitely say luck played a major part in the odd innings here and there. But , if you keep doint it – there must be something in Lax’s technique. I think it is that he rarely ever comes fully forward to pace…so there is not much momentum on the stroke/bat…either through body momentum ala Ponting or backlift ala lara….I think the batting greats like Gavaskar, Boycott etc should dissect Lax’s technique to figure this out. Like Harris said yesterday – “The Greats create their own luck”…So, whats’ Lax doing that the others are not? [[ Bit of a left-handed compliment. Probably deserving since the innings, worth its weight in Platinum, was streaky. Not the pure innings he has played in other situations. The great thing is the way he can play the sublime cover drive or turn to midwicket a ball after he has played and missed. A player who has not always been treated well. I get the feeling Dhoni has to take over his role when VVS retires. He also has the unflappable temparament and the ability to put away bad balls. India has had the luck as far as Zaheer's innings was concerned and the Steyn plumb Lbw dismissal. But they deserved their luck. As would South Africa if they get a break or two today. Ananth: ]]

• love goel on December 28, 2010, 18:09 GMT

You cannot say that Steyn is far superior to Holding based only on Strike Rate since he conceded nearly 0.75 runs per over more. ....

I beg to differ.Other parameters being equal, Steyn is definitely the better one. The reason being average and strike rate only gives the stats of that bowler. But the longer the batsmen stays, he is also going to play more balls off the other bowlers and will score some runs off them. This factor is no way included in these 2 figures. Lower strike rate helps in this regard

P.S. I will take fully fit Shane Bond any day over these 2. Both for the pace and the accuracy [[ I am afraid this argument is going past the pointless stage. I can point out many instances when keeping the runs down would have more meaning, even in a Test match. I have only pointed out that Steyn conceds 0.75 runs more per over. That is a fact and cannot be wished away. That means that no one can can say with utmost certainty that one is much more important than the other. It is possible that Strike rate should be given slightly higher weight because these are Test matches. Let us close this line of discussion. Ananth: ]]

• Abhi on December 28, 2010, 13:48 GMT

Ananth, I guess we saw the same table- but in the stats bases it shows Barnes had 189 wickets at a strike rate of 41.6

Anway , please clear this up for me. I still don’t quite get it. You state :"I have also held that the Bowling Average encompasses both strike rate and accuracy."

But for eg: 1)Bowler 1: 5 overs , 20 runs , 1 wickets = avg. 20 , SR 30 2)Bowler 2: 10 overs, 20 runs , 1 wicket = avg. 20, SR 60 Bowler 1 is clearly the far more effective bowler.

There are several bowlers with a better (lower) average than Steyn but poorer strike rates. [[ I do not know how many times I have done this. But here we go. Bowling Average = (Runs/Balls) x (Balls/Wkts) = Bowling Accuracy x Strike Rate. That is what I meant when I meant that Bowling Average encompasses both Strike rate and Accuracy. In the example you have given, the second bowler might have a poorer strike rate but much better bowling accuracy. Also never make a statement looking at a single match and 1 wicket. Look at it over a career. Steyn and Holding have approximately the same average (23.65 & 23.69). Steyn's lower level components are 40.2 (s/r) and 3.53 (rpo). Holding's components are 50.9(s/r) and 2.79 (rpo). You cannot say that Steyn is far superior to Holdinbg based only on Strike Rate since he conceded nearly 0.75 runs per over more. That is where bowling Average comes in. Ananth: ]]

• Alex on December 28, 2010, 11:36 GMT

Ananth and Abhi:

1. I think since Steyn is not that effective in ODI's, it is a mistake to rate him as McGrath's equal/superior. In tests+ODI's, SA's own Donald put up better numbers. Since the final few years of Ambrose & Donald were not as spectacular as McGrath's, these two get forgotten more easily.

2. No denying that Steyn is all class and an all-time great but, IMO, the five great WI bowlers, Imran, and Donald are arguably better him even in the more special out-and-out fast bowler category. It will be a stretch to put him among all-time Top 3 in it but he does bring back memories of Imran in full flight.

3. VVS too now has the magical 96 next to his name - hope this one is on winning side. Yet another gem from the maestro of the 3rd innings. Perhaps Ananth could enlighten and delight us all with his innings-wise stats article soon?

• Abhi on December 28, 2010, 5:32 GMT

Alex, True. But perhaps India may have been better served over the years with Dravid opening (also would have led to less mix and match and fiddling with the opening pair)...Tendulkar at No.3....Laxman at No.4.

Srikanth Re. The bowlers- In my opinion Dale Steyn is the best right arm fast bowler I've ever seen (for an extended period). The last time you saw the likes of Dravid,Tendulkar,Lax beaten by sheer pace was vs.Shoaib and Brett Lee. But those 2 weren't so relentlessly accurate. Another 4/5 years like this and Steyn will easily surpass Mcgrath as the “Greatest” Right arm fast bowler of the generation. In my opinion he is already “better” than Mcgrath…but to be considered “greater” he will have to keep it up for a while more. Steyn is Mcgrath on steroids- Relentless contol PLUS considerably faster. Amazing combination- Mcgrath Redux if you will. So,If he can keep this up for another 5 yrs you will see Steyn right up there in any World XI list.

(Also, Steyn's Strike rate ,I just heard is the best among bowlers who have taken 200 wickets- one of the prime reasons why he is so deadly- and another reason why we MUST incorporate strike rates in any bowling quality index to get a “hang” of genuine quality) [[ I have always mentioned about the strike rates of Younis and now Steyn. I have also held that the Bowling Average encompasses both strike rate and accuracy. It amuses me to see Television tables on strike rates formed in such a way that the old bowlers are excluded. For instsncae the table which I saw was for "Bowlers exceeding 200 wickets" and Barnes was excluded. Ananth: ]]

• D.V.C. on December 27, 2010, 9:18 GMT

All these discussions where people leave players out of All-Time XIs simply because they haven't seen them always amuse me. All-Time should mean All-Time. How come nobody ever mentions a player from before the birth of Test cricket? There are 100s of years people always seem to ignore.

For one example, why not consider Fuller Pilch? The man who invented the forward defensive shot? It used to be called Pilch's Poke!

I bring this up because this type of analysis, where we are comparing peers, is the best possible for seeing how the relative contributions of old-time players stack up against the modern greats. Ananth, any chance of doing this with First Class figures?

• Alex on December 26, 2010, 10:43 GMT

@Abhi - traditionally, #3 is for the batsmen who bats the longest or scores most runs (unless he be an opener) and can tackle the new ball expertly but there are plenty of exceptions. For example, Sobers mostly played at #4 through #6 (albeit his WR 365* did come at #3). GRV & Miandad made their name at #4 mostly. Over '80-'85, Lloyd was the most reliable WI batsman but batted at #5.

However, there is no doubting that if someone is capable of a real big score, his best chance to get it is probably at #3 --- just look at the abnormally low ratio of 100's to 50's for VVS.

• Alex on December 26, 2010, 8:59 GMT

@Martin: Hammond is in my all-time Top 12 after receiving a benefit of doubt (for not having seen him bat) over G Chappell & Ponting who regrettably don't make this baker's dozen. He might possibly have been a more stylish version of Kallis and that should count for something since sports are visual entertainments and not just ticker tape results (except to the bookies).

BTW, if we restrict SRT's career to Apr 1991 onwards (i.e., age 18+), SRT averages 58 in tests.

• Martin Hooks on December 26, 2010, 5:37 GMT

@ Alex, I said ( actually Wisden said)amongst top 4 all time batsman till that time. Obviously, no one has a window into the posterity. But all said and done, looking at his numbers and listening to experts, who have watched and played cricket in that era , it is quite clear that he may very well be under-rated and not over-rated. Afterall, even Ponting, Lara and Tendulkar do not have that test average. Those are the facts too! [[ Martin If we agree that the batsmen cannot decide on the quality of bowlers faced and that he plays at all places and that he may get very easy to very tough runs, it is necessary to accept that the average is a very good indicator of the batsmen quality. I only pointed out the quality of bowling Hammond faced in some matches. That way i can do that for virtually every batsman. It is fair to say that Hammond should find his place in anybody's list of the ten greatest batsmen who ever played. Ananth: ]]

• Abhi on December 24, 2010, 15:51 GMT

Alex, If we assume that it was downhill for Richards from ’83 onwards (as you mention)- then here’s the stats- From debut to beginning ‘83: At no 2- avg. 70 ; No.3 -65 ; No.4- 32 ; No.5 -60 From ’83 to retirement: At no 3- avg. 4.4; No.4 -44 ; No.5- 45.7 (Almost all the same at various positions)

Moral of the story ? - Get the best batsmen in early , preferably not later than No.3

• Alex on December 24, 2010, 12:13 GMT

1. @Ananth - re Hema's comment, perhaps you can choose not the Top 7 batsmen but the Top 7 run scorers in that particular innings. [[ Alex, I think you have hit the nail on the head. This will make sure that the late order contributions, if these are higher than the earlier batsmen, will have relevance. One other thing is that the denominator will be strengthened and the resulting Index values will fall within lower numbers. The range of 0.1 to 23 is too wide. I think I am going to take a long and hard look and come out with a very robust Index computation. I would first apply all these in the ODI computations. Ananth: ]]

2. @Abhi - Viv's average at #3 is much greater than his average at #4 because he played at #4 when he was sadly past his peak prematurely in tests (i.e., after '83) ... he himself is to be blamed for this decline. He was quite good at #5.

3. @Martin Hooks - In 1965, in Hammond's obituary, Wisden rated him, Grace, Hobbs & Don as the 4 greatest batsmen ever. Sentiment is understandable but it was not a slam-dunk that he be rated over Ranji, Trumper, Headley, & Sobers in 1965. [[ We also should not forget that Richards, Tendulkar, Lara, Ponting et al had not been born or were too young to play a Test in 1965. Ananth: ]]

Ananth- an article on SRT would be a great read now. He probably has 4 more yrs left in him. Also, Kudos to Dravid on 12K!

• shrikanthk on December 24, 2010, 3:29 GMT

unni: I agree. My point is that a batsman may end up with a low D.index simply because he happens to be a very steady, reliable performer with not too many really big hundreds under his belt!

No wonder Gooch has a high D.index of 2.2 whereas M.Waugh has a D.index of 1.5 though their averages are roughly the same! The differences in their D.index is mainly because Gooch's scores exhibit more variation (higher std.deviation) unlike M.Waugh's scores (whose highest test score was 153!!). I bet you'll find this pattern holding true regardless of the quality of batting line-ups (strong or weak). Batsmen with high sigmas have a higher D.index.

• unni on December 23, 2010, 15:49 GMT

@shrikanth : Regarding your hypothesis on avg and std.deviation. It is a function of average and std. deviation. But not limited to that. How about other guys' scores? That is what affects the dom. index very directly.

• unni on December 23, 2010, 15:41 GMT

However, I disagree to the comment because, how should we handle the case, say, a number 8 scored a century to rescue the team. Consider this case, team score is 300/9. No 8 scored 150. Opener scored 50. Now, if we go with this logic, the opener would still be considered as dominated eventhough the no 8 outscored him at least in this match.

So, in my opinion, we shouldn't complicate it further. [[ No, not really. Say, the average of the top-7 batsmen is 15. The opener would get an InnsIndex value of 3.33 and the no.8 batsman would get 10.0. It is fact that the opener outscored his 1-7 compatriots by a factor of 3.33. I will only use the top-7 average in the denominator. i never said that I would exclude a no.8 batsman. Ananth: ]]

• unni on December 23, 2010, 15:37 GMT

@Hema Singh : Thanks for the comment. @Ananath : As usual I'm against cut-offs which introduces sort of abruptness. If, I understand Hema's comments, the trouble is that a lot of lower order wickets impact's a good batsman's dom. index. I both agree and disagree to that. First agreement part. If there are two innings like 200/6 and 200/9 and one of the opener scored a 100 in this situation, he should get the same dom. index since in the second case also, whoever got out extra were mainly bowlers. But please note that if we cut the wickets off to 7, in the second case, the actual dom. index would increase since the other batsmen avg is appearing in the numerator. Hence what we could do is to weigh the batsman count with the average of the batsman. i.e, 1 opening batsman got out would be counted not as 1; but opener's avg/total team avg*11. This will generally more than 1 for proper batsmen but very less than for bowlers. This would adjust the dom.index as you expect.... to continue...

• Abhi on December 23, 2010, 13:53 GMT

Srikanth Sorry to go on!...But to me Strike rate is a critical factor in bowling quality…A bowler with a mean average may “restrict” a batsman but not be able to get him out. In the interim the batsman concerned may score against other bowlers. The strike rate is critical…if a bowler can get a batsman out quick time, it has to count for something.

So, I am a big fan of strike rates. It is the spinners for whom averages are more important because they bowl so many more overs. For a fast bowler- strike rate should take prominence.

• Abhi on December 23, 2010, 13:47 GMT

@Martin Hooks: I totally and completely agree with Martin Hooks. You play what you get- simple. Why didn't all the other batsmen in the team perform as well as Hammond?

For eg. the latest "doubt" re. Federer's place in the Tennis pantheon has been put forward by a very respected Tennis writer- that the level of competition Federer has faced is far below what Sampras has faced ! Hello ?!......What does Federer have to do now? Create imaginary opposition?...How can anyone make such a claim. All you can do is play against your contemporaries- and so as with cricket, “peer” ratios are everything.

• Abhii on December 23, 2010, 13:46 GMT

@shrikanthk: The bowling quality index thing has been a highly debated issue for a while- with no unanimous resolution in sight. One (among other) problems with even CTD figures is for eg.what do you do about the slow starters/late bloomers? For eg. Anderson and Swann are far superior bowlers to what they were initially. So, the initial hammering of their stats distorts CTD - Their current form has no bearing on their CTDs.

Then you have situations wherein selectors use a "horses for courses" policy- with certain bowlers picked purely on form, suitability to surface ,conditions etc. (only the very top bowlers are always assured of a place).

There are numerous such variables...and the current method is just too simplistic.

• shrikanthk on December 23, 2010, 9:48 GMT

Ananth: A query on the methodology you used for the analysis on Hammond's centuries. Did you just consider the mean bowling average of all the bowlers in the given innings? [[ I have already explained this. I do double weighting. First at the innings level. Sum of (Balls x Career average) for all bowlers A = ---------------------------------------------------------- Total balls bowled.

Then for the batsman Sum of (Runs scored x A) for all innings = ------------------------------------------------ Career Runs. Arjun has pointed out a fallacy in this double weighting and suggested an alternate method. May very well do an article on this.

Ananth: ]] Just a thought: We might want to take the mean bowling average weighted by the number of overs bowled. For instance, if Bowler 1 has an average of 40 and bowls 1 over. And Bowler 2 with an average of 50 bowls 9 overs. The mean bowling average = 0.9*50+0.1*40.

I got this idea while browsing through the scorecards of the Indian tour of Australia in 1947-48. It is a part of Indian folklore that Vijay Hazare scored 2 centuries in the Adeleide Test against the might of Lindwall and Miller. However, the scorecards suggest that Lindwall and especially Miller did not bowl a great deal in that test match. It is quite likely that Hazare personally may have had to negotiate not more than 10 overs by the duo in each innings!

• shrikanthk on December 23, 2010, 9:00 GMT

Hmm.. Martin: Chill. What's the fun of a Comments section without the odd passionate debate. Let's not get worked up over every other word that seems mildly provocative. I agree "overrated" was a somewhat imprecise word.

I pretty much agree with whatever you said. All I am saying is that Hammond does enjoy greater mindshare among cricket fans than several players whose records are just as impressive.

• shrikanthk on December 23, 2010, 7:25 GMT

shane: Sometimes the incentive to play safe for a draw might be stronger in a 5 day test than in a timeless game where a result is inevitable.

Yes. Hammond could score quickly on occasion. But so could Boycott! It is strange that he is remembered today by some as a compulsive strokemaker, which he wasn't.

Re the analysis: We ought to consider the distribution of scores of a batsman over his career. Suppose a person hits a big double hundred with a D.index of 10. This impacts his overall career D.index more strongly than a single digit failure (with a D.index of 0.05).

My hypothesis: The D.index is broadly a function of two things - The batting average - The Std.Dev of scores.

The higher the sigma, the higher the D.index. Take M.Waugh for instance. A very consistent batsman with not too many big 100s to prop up the average!. No wonder his index is low. It shouldn't be implied that Waugh scored runs in a pressure vaccumm. He remains one of Aus' greatest match winners of the 90s.

• Martin Hooks on December 23, 2010, 5:27 GMT

Hammand is overrated? Says who? I can't believe someone just sprouted that, just like that so after preaching respect and things like that for one another and legends of the game. He is one of the true legends of the game and suddenly few arm chair experts have decided that he is overrated? Compared to whom may I dare ask? Wisden Cricketers' Almanack described him in his obituary as one of the four best batsmen in the history of cricket. He was considered to be the best English batsman of the 1930s by commentators and those with whom he played; he averaged 58 in test cricket and after 500+ FC games averaged 56. Hammond was an effective fast-medium pace bowler and contemporaries believed that if he had been less reluctant to bowl, he could have achieved even more with the ball than he did. I generally don't express outrage except when it is truly outrageous to the core. [[ That is a personal view of one reader. At no stage have I expressed that Hammond is over-rated. What I have presented is a collection of his 100s against average-to-very weak attacks. Those are facts. Ananth: ]]

• Abhi on December 23, 2010, 5:18 GMT

I feel that for the best batsmen the best batting position is actually at No.3..From Bradman down to Richards,Lara etc the best batsmen have all done much better at No.3 than later down the order.

Probable reasons: 1)The best batsmen have the technique to play the new ball-so that is not an issue. 2)More importantly they then get all the time to construct an innings at their own pace. 3)Most big hundreds and triples have come in positions 1-3. 4)Consider Richards- avg 61 at No.3 and 48 at No.4. This is a whopping difference. The only reason I can think of is that at No.3 he could come in early and so "have his fill"...as opposed to coming in later on with runs on the board and then just going for quick runs.

So, essentially because of his No.4 position Richards comes in low on this list.The story would be quite different if we consider only his runs at No.3. Classic case of a batsman having less than optimal figures due to playing lower down the order in a strong team.

• shane on December 23, 2010, 3:24 GMT

Off the topic but Shrikanthk, I have to make two comments re your claim that Hammond is overrated. One: the two greatest bowlers of that era played for Aus. - Grimmett and O'Reilly. Secondly: a possible reason behind Hammonds slow strike rate in 28/29 was that tests in Australia were timeless so you there was no pressure to score quickly. In fact it encouraged slow methodical play as a batsman who scored a quickfire 50 before throwing his wicket away did not really benefit his team. In 1938 Hammond did score 210 in one day after England were about 3 for 40. So he could score quickly if required [[ Shane, I would not say that hammond is overrated. However he faced many very light attacks as seen in my response to Shrikant. Many of his centuries, including quite a few high ones, were against average attacks, including Australia., Grimmett's career ran from 1924 to 1936. O'Reilly's 1936 to 1946. So they did not have many overlapping years. In fact they played together in only 14 Tests, 7 against England.. Ananth: ]]

• love goel on December 22, 2010, 19:11 GMT

Ananth, for not out innings, I will suggest we take the average the same as that for out-innings. Is there any reason to think that if a not out batsmen was allowed to complete all his innnings, his average will be higher than that of completed innings? After all out-innings are same as not out innings , except that batsmen was allowed to complete them

If a batsmen has say 100 completed innings at an average of 50, and 20 not out innings with average of 35, I say we make the career average =50

It is possible that a batsmen has not-out average greater than out-average. But I think such cases will be far and few with very less number of not outs. [[ This applies to Runs. For this there are very few doubters of the current method. For Inns Index, I can also look at not including Not out innings where the Index is <1. Alternately make these as 1. These are the ones in which the batsmen has not completed the innings and might be penalized. Anyhow these are for afollow-up, incorporating the many good suggestions. Ananth: ]]

• shrikanthk on December 22, 2010, 13:17 GMT

Ananth: Thanks for the follow-up on Hammond, who I think remains heavily overrated, in sharp contrast to Bradman who deserves every ounce of adulation he receives.

Hammond scored very heavily and very slowly in 1928 against an ageing Australian attack. People often talk about his 905 runs in '28 in the same breath as Bradman's 974 in 1930. Seriously, there is no equivalence! Firstly, DGB scored those runs at a much much quicker pace. Secondly, he mastered two all-time greats in that series - Maurice Tate and Harold Larwood. Whereas, Hammond laboured at a strike-rate of 35 against ageing bowlers, with Grimmett being the only serious challenge.

He failed miserably in 1930 and 1934 series. He did make some runs in 1938 against a very weak Australian side. The Bodyline series of 32-33 is probably the only time he scored runs against a reasonably good Aussie attack.

And yet, Hammond is somehow a "legend" while nobody seems to remember the likes of McCabe and Ponsford from the same era.

• piyush on December 22, 2010, 8:04 GMT

Declaring at 550/5 Vs 600/10 is worse for the opener who lets say scored 220*. Many 2nd innings 100s(usually for team batting in the 3rd innings of the match) will not be rated highly where the goal was to set a huge 4th innings target (and perhaps quickly) as the team would often declare, than be all out. I suspect many Australian players of recent years will be at a disadvantage due to this policy.

My suggestion is to limit the number of wickets fallen to 7, than 10 when calculating the average 'rest-of-team' [[ What we should not do is to tweak the formula to suit a single instance. It really does not matter what the 220 batsmen gets. What matters is what a 100 out of 200 for 5 or 200 for 10 gets. Pl look at my response to hema. Ananth: ]]

• Arjun on December 22, 2010, 7:46 GMT

Some readers are talking about 'Dominance index'; my definition for the same is bit different. Dominance index is Highest score in the innings divided by 2nd highest score in the same team innings.(actually indicates lack of support in the same innings) eg. 2nd highest score in Kapildev's 129 was 17. Dom. index is 129/17 = 7.58 See if above ratio can incoporated in the table. [[ This is what I use in my Ratings calculation to determine the Support received. Ananth: ]]

• Abhishek Mukherjee on December 22, 2010, 5:23 GMT

Good analysis. The only aspect that I can think of is, batsmen playing in a weaker line-up (read Andy Flower) would get a higher score here than, say, Rahul Dravid (who has batted alongside some all-time greats), which does not necessarily make the former a greater batsman.

Just a thought, Trumper had scored 74* out of 122 (1 extra). So the team score becomes 47/10=4.7. Trumper's innings index turns out to be 74/4.7=15.74. A true classic, just to have missed your cut-off of 15.96, but possibly better in quality than some on the list. [[ The real cut-off is only in number of entries not an Inns Index number. Ananth: ]]

• Hema Singh on December 22, 2010, 4:50 GMT

Good effort, but even if we take peer value at its face value (frankly I have a problem with this kind of comparison) still 'Innings Index' has a problem in a way it calculates 'the average score of the other batsman'. Average score of all players are not reflective of peer value even within the team since it also takes into accounts batting average of tail-enders and pure bowlers. For example, in great WI team we should do peer evaluation of Richards, Llyod, Greenidge, Haynes and the likes; not mixing them with batting average of Garner, Holding and Andy Roberts. Think about a cutoff in terms of career average or something like that to fix the problem. Otherwise a batsman batting in a a team that has long tail will suffer. [[ Many thanks. Instead of making negative comments and leaving it there it is nice of you to look at the problem deeply and come out with an excellent suggestion. This is a first attempt. I think this can be improved considerably. I agree that the impact of the scores of the late order has to be minimized. Especially in teams with average tail who are likely to score poorly. I am not comfortable with an average-cutoff. Instead why cannot I take only the first 7 batsmen's scores. The advantage is that top order batsmen would not be excluded because of low averages. Also in weaker teams, the top-7 is top-7, whatever be their averages. This will strengthen the Peer comparison aspect immensely.Inns Index factors like 20 will disappear. But will have real relevance. Let me hear from Unni, Arjun, Shri and anyone else. Once again thanks. Ananth: ]]

• shrikanthk on December 22, 2010, 3:47 GMT

Wasp, Ananth: I think we should just consider the averages of the contemporaries during the batsman's career. i.e If you wish to benchmark Sobers against Barrington, Kanhai and Harvey, we should consider the averages of these three during Sobers' career which lasted from 1953 to 1973.

This should yield interesting results. Mark Waugh may have averaged just 42 in the nineties. However, if you consider Ponting's average during M.Waugh's career, it won't be much higher than 42 I bet!

By the way, we should also have a list wherein batting averages are appropriately weighted using the opposition bowling "quality".

Take for instance Bradman and Hammond - two giants from the 30s. Bradman's average of 99.94 remains a cricketing monument. However, I believe Hammond's 58.xx is a little inflated. This is because throughout the 30s, England had a better bowling side than Australia (especially in the pace department). [[ Given below is part of a reply to Arjun. This is based on my current method. Still the figures are telling. It is true that the English bowling of 20s was much better than that of Australia. I have given below some of the Hammond 100s scored off average bowling. You must note two things, One is that I have strict average value adjustments for the career start situations until 50 wickets are captured. The other is that during Hammond's career the Australian attacks were quite average quite often. Aus-200-46.47 Aus-119-45.41 Aus-119-45.41 Saf-121-59.80 Aus-113-44,42 Nzl-100-61.34 Aus-100-57.75 Nzl-227-81.74 Nzl-336-73.75 Ind-167-58.96 Ind-217-41.85 Aus-231-42.85 Nzl-140-45.88 Saf-181-46.04 Saf-120-47.61 Saf-140-48.29 The two New Zealand big innings single-handedly added about 3 to the final number. It was indeed amongst the three worst bowling attacks ever. In these two matches New Zealand fielded 5 bowlers whose combined CAREER wickets were 49 at an average of 80+. The Bangladeshi/Zimbabwe attacks were deadly compared to this. The 81.74 is the WORST ever Test bowling attack put on field. Ananth: ]]

• Waspsting on December 21, 2010, 21:57 GMT

we'll need some way of defining what "contemporaries" are - maybe the two players have to have had 50% of their careers overlap?) Imran wasn't Cowdrey's contemporary and Tendulkar wasn't Imran's, thought they've played in the same matches.

Thoughts? [[ One of the best analytical bits I have done was the Peer Analysis last year. If you have not done so already, please go back and have a look at it. There the player is compared to his Peers, during his career, to a day, and ratios drawn. The entity of comparison was an amorphous mass of peer players. What you are suggesting is something little different. You want to take peer players, but as an identified and named entity, and do a comparison. Nice idea. I have to see how to identify this group and how to handle the overlaps. Probably identify a set of periods and then allot the players to each such set based on their spending over 50% of their career in a specific group, accepting overlaps into other periods and then do their complete career comparisons. Will do (and arm myself with an armour to take the brickbats, which, now, are becoming personal ???). Ananth: ]]

• Waspsting on December 21, 2010, 21:53 GMT

Hi Ananth,

I was thinking about a stats analysis along these lines - comparing players averages, scores AGAINST their own contemporaries - as a judge.

Don't know how the specifics would work out (your the expert there), but going on the assumption that each generations greats would be greats in other generations, it sets up a marginof comparison across generations (given how its cliche to start any such comparisons with "its very hard to compare across generations...")

For example, Sachin Tendulkar is the best bat i've seen, I notice he has not outstripped his contemporaries. Ponting, Kallis, Sangakarra have records comparable to his (something weighing longetivity should also be in the final stat). Sobers, meanwhile, outstripped most.He averaged 10-15 more than Kanhai and Cowdrey. Only Barrington matched him (played a bit less), and G. Pollock (played a lot less). Weekes, Walcott, Hutton weren't really his contemporaries (we'll need...

• Abhi on December 21, 2010, 17:08 GMT

Ananth, No problems with taking the numbers in our stride etc. It's just that it is reasonably obvious that if a 50+ avg. batsman plays in a side with batsmen who average mostly 30, it is certain that the 50+ avg. batsman will outperform the others. I wouldn't be too surprised if there is a direct correlation with batting averages.

The point I was trying to make (perhaps for another analysis)is the best "peer" comparison would be to see how "similar" quality batsmen have performed.

As rgds. Tendulkar's (or anyone's) batting average- the peer averages fix that. Also, we assume that over long periods as you say things will "balance" out as rgds conditions, bowlers etc.

However, with a situation such as a good batsman in a weak team- this will never balance out unless the rest of the team improves. This has absolutely nothing to do with external conditions, bowlers etc.

• shrikanthk on December 21, 2010, 16:51 GMT

Ananth: Just an arbitrary query. This Bradman pic is from his 1930 Lords' innings of 254 right? Or is it from his 334 at Leeds the same year? Just curious to see if memory serves me right. [[ I have no idea since Cricinfo puts up these pictures. I will pass your comment to them to see whether they can answer. Ananth: ]]

By the way, I missed out on mentioning Gambhir's name in that list I compiled in the previous comment. An average of 50 but a "dominance index" of barely 1.5. To my mind, that is a validation of your analysis. A very typical case of a batsmen who scores runs in a pressure vaccumm on batsmen friendly conditions.

It's also interesting to find Hobbs high up on the list. Could we look at the dominance index of pre WWI Hobbs? I bet it will be as high as that of Bradman or Headley. Maybe higher! I think his pre War average was also 50+. And during that period, he had no competition within the team from the likes of Sutcliffe and Hammond. [[ Hobbs's pre-ww1 average was 57.06 in 28 matches. Post-ww1, his average was 56.85 in 33 matches. It is possible that his Inns Index average might have been higher during the pre-ww1 period. I have to see how this can be done for part of the career since this was done as a one-off exercise. Let me see. Ananth: ]]

• jignesh on December 21, 2010, 16:18 GMT

Very poor measure for test match - discredits the innings from Dravids and lower order resistances as only runs are accounted. Additional rigorous analyses might help: 1) % of total innings' balls delivered during the batsman's stay 2,3) % of resources depleted at the beginning and end of batsman's innings - How about sum of averages of players out/sum of team batting averages as resource indicator? 4) partnership contribution to total during the stay of the batsmen <--very imp for test matches. While VVS might have done bulk of scoring, Ishant's resistance in recent Aussie series should be credited 5) % of that batsman's contribution to total (currently used) 6) % of resources used for the inning One set of results with equal weightage for each. Another set of results with ur choice of weightage...ideally something that would give similar to current results. Additional info on fall of wickets and player averages before the innings sufficient. Can I get these info for picked innings? [[ This is one type of analysis based on runs, which at the end of the day are the MOST important contribution a batsman can make for his team. What you are asking for is a host of different analysies some of which have already been covered. Also where is the Balls played information for 1200+ Test matches (around two-thirds). Just not available. I am very wary of doing anything covering only the recent 15 years. Ananth: ]]

• Navin Chandar on December 21, 2010, 16:16 GMT

Can you please do the same analysis for bowlers as well? Again, like the batting postion problem, we will have the problem with spinners coming into action only after the fast bowlers. But atleast we can have a look at who takes the most number of wickets among his peers. For individual inns Kumble will be at the top. [[ Navin, Bowlers have to be handled with care because of the upper limit of 10. Laker and Kumble will be infinity. Also division of balance wickets by number of bowlers who bowled may not make sense. Your suggestion is good. However we would have to think of a better metric than just wickets. Ananth: ]]

• Abhi on December 21, 2010, 16:06 GMT

Good observation by Arjun. And equally good quick "fix" by D.V.C.

Here's a fanciful solution to "neutralise" the good batsman in weak team factor. What if we only compare scores of batsmen with similar averages? i.e say we bracket batsmen into averages of say every 10 runs. 1-10; 11-20; etc...STOP at 50+ (i.e All 50+ batsmen are considered equally good and form a single bracket) THEN we compare innings of only "like" batsmen. Don't know if you have the data to do this though. [[ I am not sure whether these fixes will work. I think we should just take these numbers in our stride and move on. The same way you are willing to accept Tendulkar's 56.93, irrespective of which of these runs were scored off good bowlers, poor bowlers, home, away, flat tracks, tough pitches, in winning causes, losing causes, first day, last day, Australia, Bangaldesh and so on. Same with 50 centuries. My personal feel, and probably different drom Unni's, is that this has more relevance as an Innings evaluation measure than the Batsman evaluation measure. I have already talked about it. this is proved by the following table of recent centuries and their Inns Index values. Hussey 195 7.35 Hussey 116 6.41 Haddin 136 4.19 Pietersen 227 3.74 Tendulkar 111 3.50 Cook 148 2.00 Kallis 201 1.97 Amla 140 1.19 De Villiers 128 1.06 Barring minor disagreements these numbers represent the true value of the innings. Note how the two double centuries differ. Also how the 111 scores over four higher scores. So we could look at this method from the innings point of view. Ananth: ]]

• Natwar Modani on December 21, 2010, 13:33 GMT

Nice work. However, I also agree with lot of readers about the fact that it favour batsman in weak teams. One possibility could be to include the overall match average in the formula, probably with little less weight than the same inning averages. After all, it is the same match. Also, there was some defense of the formula saying it is a dominance index, but the point is that dominance has to be defined with respect to the match and not just the team the batsman is playing for. I think the way suggested to handle not outs is OK, but not perfect. I also like the idea of having a position index (average of all innings in that position) which can be used for normalizing the contribution of the batsman based on his batting position. And this can be done on per inning basis, so no issues in handling that (unlike what some comments are suggesting).

• Lakshan Wanniarachchi on December 21, 2010, 8:58 GMT

Hi Ananth, I'm a regular fan of your blog, and I've enjoyed and received a lot of insight from your analysis. I have a request for a particular analysis. Could I have your email address to send my idea across? You may email me at the above address to which I will reply with my idea.

Thanks and best regards [[ You could mail me at ananth.itfigures@gmail.com Ananth: ]]

• shane on December 21, 2010, 6:38 GMT

Interesting analysis Ananth. But I am disappointed at your description of Gurusinha as an average performer and that his innings listed here was his 15 minutes of fame. Gurusinha averaged 39 at test level, scored 7 centuries (including one against Warne and McGrath at MCG) and played a major role in the 96 World Cup victory. Unfortunately, you have fallen into the trap committed by many Cricinfo contributors of believing that if you are not a champion, then you are an ordinary player. There are a lot of players who may not have been elite but had good solid careers at Test level [[ Absolutely my mistake. I plead guity to all what you have mentioned. I should have checked deeply into Gurusinha's career. The article has been corrected, My apologies to you (and mainly Asanka) Ananth: ]]

• Rahul Bansal on December 21, 2010, 5:11 GMT

Great work Anantha !! Though it is impossible to compare different players of different times, this is a nice piece of work !!

(There are so many psychological aspects that no statistical analysis can imbibe them)

• No Country for Grown Man on December 21, 2010, 4:41 GMT

I used to like your work anantha but never commented on it earlier. I am one of silent admirers of this blog. However, I completely agree with Anjum on this on., it is not fair to make comparisons like this as even though it is surely hard work for the analyst, is basically devoid of any meaning as such.

• shrikanthk on December 21, 2010, 3:48 GMT

Ananth: The index cutoff is not featuring in my comment for some strange reason. My criteria for that list was : batting average greater than 45 and index less than One point eight. (1.8). Not taking chances this time since the 1.8 went missing in the previous comment :) [[ You should not use the gt and lt signs in these comments since these are html based. Ananth: ]]

• shrikanthk on December 21, 2010, 3:32 GMT

Ananth: Here's a list of batsmen with a batting average gt 45 and relative index lt 1.8 :

Lloyd, Hassett, Dean Jones, Patsy Hendren, Hayden, Woodfull, Leyland, Azharuddin, Barlow, Worrell, Dexter, Samaraweera, Graeme Smith, Hussey, Ponting, AB De Villiers, Alistair Cook, Steve Waugh, Katich, Langer, Ponsford, Hashim Amla, Hunte, Damien Martyn, Gilchrist, Michael Clarke

Note: We should filter out Hassett, Woodfull and Ponsford. They played a significant number of Tests in the same team as Bradman (an outlier). Hence, their innings indices are low.

Among the rest : There are three reasons why a batsman is on this list -

- He played in a very strong batting side, due to which his performances don't stand out. Eg: Lloyd, Hussey, Ponting, Worrell

- He scored a lot of his runs in a pressure vaccumm in favourable conditions. i.e The batting average overrates him. Eg: Samaraweera, Cook, Smith, Azhar, Amla, De Villiers.

- He batted down the order Eg: Gilchrist [[ Many thanks, Shri. "Good as gold" extraction and comments. Ananth: ]]

• Anjum Kabir on December 21, 2010, 2:34 GMT

These calculations are utterly wrong way of looking at a batsman. It takes away the quality of his batting when compared to other premiere batsmen of his era. It can be well used to calculate, to an extent, batsmanship with in a team. But comparing different batsman from different teams is too far fetched. A team might have had batsmen who were not very far behind in talent to their premier batsman, as is the case with India. Meanwhile, there may be a team which had batsmen far behind in talent then their premiere batsman, as was the case with Lara's West Indies at the start of the century.

No comparisons like this please. [[ One thing is certain. EVERY comparison will warrant a comment like this. I suggest do not look at the batsman-level table. Instead look at the innings level table. Ananth: ]]

• unni on December 21, 2010, 2:15 GMT

@shrikanthk : more proper name for this would have been dominance index. And that is what it conveys more than average. It is simply how a batsman has dominated his team which was playing in the same condition as him. Nothing more.

@Ananth : I was bit apprehensive in using this index to analyse individual innings as it would require special case tweaks. Along career it would get adjusted itself. Further thinking about it, still you could avoid the anomalies of low scores by giving due weight-age to the dominance index according to the team score. i.e, higher the team score, the index will be higher. This could be achieved by the following modification in the formula. Let InnAverage = average innings score across all matches (as a bench mark). Let TeamScoreFactor = TeamScore/InnAverage. Multiply the dominance index by TeamScoreFactor and higher indices in the context of low team scores will get pushed down. [[ Unni, I think the Index is fine as it is. There are only two innings within the top-30 in which the team total is less than 100. One is the Hutton classic. Many people feel that it is his finest innings, at least amongst his finest. So its place is very well-deserved. Gurusinha's innings is no less. He was a competent Test performer. To come out and play an innings like that was surely heart-warming and more people will know of Gurusinha now. The real problem situations like Aamir Sohail's have been taken care of. I personally think the Innings Index itself has as much value as the batsman level compilations. Note how quickly we can get a handle on the value of Hussey's innings. Ananth: ]]

• Sanchez on December 21, 2010, 1:17 GMT

Surely someone has to point out that in nearly half of his innings, Tendulkar has performed worse than his teammates? Cant be that good then, can he? Just kidding. [[ Not even in jest. You will be pilloried. The fact is that Tendulkar's 50 centuries and the absence of very high and very low scores means that his consistency has been phenomenal. Another batsman like Lara probably has many highs and many lows also. This is probably reflected in these numbers. Ananth: ]]

Lara was amazing. It takes a special sort of batsmen to not fall into the rot, and to continue to make runs even when others around you cannot.

• shrikanthk on December 20, 2010, 17:31 GMT

I've been wondering - What does this exercise tell us which is not conveyed by the career batting averages? Ofcourse, it does help us single out those high-performing batsmen who've played in weak teams.

The other objective of this exercise is to weed out those batsmen with very high career averages, but low relative innings indices. Are we succeeding in this regard? Could we list batsmen who have a career average of over 45 but an index < 1.5. To my mind, these are the batsmen whose stature is somewhat overrated on account of the speciously good batting average. [[ Shrikanth Let us have a deal. I will add the Batting average to the table and upload the same. You can download the same, do a bit of viewing and come back with the selection. If you look at the innings methodolgy, this gives an immediate insight into the value of an innings. Note the high value of 6.41 for Hussey's innings. The revised tables have since been posted. Ananth: ]]

• unni on December 20, 2010, 16:58 GMT

@Arjun : I get the point. I could suggest a fancy way to offset it, for e.g, weigh the index by batting position/11. But I have a feeling that this aspect(lower order batsman single handedly playing) doesn't count much in a 'dominance index'. Afterall what is the significance of this? The current index computation anyway gives him the due weightage. Additional aspect which comes out is that the rest of the lower order batsmen gave him company to perform (like Sreesanth's effort along with Harbhajan). So, ideally the credit should really go to the other lower order batsmen, right? [[ Arjun/Unni, I also feel that the index should be simple and easy to understand. At a pinch the average enthusiast should be able to do the Index for Hussey's 116 quickly 116/(181/10)=6.41. It is that easy. You will know the value of the innings immediately. Complicating it will make it lose the impact. Ananth: ]]

• shrikanthk on December 20, 2010, 16:52 GMT

One way of partly removing the bias towards great batsmen in weak batting line-ups is to consider only the avg innings score of the top 6 batsmen in each innings. A batsman like Lara or Headley invariably witnesses their teams losing 20 wickets in most matches. Hence, their avg innings score vis-a-vis the "peers" will be somewhat exaggerated due to the poor tail end averages.

Ofcourse there will still be a bias if the remaining batsmen in the top 6 aren't particularly good. I suppose that can't be helped. [[ I have already used the top-10 batsmen scores across the test, I think suggested by Arjun, earlie, in a different article. This is a good idea since the comparisons would be tough. Let me also wait for Arjun's and Unni's comments. Ananth: ]]

• Swaminathan on December 20, 2010, 16:20 GMT

Another illuminating comparison would be the innings index compared to batsmen of both teams. It would reduce the strong player in a weak side effect. In fact, a comparison purely to opponents could also be interesting, since across a career one plays against a wide variety of opposing teams whose batting strength averages out. [[ Swami, the problem would be the wide disparity in the bowling strength of the two teams. So the comparisons will not be correct.. We will lose the "almost identical conditions" benefit. Ananth: ]]

I am really pleased at the quality of analysis you produce. I have long followed sabermetrics (baseball analytics) because they do brilliant statistical analytics work, producing findings that are absolutely mindboggling. Good to see similar excellent quality work being done on cricket statistics, though of course given the more limited base of teams and matches, it is tough to do quite as much with the numbers.

• shrikanthk on December 20, 2010, 16:20 GMT

Nice piece. Some Observations :

- The absence of Headley! : Am I guilty of oversight here? Can't spot his name in the list! Surely, his Index should be greater than 2! [[ Shri, I owe you (and the other readers) a big aplogy. I forgot to remove an extra filter I had put in and a few 2000+ batsmen were left out. Unfortunately Headley was the most important one. In fact he is in second position with a 3+ figure. the article has since been corrected and the download file has been uploaded. Many thanks for a prompt comment. Once again my apologies. Ananth: ]]

- The best thing about this analysis is that it appropriately discounts those batsmen who score very heavily in collective runfests on flat tracks (Zaheer Abbas and Ganguly for instance)

- Nevertheless, there will remain a bias towards batsmen who've been dominant in somewhat weak batting line-ups. (Eg: Lara, HW Taylor among others). Which is why I'm astonished to find Headley missing altogether!

- Bradman remains the Don. No surprises there. There was a phase in his career when that figure might have been well over 4. I'm referring to the keenly contested 1936-37 and 1938 Ashes series, in which Bradman alone was the primary reason for Australia's competitiveness. (Ananth can confirm this hypothesis) [[ Let me look at this short period. Ananth: ]]

• David Reinecke on December 20, 2010, 16:11 GMT

To adjust for batting position: calculate the overall average innings index per batting position and then indicate the individual batsmans innings index relative to the average for his position. What makes this tricky of course is that players do not always bat in the same position, so good luck with the math :-) [[ Fine idea but a tough task, David. Let me see. Ananth: ]]

• Ananth on December 20, 2010, 15:32 GMT

Arjun The tables have been re-posted after incorporating Batting Position.

• D.V.C. on December 20, 2010, 14:11 GMT

When it comes to not outs, why not just do what is usually done with not outs? i.e. If someone has batted 80 times but been not out 10 times, then divide their total Innings Index by 70 instead of 80. It's a simple solution – solves the problem of where to draw the line as to when to count an innings and when not to, and helps balance out the current bias toward those who bat higher in the order. If I bat at 6 then I have less chance to score runs than the openers, but I have more chance of being not out. [[ Daniel Your response has come after quite some gap. Let me wait for some ore responses. In principle looks good. Ananth: ]]

• Arjun on December 20, 2010, 13:54 GMT

For individual innings add a new column of batting position. looks like opnening batsman or top order have an advantage; in which case innings played by kapildev(129) at bat pos # 7 is extraordinary. To outscore your teammates by 16 times batting at no.7 is something. [[ Excellent idea. Will complete and post the revised table by morning. Ananth: ]] Can 'inn-index' be tweaked/adjusted somewhat according to batting position batted ? [[ I cannot off-hand think of a satisfactory method. Let me see what Unni reponds. Ananth: ]]

• No featured comments at the moment.

• Arjun on December 20, 2010, 13:54 GMT

For individual innings add a new column of batting position. looks like opnening batsman or top order have an advantage; in which case innings played by kapildev(129) at bat pos # 7 is extraordinary. To outscore your teammates by 16 times batting at no.7 is something. [[ Excellent idea. Will complete and post the revised table by morning. Ananth: ]] Can 'inn-index' be tweaked/adjusted somewhat according to batting position batted ? [[ I cannot off-hand think of a satisfactory method. Let me see what Unni reponds. Ananth: ]]

• D.V.C. on December 20, 2010, 14:11 GMT

When it comes to not outs, why not just do what is usually done with not outs? i.e. If someone has batted 80 times but been not out 10 times, then divide their total Innings Index by 70 instead of 80. It's a simple solution – solves the problem of where to draw the line as to when to count an innings and when not to, and helps balance out the current bias toward those who bat higher in the order. If I bat at 6 then I have less chance to score runs than the openers, but I have more chance of being not out. [[ Daniel Your response has come after quite some gap. Let me wait for some ore responses. In principle looks good. Ananth: ]]

• Ananth on December 20, 2010, 15:32 GMT

Arjun The tables have been re-posted after incorporating Batting Position.

• David Reinecke on December 20, 2010, 16:11 GMT

To adjust for batting position: calculate the overall average innings index per batting position and then indicate the individual batsmans innings index relative to the average for his position. What makes this tricky of course is that players do not always bat in the same position, so good luck with the math :-) [[ Fine idea but a tough task, David. Let me see. Ananth: ]]

• shrikanthk on December 20, 2010, 16:20 GMT

Nice piece. Some Observations :

- The absence of Headley! : Am I guilty of oversight here? Can't spot his name in the list! Surely, his Index should be greater than 2! [[ Shri, I owe you (and the other readers) a big aplogy. I forgot to remove an extra filter I had put in and a few 2000+ batsmen were left out. Unfortunately Headley was the most important one. In fact he is in second position with a 3+ figure. the article has since been corrected and the download file has been uploaded. Many thanks for a prompt comment. Once again my apologies. Ananth: ]]

- The best thing about this analysis is that it appropriately discounts those batsmen who score very heavily in collective runfests on flat tracks (Zaheer Abbas and Ganguly for instance)

- Nevertheless, there will remain a bias towards batsmen who've been dominant in somewhat weak batting line-ups. (Eg: Lara, HW Taylor among others). Which is why I'm astonished to find Headley missing altogether!

- Bradman remains the Don. No surprises there. There was a phase in his career when that figure might have been well over 4. I'm referring to the keenly contested 1936-37 and 1938 Ashes series, in which Bradman alone was the primary reason for Australia's competitiveness. (Ananth can confirm this hypothesis) [[ Let me look at this short period. Ananth: ]]

• Swaminathan on December 20, 2010, 16:20 GMT

Another illuminating comparison would be the innings index compared to batsmen of both teams. It would reduce the strong player in a weak side effect. In fact, a comparison purely to opponents could also be interesting, since across a career one plays against a wide variety of opposing teams whose batting strength averages out. [[ Swami, the problem would be the wide disparity in the bowling strength of the two teams. So the comparisons will not be correct.. We will lose the "almost identical conditions" benefit. Ananth: ]]

I am really pleased at the quality of analysis you produce. I have long followed sabermetrics (baseball analytics) because they do brilliant statistical analytics work, producing findings that are absolutely mindboggling. Good to see similar excellent quality work being done on cricket statistics, though of course given the more limited base of teams and matches, it is tough to do quite as much with the numbers.

• shrikanthk on December 20, 2010, 16:52 GMT

One way of partly removing the bias towards great batsmen in weak batting line-ups is to consider only the avg innings score of the top 6 batsmen in each innings. A batsman like Lara or Headley invariably witnesses their teams losing 20 wickets in most matches. Hence, their avg innings score vis-a-vis the "peers" will be somewhat exaggerated due to the poor tail end averages.

Ofcourse there will still be a bias if the remaining batsmen in the top 6 aren't particularly good. I suppose that can't be helped. [[ I have already used the top-10 batsmen scores across the test, I think suggested by Arjun, earlie, in a different article. This is a good idea since the comparisons would be tough. Let me also wait for Arjun's and Unni's comments. Ananth: ]]

• unni on December 20, 2010, 16:58 GMT

@Arjun : I get the point. I could suggest a fancy way to offset it, for e.g, weigh the index by batting position/11. But I have a feeling that this aspect(lower order batsman single handedly playing) doesn't count much in a 'dominance index'. Afterall what is the significance of this? The current index computation anyway gives him the due weightage. Additional aspect which comes out is that the rest of the lower order batsmen gave him company to perform (like Sreesanth's effort along with Harbhajan). So, ideally the credit should really go to the other lower order batsmen, right? [[ Arjun/Unni, I also feel that the index should be simple and easy to understand. At a pinch the average enthusiast should be able to do the Index for Hussey's 116 quickly 116/(181/10)=6.41. It is that easy. You will know the value of the innings immediately. Complicating it will make it lose the impact. Ananth: ]]

• shrikanthk on December 20, 2010, 17:31 GMT

I've been wondering - What does this exercise tell us which is not conveyed by the career batting averages? Ofcourse, it does help us single out those high-performing batsmen who've played in weak teams.

The other objective of this exercise is to weed out those batsmen with very high career averages, but low relative innings indices. Are we succeeding in this regard? Could we list batsmen who have a career average of over 45 but an index < 1.5. To my mind, these are the batsmen whose stature is somewhat overrated on account of the speciously good batting average. [[ Shrikanth Let us have a deal. I will add the Batting average to the table and upload the same. You can download the same, do a bit of viewing and come back with the selection. If you look at the innings methodolgy, this gives an immediate insight into the value of an innings. Note the high value of 6.41 for Hussey's innings. The revised tables have since been posted. Ananth: ]]

• Sanchez on December 21, 2010, 1:17 GMT

Surely someone has to point out that in nearly half of his innings, Tendulkar has performed worse than his teammates? Cant be that good then, can he? Just kidding. [[ Not even in jest. You will be pilloried. The fact is that Tendulkar's 50 centuries and the absence of very high and very low scores means that his consistency has been phenomenal. Another batsman like Lara probably has many highs and many lows also. This is probably reflected in these numbers. Ananth: ]]

Lara was amazing. It takes a special sort of batsmen to not fall into the rot, and to continue to make runs even when others around you cannot.