Trivia - bowling February 15, 2008

Bowler consistency analysis - a follow-up

My previous post on bowler consistency has attracted plenty of comments

The post on "Bowler Consistency" received many comments, some silly, some sceptical, some dismissive, some appreciative and some happy at the thinking process it initiated. There were many relevant comments, which warrant a follow-up post.

First of all, an apology to the readers. I used "spell" when I really meant "innings spell". A spell is an uninterrupted stint of bowling. What I really meant was the bowling done during an innings. So I have coined an alternate term called "innspell" which is exactly what it means, the complete bowling effort during an innings, often consisting of multiple spells. Many thanks to the readers who took me to task on this issue.

There were many relevant comments on bowler strike-rates and other pertinent measures such as bowling support, pressure situations, bowling accuracy etc. I do not want to mix up the criteria. Bowler strike-rate is not to be confused with the ability of bowlers to be more consistent. That is one of the most important of bowler measures and warrants a separate post. A similar situation exists with the other measures as well.

I am also determined that I will keep the analysis as simple as when the post started. Finally, one factor should not be forgotten: what I have stated is that if a spin bowler bowls 11 overs or above and takes at least a wicket, this innspell is considered to be a success as compared to a bowler who bowls a similar innspell and comes out wicketless. There is nothing fundamentally wrong with this statement.

However the one comment which also impressed me a lot was the suggestion to reward bowlers for taking more than one wicket in an innsspell. Hence I have added a simple linear weightage factor to a successful innsspell. For each wicket captured, a weightage of 10% is given. Thus a 1-wkt capture makes this value as 1.1, 2-wkt capture as 1.2 and so on upto the two 10-wicket captures of Laker and Kumble as 2.0. This tweak should satisfy most readers. The Bowler Consistency Index value is computed as a % of "number of relevant spells x 2.0".

Readers will note that this method will increase the consistency ratings of bowlers who have captured more wickets in their career. An alternative would have been to consider a one-wickett haul to have a value of 1.0 and anything more as 1.1. This would still have left the disparity between insspells which fetched two wickets, and those that fetched more than two.

The most significant comment was that 30 innspells are not sufficient. The lowering of the bar has allowed quite a few relatively insignificant bowlers to walk through at the expense of bowlers who have served for longer periods. Hence I have raised the bar, and considered only the bowlers who have bowled a minimum of 50 innspells, which translates to around 30 Tests, possibly 4-5 years of Test cricket. A total of 143 bowlers have qualified under this criteria compared to 276 bowlers with the lower cut-off.

There were a few (mostly unwarranted) comments on Mervyn Dillon. He has taken 131 wickets in 38 Tests, not a bad haul. Just taking a "wickets per Test" criterion, he is ahead of more well-known bowlers such as Derek Underwood, Kapil Dev, Vaas, Zaheer Khan, Daniel Vettori, Sarfraz Nawaz, Michael Kasprowicz and S Venkataraghavan. Comments should be made only after verifying facts and making allowance for his playing in a weak West Indies team all his career.

The revised table is given below.

Bowler Consistency Analysis (Revised with weightage) - Min 50 relevant spells
SNo Bowler            Bow Ctry  Mat Relevant Successful BCIdx  Wkts Wkt/IS
1.Muralitharan M    ROB  Slk  118    201     267.3    66.49   723  3.60
2.Grimmett C.V      RLB  Aus   37     66      84.6    64.09   216  3.27
3.Kumble A          RLB  Ind  125    215     262.4    61.02   604  2.81
4.Hadlee R.J        RFM  Nzl   86    138     168.1    60.91   431  3.12
5.Bedi B.S          LSP  Ind   67    107     129.6    60.56   266  2.49
6.Warne S.K         RLB  Aus  145    250     302.8    60.56   708  2.83
7.Donald A.A        RF   Saf   72    124     150.0    60.48   330  2.66
8.Trueman F.S       RF   Eng   67    115     138.7    60.30   307  2.67
9.Lillee D.K        RF   Aus   70    124     149.5    60.28   355  2.86
10.Gupte S.P         RLB  Ind   36     54      64.9    60.09   149  2.76
11.Dillon M          RFM  Win   38     55      66.1    60.09   131  2.38
12.Imran Khan        RF   Pak   88    125     150.2    60.08   362  2.90
13.MacGill S.C.G     RLB  Aus   42     76      91.3    60.07   203  2.67
14.Wasim Akram       LFM  Pak  104    161     192.4    59.75   414  2.57
15.Danish Kaneria    RLB  Pak   51     84     100.0    59.52   220  2.62
16.Croft C.E.H       RF   Win   27     50      59.5    59.50   125  2.50
17.Chandrasekhar B.S RLB  Ind   58     87     103.2    59.31   242  2.78
18.Saqlain Mushtaq   ROB  Pak   49     78      91.8    58.85   208  2.67
19.Gough D           RF   Eng   58     91     106.9    58.74   229  2.52
20.Laker J.C         ROB  Eng   46     77      90.3    58.64   193  2.51
21.Trumble H         ROB  Aus   32     53      62.1    58.58   141  2.66
22.Ambrose C.E.L     RF   Win   98    164     191.5    58.38   405  2.47
23.Marshall M.D      RF   Win   81    147     171.6    58.37   376  2.56
24.Bishop I.R        RF   Win   43     67      78.1    58.28   161  2.40
25.Cork D.G          RFM  Eng   37     57      66.1    57.98   131  2.30
(Click here for the full table.)

Muralitharan is on top, followed by Grimmett, Kumble, Hadlee and Bedi. This is not a bad quintet. The next five bowlers are Warne, Donald, Trueman, Lillee and Subash Gupte. There should be no complaints there either. The top 25 table now includes most bowling stalwarts, which should satisfy most readers. The weightage of values and raising the cut-off bar has taken away most of the lesser, shorter-duration bowlers.

I have added another variable, the average wickets per innspell, which is indicative of the bowler performance. Muralitharan is way ahead of the other bowlers, having taken 3.6 wickets per innspell. Compare the numbers for some of the other greats. Donald - 2.66, Kumble - 2.81, Pollock - 2.30, Warne - 2.83, Lillee - 2.86 and Ambrose - 2.47. It shows what a wonderful wicket-taking bowler Muralitharan is, even if you concede that the bowler at the other end was not always competing with him in taking wickets. Note also that Flintoff is the only bowler in the top-50 whose wickets per innsspell figure is less than two.

The unfortunate aspect of increasing the cut-off is that this has excluded the pre-World War I all-time greats such as Sydney Barnes, George Lohmann etc. In order to be fair to these great bowlers I have given below a list of five such pre-WW I bowlers.

Barnes S.F          RFM  Eng   27     48     64.9     67.60   189  3.94
Lohmann G.A         RFM  Eng   18     32     40.2     62.81   112  3.50
Giffen G            ROB  Aus   31     35     43.3     61.86   103  2.94
Briggs J            LSP  Eng   33     42     51.8     61.67   118  2.81
Blythe C            LSP  Eng   19     35     43.0     61.43   100  2.86
Look at Barnes' wickets per innspell figure - it is higher than Muralitharan's.

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