After the RSA series, I began to think about whether Dale Steyn belongs in the discussion of ATG (All Time Great) bowlers. So the fundamental question I got was "how one would project Steyn's career?"
Steyn's stats to date bear repeating: 46 Tests, 238 wickets @ 23.2, a mind-boggling strike-rate of 40 and 5.2 wickets / match. So, the first step was to find bowlers who were "similar" to Steyn.
So, lets begin by finding pace bowlers who ended up with more than Steyn's current haul of 238 wickets. That gives us 25+ names, so lets cull it further by restricting it to quicks who took their wickets at the rate of at least 4 wickets / match over their career.
This criterion eliminated a whole bunch of folks, including Walsh, Kapil, Pollock, Ntini, Botham, Vaas, Willis, Gillespie, and Statham. This also eliminated the incomparable Barnes who took 189 wickets in
27 Tests at 7 wickets / Test!!!! Since I've used "mind-boggling" for Steyn's 5.2 w/Test, no suitable adjective remains. Which is fine, as the goal is to find bowlers who are similar to Steyn, and Barnes isn't similar to anyone else, in the sense of Bradman.
I would like to include Pollock in the analysis at least initially as he played in very similar conditions against similar opposition, and that can be achieved by relaxing the cut-off to 3.9 wickets / Test. That adds Pollock & Courtney Walsh back into the list. This first cut of the list, leaves us with a list of 16+1 pace bowlers, ranging from the good to fabulous -- with one criterion in common; they took a lot of wickets at ~4w/Test or higher.
Ambrose, Donald, Garner, Hadlee, Holding, Imran, Lee, Lillee, Marshall, McDermott, McGrath, Pollock, Steyn, Trueman, Walsh, Waqar, Wasim
Not all of them should belong in the discussion, but lets try to find objective criteria.
So, how would one go about finding "similar" bowlers to Dale at the same stages of their career? How about looking at everyone at the 46-match stage in their career. Here are some criteria to consider:
A Bowling Average (the traditional measure of a bowlers success)
B Wickets / match (or total wickets as we are looking at the same # of Tests)
C Strike rate (deliveries / wicket)
D Overs bowled (a measure of the wear and tear on the bowler)
E Dominance score (# of 5-fers & 3x10-fers)
F PI score (courtesy Mike Holmans) = average * strike rate
Clearly some of the criteria are overemphasized (D can be computed from B & C, F from A & C) but for now, let them stay in for simplicity -- can be fixed by eliminating it and adding more weight to other criteria. You can also argue if 3x is the right weight for a 10-fer, given that the 5-fers are double-counted in the 10-fer. The idea isn't to get the criteria precisely right, but to create a basis for discussion.
So lets now look at each of these players at the 46 Test stage of their careers, and see where Steyn fits in
this list & find potential outliers to eliminate from the list.
A) Average: Steyn @ 23.2 is #9 of 17 flanked by McGrath & Imran -- also, at the very bottom of the list are a couple of clear outliers in Lee (31.8) and McDermott (28.2)
B) Wickets / match or (total wickets): Steyn is #1 of 17, but closely followed by Lillee, Waqar & Donald, all > 5w/Test. Walsh is a clear outlier with only 161 wickets (3.5w/Test) and Lee @ 3.9 w/Test is below the original 4w/Test cutoff. (Interestingly Walsh did a lot better in his last 80-odd Tests than his first 46)
C) Strike Rate: Steyn & Waqar head the list, but no real outliers with all but 4 of the list having 50+ strike rates. This is interesting because it indicates that only Waqar & Marshall are comparable to Steyn in this
attribute.
D) Overs: Walsh with <1500 overs bowled the least (probably a factor in his longevity) and Lee, Waqar (that name again) and Steyn, all have below 1600 overs.
E) Dominance Score (5-fers + 3x 10-fers): Lillee and Waqar head the list with 32 and 31 respectively, followed by Steyn, Imran and Ambrose in a clump ahead of the pack. Changing the 10-fer multiplier to 2, obviously changes the scores, but provides the same qualitative description and separation. (A general test to see how stable a measure is, is to tweak it somewhat and determine if it still provides a similar qualitative result.)
F) PI: Waqar, Steyn and Marshall head the list (only one's below 1000), with Lee & McDermott clear outliers, and possibly Hadlee.
So what have we learned so far. Actually, not a whole lot: Lee & McDermott don't belong in this list which is hardly surprising news. Steyn is near the top in all measures except average, but since all the other measures capture strike rate in some way, he figures to be on the top of the list.
Steyn seems quite similar to Waqar, and possibly Lillee and Marshall, but we will figure that out later.
Now lets look at their entire careers. It would be good to have the ages of each bowler at their 46th Test, but I didn't have an easy way to add the data, so I'm going to use the total Tests played as a proxy.
Although Steyn's career could end in the next game he plays, it is reasonable to expect he is right now in his prime, and will play another 20-40 Tests. So, lets eliminate the bowlers who were towards the end of their careers at the 46th Test by using total Tests played as proxy, which eliminates Holding and Garner.
So culling this list, we get a murderer's row of a dozen bowlers (no Lee, Walsh, McDermott, Holding, Garner), many of whom would be part of any discussion on "Best bowlers ever", and in fact were discussed in CricInfo's all-time XIs:
Ambrose, Donald, Hadlee, Imran, Lillee, Marshall, McGrath, Pollock, Steyn, Trueman, Waqar, Wasim
Lets look to see if Steyn belongs in this august list, which was were this discussion started. There is NO question that he does. His average of 23.2 is "only" 8th, but better than Imran, Lillee, Wasim and Hadlee at a similar stage of their careers. On all other measures, PI, SR, Overs bowled, he stands at or near the
top of the lists.
Pollock is perhaps an anomaly in that we would not figure in anyone's list of the greatest ever bowlers, but we also forget what a fabulous bowler the young Pollock was -- 190w @ 20.5 is very very very good.
So the purpose of this post is to identify the basic list of top fast bowlers, and use them as a basis for
a) defining similarity metrics to find bowlers similar to Steyn (or others) at the same stage of their careers (46 Tests)
b) project Steyn (and other bowler's performances) looking at these criterion
c) verifying projections for bowlers whose career's we know are over to determine if the criteria we use are reasonable
These verified criteria would be the basis of a future (no promises when) analysis of best bowlers.
The Dynamic Dozen again (heavily weighted towards modern bowlers because of the 238 wicket cut off!):
Ambrose,
Donald,
Hadlee,
Imran,
Lillee,
Marshall,
McGrath,
Pollock,
Steyn,
Trueman,
Waqar,
Wasim
I haven't done the stats analysis yet, but guesses would be welcome as to
a) Who are the most similar (and dis-similar) bowlers from this list to Steyn at the 46-Test stage of their careers?
b) Which two bowlers from this list would be the most similar? most dissimilar? Again at the 46-Test stage of their careers.
Comments on the selection methodology welcome. Also, once set up, it will then be appropriate to talk of further adjustments for opposition quality, # of Tests played in a career, different eras. But as all these folks with the exception of Lillee and Trueman bowled in the last 25 years, that may not be needed now.
Bharat
ObDisclaimer: I may at some point choose to write a book on cricket statistics. It is one of my "bucket list" items. So any material you choose to post publicly may well lead to insights that find its way in
the book, certainly without obligation of any sorts, and possibly without attribution due to oversight (though I will certainly try to remember if someone provided starting insights and thank them). All this may well be moot, as my bucket list keeps growing faster than the items I cross off...
Also, thanks to the fabulous folks who provided CricInfo's Statsguru tool, without which none of this analysis would be easy. stats.espncricinfo.com/guru
Steyn's stats to date bear repeating: 46 Tests, 238 wickets @ 23.2, a mind-boggling strike-rate of 40 and 5.2 wickets / match. So, the first step was to find bowlers who were "similar" to Steyn.
So, lets begin by finding pace bowlers who ended up with more than Steyn's current haul of 238 wickets. That gives us 25+ names, so lets cull it further by restricting it to quicks who took their wickets at the rate of at least 4 wickets / match over their career.
This criterion eliminated a whole bunch of folks, including Walsh, Kapil, Pollock, Ntini, Botham, Vaas, Willis, Gillespie, and Statham. This also eliminated the incomparable Barnes who took 189 wickets in
27 Tests at 7 wickets / Test!!!! Since I've used "mind-boggling" for Steyn's 5.2 w/Test, no suitable adjective remains. Which is fine, as the goal is to find bowlers who are similar to Steyn, and Barnes isn't similar to anyone else, in the sense of Bradman.
I would like to include Pollock in the analysis at least initially as he played in very similar conditions against similar opposition, and that can be achieved by relaxing the cut-off to 3.9 wickets / Test. That adds Pollock & Courtney Walsh back into the list. This first cut of the list, leaves us with a list of 16+1 pace bowlers, ranging from the good to fabulous -- with one criterion in common; they took a lot of wickets at ~4w/Test or higher.
Ambrose, Donald, Garner, Hadlee, Holding, Imran, Lee, Lillee, Marshall, McDermott, McGrath, Pollock, Steyn, Trueman, Walsh, Waqar, Wasim
Not all of them should belong in the discussion, but lets try to find objective criteria.
So, how would one go about finding "similar" bowlers to Dale at the same stages of their career? How about looking at everyone at the 46-match stage in their career. Here are some criteria to consider:
A Bowling Average (the traditional measure of a bowlers success)
B Wickets / match (or total wickets as we are looking at the same # of Tests)
C Strike rate (deliveries / wicket)
D Overs bowled (a measure of the wear and tear on the bowler)
E Dominance score (# of 5-fers & 3x10-fers)
F PI score (courtesy Mike Holmans) = average * strike rate
Clearly some of the criteria are overemphasized (D can be computed from B & C, F from A & C) but for now, let them stay in for simplicity -- can be fixed by eliminating it and adding more weight to other criteria. You can also argue if 3x is the right weight for a 10-fer, given that the 5-fers are double-counted in the 10-fer. The idea isn't to get the criteria precisely right, but to create a basis for discussion.
So lets now look at each of these players at the 46 Test stage of their careers, and see where Steyn fits in
this list & find potential outliers to eliminate from the list.
A) Average: Steyn @ 23.2 is #9 of 17 flanked by McGrath & Imran -- also, at the very bottom of the list are a couple of clear outliers in Lee (31.8) and McDermott (28.2)
B) Wickets / match or (total wickets): Steyn is #1 of 17, but closely followed by Lillee, Waqar & Donald, all > 5w/Test. Walsh is a clear outlier with only 161 wickets (3.5w/Test) and Lee @ 3.9 w/Test is below the original 4w/Test cutoff. (Interestingly Walsh did a lot better in his last 80-odd Tests than his first 46)
C) Strike Rate: Steyn & Waqar head the list, but no real outliers with all but 4 of the list having 50+ strike rates. This is interesting because it indicates that only Waqar & Marshall are comparable to Steyn in this
attribute.
D) Overs: Walsh with <1500 overs bowled the least (probably a factor in his longevity) and Lee, Waqar (that name again) and Steyn, all have below 1600 overs.
E) Dominance Score (5-fers + 3x 10-fers): Lillee and Waqar head the list with 32 and 31 respectively, followed by Steyn, Imran and Ambrose in a clump ahead of the pack. Changing the 10-fer multiplier to 2, obviously changes the scores, but provides the same qualitative description and separation. (A general test to see how stable a measure is, is to tweak it somewhat and determine if it still provides a similar qualitative result.)
F) PI: Waqar, Steyn and Marshall head the list (only one's below 1000), with Lee & McDermott clear outliers, and possibly Hadlee.
So what have we learned so far. Actually, not a whole lot: Lee & McDermott don't belong in this list which is hardly surprising news. Steyn is near the top in all measures except average, but since all the other measures capture strike rate in some way, he figures to be on the top of the list.
Steyn seems quite similar to Waqar, and possibly Lillee and Marshall, but we will figure that out later.
Now lets look at their entire careers. It would be good to have the ages of each bowler at their 46th Test, but I didn't have an easy way to add the data, so I'm going to use the total Tests played as a proxy.
Although Steyn's career could end in the next game he plays, it is reasonable to expect he is right now in his prime, and will play another 20-40 Tests. So, lets eliminate the bowlers who were towards the end of their careers at the 46th Test by using total Tests played as proxy, which eliminates Holding and Garner.
So culling this list, we get a murderer's row of a dozen bowlers (no Lee, Walsh, McDermott, Holding, Garner), many of whom would be part of any discussion on "Best bowlers ever", and in fact were discussed in CricInfo's all-time XIs:
Ambrose, Donald, Hadlee, Imran, Lillee, Marshall, McGrath, Pollock, Steyn, Trueman, Waqar, Wasim
Lets look to see if Steyn belongs in this august list, which was were this discussion started. There is NO question that he does. His average of 23.2 is "only" 8th, but better than Imran, Lillee, Wasim and Hadlee at a similar stage of their careers. On all other measures, PI, SR, Overs bowled, he stands at or near the
top of the lists.
Pollock is perhaps an anomaly in that we would not figure in anyone's list of the greatest ever bowlers, but we also forget what a fabulous bowler the young Pollock was -- 190w @ 20.5 is very very very good.
So the purpose of this post is to identify the basic list of top fast bowlers, and use them as a basis for
a) defining similarity metrics to find bowlers similar to Steyn (or others) at the same stage of their careers (46 Tests)
b) project Steyn (and other bowler's performances) looking at these criterion
c) verifying projections for bowlers whose career's we know are over to determine if the criteria we use are reasonable
These verified criteria would be the basis of a future (no promises when) analysis of best bowlers.
The Dynamic Dozen again (heavily weighted towards modern bowlers because of the 238 wicket cut off!):
Ambrose,
Donald,
Hadlee,
Imran,
Lillee,
Marshall,
McGrath,
Pollock,
Steyn,
Trueman,
Waqar,
Wasim
I haven't done the stats analysis yet, but guesses would be welcome as to
a) Who are the most similar (and dis-similar) bowlers from this list to Steyn at the 46-Test stage of their careers?
b) Which two bowlers from this list would be the most similar? most dissimilar? Again at the 46-Test stage of their careers.
Comments on the selection methodology welcome. Also, once set up, it will then be appropriate to talk of further adjustments for opposition quality, # of Tests played in a career, different eras. But as all these folks with the exception of Lillee and Trueman bowled in the last 25 years, that may not be needed now.
Bharat
ObDisclaimer: I may at some point choose to write a book on cricket statistics. It is one of my "bucket list" items. So any material you choose to post publicly may well lead to insights that find its way in
the book, certainly without obligation of any sorts, and possibly without attribution due to oversight (though I will certainly try to remember if someone provided starting insights and thank them). All this may well be moot, as my bucket list keeps growing faster than the items I cross off...
Also, thanks to the fabulous folks who provided CricInfo's Statsguru tool, without which none of this analysis would be easy. stats.espncricinfo.com/guru
