Don't think you give Starc anywhere near the credit he deserves. I agree Boland is as good but I'd never pick him over Starc (I would pick him over Hazlewood or maybe Cummins). Starc adds mongrel to the side that is unquantifiable that Boland doesn't have. And has a pretty well proven knack of picking up wickets at crucial points. You love your stats, yet don't refer to them with Starc. 382 at 27 and a SR of 48 is exceptional.
I would also say that's a very unfair assessment of Brook, who has been game-changing in his role. Bavuma may have a higher average but it's silly not to consider Brook. I can't stand him (and I love Bavuma) but you can't deny him. I'd back a WTC XI with Brook over one with Bavuma.
I don't say any of this to be mean, by the way, it's all fun debate and you do a great job.
Here's hoping England can win a fucking Test. Wild they have won one down here since that 11 series... which I remember very well. Everyone remember's Siddle's hat trick, but not us bowling on Trott's pads for the days thereafter as England notched up 1/500.
It’s true that in the past I’ve criticised Starc only to see him excel.
I’d love to pick him but in all honesty think the other 3 squeeze him out. His career average of 27.6 is very good. I know many people are fans of bowlers’ Test strike rates but I am less so as I think average already covers it.
Last summer Boland took his wickets at 13, Starc at 29. I just think Boland is more likely to do well than Starc - although I expect Starc will do well himself.
As for Brook - I wasn’t criticising him, just that CA were using averages until it didn’t suit them!
Some good stuff in here. On Ball 5: the average ‘cost’ of a wicket is 5 runs, but as I understand it this figure starts higher and diminishes as the innings goes on. So a wicket first ball of the innings has a cost much higher than 5, whereas a wicket off the last ball has a cost much closer to zero. Is it possible to build this into your model?!
Hey Ben - thank you! Yes, it is an interesting point. I was going to include a paragraph on it but it was getting a bit long so I didn’t. It would be possible to do, especially if one had access to the professional Duckworth Lewis Stern tables. Although it could also be argued that even though wickets at different points have differing degrees of value this should not be taken into account when purely considering how good a bowler is.
Maybe a better approach would be to factor in the ability of the batter dismissed. Then there is also the fact that runs conceded in different overs are not equal either. 10 runs off the 20th over might well be good whereas eight runs off the seventh might not be good. And then there are all the features of luck that go into it – dropped catches, beating the bat by a centimetre rather than picking up the edge and so on. I suspect cricviz probably have some sort of metric that takes into account all those things. And it might be pretty valuable. The more blunt instruments that cricket tends to use hopefully become pretty close to accurate once the sample size gets large enough.
Thanks for the detailed reply Paul! I guess using the batsman’s control percentage could incorporate some of the ‘luck’ element you talk about – I.e. a false shot is counted as such whether or not it results in a wicket.
Yes, it is fascinating. It would be nice to have all the raw data - and either stats expert or smart AI - to produce something definitive, and then see if ultimately this proved more useful than the blunt tools we have now.
I think you've got some click bait too! 😉
Don't think you give Starc anywhere near the credit he deserves. I agree Boland is as good but I'd never pick him over Starc (I would pick him over Hazlewood or maybe Cummins). Starc adds mongrel to the side that is unquantifiable that Boland doesn't have. And has a pretty well proven knack of picking up wickets at crucial points. You love your stats, yet don't refer to them with Starc. 382 at 27 and a SR of 48 is exceptional.
I would also say that's a very unfair assessment of Brook, who has been game-changing in his role. Bavuma may have a higher average but it's silly not to consider Brook. I can't stand him (and I love Bavuma) but you can't deny him. I'd back a WTC XI with Brook over one with Bavuma.
I don't say any of this to be mean, by the way, it's all fun debate and you do a great job.
Here's hoping England can win a fucking Test. Wild they have won one down here since that 11 series... which I remember very well. Everyone remember's Siddle's hat trick, but not us bowling on Trott's pads for the days thereafter as England notched up 1/500.
Hi Daniel - thanks for the comment:)
It’s true that in the past I’ve criticised Starc only to see him excel.
I’d love to pick him but in all honesty think the other 3 squeeze him out. His career average of 27.6 is very good. I know many people are fans of bowlers’ Test strike rates but I am less so as I think average already covers it.
Last summer Boland took his wickets at 13, Starc at 29. I just think Boland is more likely to do well than Starc - although I expect Starc will do well himself.
As for Brook - I wasn’t criticising him, just that CA were using averages until it didn’t suit them!
Some good stuff in here. On Ball 5: the average ‘cost’ of a wicket is 5 runs, but as I understand it this figure starts higher and diminishes as the innings goes on. So a wicket first ball of the innings has a cost much higher than 5, whereas a wicket off the last ball has a cost much closer to zero. Is it possible to build this into your model?!
Hey Ben - thank you! Yes, it is an interesting point. I was going to include a paragraph on it but it was getting a bit long so I didn’t. It would be possible to do, especially if one had access to the professional Duckworth Lewis Stern tables. Although it could also be argued that even though wickets at different points have differing degrees of value this should not be taken into account when purely considering how good a bowler is.
Maybe a better approach would be to factor in the ability of the batter dismissed. Then there is also the fact that runs conceded in different overs are not equal either. 10 runs off the 20th over might well be good whereas eight runs off the seventh might not be good. And then there are all the features of luck that go into it – dropped catches, beating the bat by a centimetre rather than picking up the edge and so on. I suspect cricviz probably have some sort of metric that takes into account all those things. And it might be pretty valuable. The more blunt instruments that cricket tends to use hopefully become pretty close to accurate once the sample size gets large enough.
Thanks for the detailed reply Paul! I guess using the batsman’s control percentage could incorporate some of the ‘luck’ element you talk about – I.e. a false shot is counted as such whether or not it results in a wicket.
Yes, it is fascinating. It would be nice to have all the raw data - and either stats expert or smart AI - to produce something definitive, and then see if ultimately this proved more useful than the blunt tools we have now.