*Official* Warne vs Murali Discussion

[Richard]

Yeah, meant "in one continuous thread with (against?) the same people over a continuous age" TBH. If by some chance it was possible to find every post exclusively on the subject of first-chance scores and add them to one thread, I'd be surprised if I had less than 2,000 in there TBH.

That's one subject that does remain enjoyable to me, unlike Warne vs Murali.

 

[Richard]

The minute I screwed up the smilie I figured you'd correct me.

You're quite the predictable character sometimes Richard

It truly does amaze me actually how many people screw-up the smiley so regularly. It must be easily the most common incorrectly-notated smiley on the CW forums.

(And it's exaggerated again by the fact that it DID used to be notated as :P so all smilies before The Great vBulletin Smiley Reorganisation Of '04 are now wrong. Another example is that used to be notated as :rolleyes: so hence you'll see ":rolleyes:" on all posts of a certain age.)

 

[GIMH]

Yeah, and on Heath's posts as well

Wish smilies weren't case-sensitive tbh

 

[pasag]

It truly astounds me that people have the will to live to argue this crap for 54 (and counting) pages.

We made this thread so people can discuss the topic without fear of persecution (lol!) and so they don't clog up the forum with this, as long as they keep it in here they can knock themselves out, as should be their right on a cricket forum.

 

[Jakester1288]

Warne v Murali threads are terrible, and I have seen them closed on other forums. I don't inderstand how this one has survived.

 

[Top_Cat]

I'm not too sure about the stats above - meaning my knowledge is just ordinary. Maybe Top_Cat can provide some info.

SJS's response about sums up my thoughts on the matter.

Where is the proof of association between number of balls bowled per innings and number of wickets per innings? Testing of hypotheses? Significance/non-significance (p-value) of the linear relationship and for that matter showing a significant binomial relationship betwen the two variables? F-ratio's to show that a non-linear relationship exists and a binomial plot is justified? Why is there an assumption that these are the only two variables which explain what we want to know (I highly doubt they are)? Where's the MHLR with significance values to see how much of the variation is explained by other factors?

The whole issue of tiredness is somewhat difficult to quantify too. It seems pretty clear to me that you can't make any assumptions about whether they were tired when they were removed from the attack in any given instance and we certainly can't quantify their level of fitness/injuries by innings (it probably does vary significantly from match-to-match, tour-to-tour, etc. so how do we know that Warne bowling 30 overs in a day in 1993 = Warne bowling 30 overs in a day in 2005). In terms of 'tiredness', is 4/120 in two 20 overs spells the same as 4/120 in a 10-over spell + 2x15 overs spells? How do we know? Is the time of day where the bulk of bowling done significant? God, there's so many variables to worry about here. And now, having taken all of that into account or normalised the data so you can make the different innings/matches/years comparable, does the number you come up with actually mean anything? And then, how does that figure vary with the other data? Is there a significant association.

Wow, that's a whole load of blarney, isn't it?

 

[pasag]

Wow at P-values, F-ratios and r-squared coefficients. I come to CW to get away from the stuff that's made my life living hell through the semester . What's next, Chi Squared and Anova tests?

 

[Top_Cat]

Wow at P-values, F-ratios and r-squared coefficients. I come to CW to get away from the stuff that's made my life living hell through the semester . What's next, Chi Squared and Anova tests?

Already covered ANOVA's but no chi-square, mate; no categorical variables here.

 

[pasag]

Already covered ANOVA's but no chi-square, mate; no categorical variables here.

 

[Top_Cat]

Gun call.

 

[honestbharani]

Time to argue Casson Vs Mendis, I feel.



Forget the has beens...

 

[vic_orthdox]

OMG, ITS 316-130 QUANTITATIVE METHODS

Am having nightmares aws.

 

[Richard]

We made this thread so people can discuss the topic without fear of persecution (lol!) and so they don't clog up the forum with this, as long as they keep it in here they can knock themselves out, as should be their right on a cricket forum.

Wasn't criticising the thread, I'm massively in favour of it. Wish it'd been started 2 years before it was. I was just commenting that I'm truly amazed people don't lose the will to live trying to argue what I see as such a futile case.

 

[Migara]

His SR is inflated because of Bangladesh and Zimbabwe amongst other things described in this thread.

It was proven beyond doubt that on same conditions, Murali's SR is lower than that of Warne by about 4.

 

[Migara]

How do you measure the 'Tiring' Quotient here ? Is it even possible to measure that ? Also how do you measure 'tiring' factor in situations like ' Bowler bowled total 35 overs in an innings over two days as ooposed to Bowler B bowling 25 overs in one day.'

That was only an idea. Above stats do not look at tiring factor. But I assumed there has to be such a factor even it is minute. And interestingly, the trendlines are quite similar in shape. If I have bit of more time, I'll dig up Kumble's stats as well to see whether there is such a trend.

Secondly, I would be more interested in knowing how both bowlers compare in real life by wickets in no. of overs bowled. I.e. How many wickets Murali/Warne take in their first 5/10/20 etc over.

OK. I'll try to do that.

 

[Migara]

SJS's response about sums up my thoughts on the matter.

Where is the proof of association between number of balls bowled per innings and number of wickets per innings? Testing of hypotheses? Significance/non-significance (p-value) of the linear relationship and for that matter showing a significant binomial relationship betwen the two variables? F-ratio's to show that a non-linear relationship exists and a binomial plot is justified? Why is there an assumption that these are the only two variables which explain what we want to know (I highly doubt they are)? Where's the MHLR with significance values to see how much of the variation is explained by other factors?

The whole issue of tiredness is somewhat difficult to quantify too. It seems pretty clear to me that you can't make any assumptions about whether they were tired when they were removed from the attack in any given instance and we certainly can't quantify their level of fitness/injuries by innings (it probably does vary significantly from match-to-match, tour-to-tour, etc. so how do we know that Warne bowling 30 overs in a day in 1993 = Warne bowling 30 overs in a day in 2005). In terms of 'tiredness', is 4/120 in two 20 overs spells the same as 4/120 in a 10-over spell + 2x15 overs spells? How do we know? Is the time of day where the bulk of bowling done significant? God, there's so many variables to worry about here. And now, having taken all of that into account or normalised the data so you can make the different innings/matches/years comparable, does the number you come up with actually mean anything? And then, how does that figure vary with the other data? Is there a significant association.

Wow, that's a whole load of blarney, isn't it?

What you are telling is exactly correct. All those statistical data is necessary to get in to a better understanding. I was testing a hypothesis put by another person. When I am little free I'll put this in to a SPSS table and find some of those values.

There may be many variables, and some can be unknown. But isn't it what we use linear/polynominal regression for? A low R square value will show that there are many other factors contributing. And infact the R2 of above plots tell the same story. R2 around 0.25 would say that there are other factors involved.

And I repeat that I did not put the factor of "tiring" in to my equation.

 

[Sanz]

OK. I'll try to do that.

Thank You. IMO that would settle the 'Murali's SR is better because he bowls more overs' debate for good.

 

[Top_Cat]

What you are telling is exactly correct. All those statistical data is necessary to get in to a better understanding. I was testing a hypothesis put by another person.

Correlation != causation.

There may be many variables, and some can be unknown. But isn't it what we use linear/polynominal regression for? A low R square value will show that there are many other factors contributing. And infact the R2 of above plots tell the same story. R2 around 0.25 would say that there are other factors involved.

No. The correlation between two variables only speaks to those two and their strength of association. All one of 0.25 tells you is there is a weak to moderate positive linear relationship between those two variables but then, without the p-value it's meaningless anyway. Also, that there's such a wide variation in correlations between just two bowlers who are in the very top echlon of bowlers, who took a similar number of wickets, this surely brings into question whether there is a real association between number of deliveries and wickets and whether it says anything meaningful about their abilities to get batsmen out. For all we know, the rest of the variation in what determines whether a bowler gets a batsman out could have little to do with the bowler.

 

[GIMH]

EPO please

 

[Neil Pickup]

Thought of bumping this.

Some people have been saying that Murali takes more wickets because he bowls more. When you bowl more, oyu can have more opportunities. But you'll be tired and your performance will drop after some time. So thought of testing this out statistically.

I plotted number of wickets against number of deliveries per each inning. On average Warne takes about 2.6 wickets per innings and Murali 3.5 per innings. Warne bowls 149 deliveries per innings whereas Murali bowls 192.

Resulting scatter digrams were as following

The trend lines were drawn using linear egression and polynominal regression analysis. As shown by the R square value, polynominal thrend line describes the scatter better.

This shows that Murali will achieve his highest potential (i.e 4 wickets per innings) if he bowled around 300 deliveries. Warne also has the maximum potential at 4 wickets per innigs. but it would take about 350 deliveries for that.

IF Warne was to bowl 192 deliveries on average per innigs, still he would have picked around 3.0 wickets per innigs. If Murali was to bowl 149 deliveries, he would have picked about 3.1 wickets per innigs.

Conclusion: Long bowls would have helped Warne to achieve better figures. The trend is less apparent with Murali. But for about up to 350 deliveries per innigs, Murali will pick more wickets than Warne.

Those R squared values are so low that they are no evidence of anything causal - or even significantly correlative - whatsoever.