Sachin Tendulkar better than Don Bradman, new study shows

[ankitj]

Ankit what is the cut off on the sufficient size of the sample for CLT to apply.

30 is the bare minimum. But like many other things, it depends.

 

[Hurricane]

Regarding ponting - Ponting will be dropped before his average drops below 50.

And likewise if Bradman's average went south too much over innings 80-180 he would have been dropped too (probably).

For example for him to have his average drop to 86 over his next say 20 innings he would have have to have averaged approx 30 over those innings (a case for being dropped)
Thus you can assume that while it, his average, may not have stayed at 99.94 worst case scenario It is unlikely it would have dropped to 55 or 60 either. Therefore I suggest you can with some safety use a high average like 80 or 90 if you want to be conservative in extrapolating to 180 innings.

Regarding hussey we have already dealt with him his average declined long before inning #80

 

[Ruckus]

think that's slightly beside the point - if you're gonna say that he'll average 95-100 over 180 test innings based on his fc average, that's fine, but it's a whole other thing to say the same thing based on an average of 95-100 over 80 innings.

Well you can't say he would, unequivocally, but all the evidence points towards his test career as not simply being a purple patch but rather an accurate indication of just how good he was. I.e. there is no reason to believe if he had played more innings his average would have dropped by very much. That's all I was trying to say.

 

[ganeshran]

Yeah, but when the difference in averages (i.e. estimated means, let's say) is so large, uncertainty due to smaller sample will do little to challenge the conclusion that Bradman's true average is better than Tendulkar's. It might be harder to do that with respect to Graeme Pollock vis-a-vis Tendulkar may be. Also, 80 is a fairly large sample. There is a diminishing marginal benefit on sample sizes too.

Again, my posts in this thread are not to argue SRT vs Bradman.

I am just interested in the extrapolation of aggregate totals with respect to the number of innings. Beyond what number of innings does a batsmen's average become a statistically sound representation of his average in any number of future innings, so that we can make a accurate estimate of his aggregate if he had played that many number of innings more.

 

[Hurricane]

Again, my posts in this thread are not to argue SRT vs Bradman.

I am just interested in the extrapolation of aggregate totals with respect to the number of innings. Beyond what number of innings does a batsmen's average become a statistically sound representation of his average in any number of future innings, so that we can make a accurate estimate of his aggregate if he had played that many number of innings more.

Well the thread is on Bradman vs SRT so that is the example people want to use.

 

[The Sean]

A variation on something I’ve posted previously, with regard to extrapolation or a greater sample size.

If Bradman had followed his famous final innings duck with 54 more – and so finished his Test career with 55 consecutive ducks – his Test average would still be higher than Tendulkar's.

Just saying.

 

[Spark]

Well you can't say he would, unequivocally, but all the evidence points towards his test career as not simply being a purple patch but rather an accurate indication of just how good he was. I.e. there is no reason to believe if he had played more innings his average would have dropped by very much. That's all I was trying to say.

but then the extrapolation ceases to be the major part of your argument - so there's little point in doing it regardless.

 

[Ruckus]

I'm not sure it was ever part of my argument

 

[ganeshran]

Haha ruckus was only joking on the direct extrapolation part. But its an interesting discussion - the impact of longevity on a player's numbers.

 

[Spark]

I'm not sure it was ever part of my argument

this is what i get for skim reading posts.

i blame sachin

 

[Top_Cat]

****ing yawn. Economist in 'slavish adherence to the use of numeric data without bothering to consider the limits of said data' shocker.

 

[ganeshran]

30 is the bare minimum. But like many other things, it depends.

So if we apply this to historical data, i.e comparing batsmen's averages at 30 innings and at the end of their careers. what would be the acceptable error percentage in the predicted values vis-a-vis their actual aggregates.

 

[Hurricane]

I don't think you can use 30. 30 is for randomly selected data elements.

we are talking about X innings from the start of someones career. 30 innings is when they are either inexperienced or too naieve to know how to fail (fancy way of saying beginners luck). I base this statement that 30 is too low on Hussey in particular.

 

[NasserFan207]

Murali, Warne, Marshall, Mcgrath >>>>> both

 

[ganeshran]

What would be the right number then? And how many innings in the future can we accurately predict based on that

 

[Jono]

Lol @ this coming from The Australian

To the bloke saying Sachin only has so many runs because he has played so many matches... well ****ing duh! The point is he's been good for so long.

 

[The Sean]

I suppose the more pertinent issue here is how some hack - who calls himself an academic - from the Gold Coast campus of Griffith University has managed to get this level of publicity for something so academically unsound?

 

[The Sean]

Lol @ this coming from The Australian

To the bloke saying Sachin only has so many runs because he has played so many matches... well ****ing duh! The point is he's been good for so long.

Ha ha, yeah - I find the Bradman comparisons as odious as anyone, but I don't see how the fact that Tendulkar has been good enough to play international cricket for 22 years can be used as an argument against him.

 

[Arachnodouche]

I'm as sick of Bradman veneration as of Sachin fanboys. The guy played in two countries against four countries for ****sakes, in a vastly different era with vastly differing levels of fitness, competitiveness, remuneration which directly ties in with professionalism. Even granting that you can only eat the gruel on your plate, there is just no way one can dismiss all other claims to greatness with such flippancy as is wont among the Don's worshippers.

 

[Hurricane]

What would be the right number then? And how many innings in the future can we accurately predict based on that

How many has trott played?