If Sir Donald Bradman was born an Indian

[mishra]

Don't know what you are arriving at, but don't think if speed, that is speed of running, has improved by 25% or so. To get a better answer, you should ask more qualified question. 1.5 billion is the exact reason why there can not be any more Bradman.

Ver well said

[bulbul]

Well if Bradaman was indian firstly we would not now about thim' date=' as [b']history would have forgotten him. Also he would simply be to me a historical relic from the past, who is fun to talk about but certainlky cant be compared to people in modern world. Also would npt have seen him play etc so would not have the feel for him. Of course the self loathers would hate him and smash his rep to all places while praising the gora tendulkar:--D

so after 20/25 years you will also forget Sachin ... seems like what outsider saying all these times is true after all

[mishra]

OP is epic :hysterical:

[bulbul]

OP is epic :hysterical:

Not in the same league as you :hysterical: :giggle: :hysterical:

[akshayxyz]

Some people can't appreciate a well written satire... @OP - thanks for the good read. :two_thumbs_up:

[bulbul]

Some people can't appreciate a well written satire... @OP - thanks for the good read. :two_thumbs_up:

this has to applicable evrywhere we cant just select whats suit us :winky:

[akshayxyz]

Having a huge advantage over second best in terms of statistical measure doesn't necessarily mean that difference in quality was also of same level. This kind of gap may be result of some kind of statistical anomaly which is more likely to arise if field under consideration lacks in competitiveness or data sample is not really huge. I am not telling that Bradman is best ever or not, but highlighting there may be fallacies in this gap with gap with second best argument.

++ I don't understand why is it difficult for stats experts to understand this simple concept.

Which rearest rivals I have chosen to ignore?? I had set a random qualification mark of 1500 runs, which I think is fair enough. What opposing teams you want? I can include stats of Bangladesh batsmen as well to make it the same as the number of active cricket nations in the time of Don Bradman. (Cricinfo is blocked in my office, so can't include stats here,however) Actually you are missing the whole point because of your pre-conceived notions. I am not trying to conclude Bradman was inferior to any of current batsmen or if he was definite case of statistical anamoly. I am only questioning that argument of "gap over second best" can not be taken as gospel truth, particularly when sample size is so small. Just two teams playing each other for years with each having 5-6 batsmen. I am not very sure how much sense comparison, of statistical parameters for this small sample size, would make.

++ again.

[mishra]

Not in the same league as you :hysterical: :giggle: :hysterical:

Thanks Bro..... Lehman Brothers used to sell Black boxes at very high prices untill someone recognised that under the packaging they were all sub prime mortgages....

[CSK Fan]

OP is epic :hysterical:

OP is always epic

[Sachin=GOD]

@OP :nice:

[The Outsider]

++ I don't understand why is it difficult for stats experts to understand this simple concept.

1.5 billion is the exact reason why there can not be any more Bradman.

Without counting myself as a "stats expert", such an exercise is a a simple and quantifiable one in statistics. A fairly straightforward calculation is enough to show whether value "n" in a data set of "a" samples is equivalent to value "m" in a data set of "b" samples and this point has already been made by me in previous threads. From the top of my head, given the larger statistical pool the sample is drawn from today the equivalent number of 100 would be around 75 or so. Above has nothing to do with cricket or subjectivity, but is a simple statistical exercise in identifying the degree to which your data point is an outlier.

[Bumper]

If Bradman were an Indian, some Fingleton/Bingleton's account of his exploits would have counted for a whole lot of nothing. Cricket historians of those years were predominantly white (from: Australia and England) and that meant they told us what we should believe and what we should not. The whole Bradman story comes down to this one simple fact. Looking beyond this and arguing over competitiveness of cricket, conditions, relative greatness of players is simply a waste of time and is mostly an attempt at selling one's conjecture. We don't have any data or reference point to judge Bradman's relative greatness. As for my opinion, Bradman was a great of his own era, but that era wasn't great enough for me. I'd stop there!

[The Outsider]

Here is a back of the envelope calculation that can be done to identify what would constitute an outlier in different sample sizes. Let's take the DFFITS method of identifying statistical outliers for it's simplistic scaling. It scales as 1/sqrt(n). Say, in Bradman's time there were 20 international class batsmen(5 teams, but allowing for the argument that 3 were "minnows" so choosing just 3 batsmen each from them - the likes of Headley and Hazare played for the "minnows" - and today there are 50(being generous here and counting the likes of Dhoni :winky:). The 1/sqrt(n) scaling can give a simple estimate of 0.65 as the ratio then and now of a statistical outlier. Applying that to Bradman's average of 100 it would be an average of 65 today. This is when I've weighed the calculation heavily in today's favor. More realistically, there are only 30-40 international class batmen today if one looks at the batting line ups of Pakistan, West Indies, Zimbabwe, Bangladesh, and New Zealand with a bit more scrutiny. Taking the number to be 40, the ratio goes to 0.71 or implying an average of 71 for someone to be a statistical outlier like Bradman. For 30, it is around 81. Pick whichever one you like.

[Sachin=GOD]

Here is a back of the envelope calculation that can be done to identify what would constitute an outlier in different sample sizes. Let's take the DFFITS method of identifying statistical outliers for it's simplistic scaling. It scales as 1/sqrt(n). Say, in Bradman's time there were 20 international class batsmen(5 teams, but allowing for the argument that 3 were "minnows" so choosing just 3 batsmen each from them - the likes of Headley and Hazare played for the "minnows" - and today there are 50(being generous here and counting the likes of Dhoni :winky:). The 1/sqrt(n) scaling can give a simple estimate of 0.65 as the ratio then and now of a statistical outlier. Applying that to Bradman's average of 100 it would be an average of 65 today. This is when I've weighed the calculation heavily in today's favor. More realistically, there are only 30-40 international class batmen today if one looks at the batting line ups of Pakistan, West Indies, Zimbabwe, Bangladesh, and New Zealand with a bit more scrutiny. Taking the number to be 40, the ratio goes to 0.71 or implying an average of 71 for someone to be a statistical outlier like Bradman. For 30, it is around 81. Pick whichever one you like.

Wah!! dhanya hain aap :adore:

[bulbul]

Thanks Bro..... Lehman Brothers used to sell Black boxes at very high prices untill someone recognised that under the packaging they were all sub prime mortgages....

ok bra.. Now the main concern is Euro default..dont worry about mortgages unless u are exposed to it

[Raghav_12]

Here is a back of the envelope calculation that can be done to identify what would constitute an outlier in different sample sizes. Let's take the DFFITS method of identifying statistical outliers for it's simplistic scaling. It scales as 1/sqrt(n). Say, in Bradman's time there were 20 international class batsmen(5 teams, but allowing for the argument that 3 were "minnows" so choosing just 3 batsmen each from them - the likes of Headley and Hazare played for the "minnows" - and today there are 50(being generous here and counting the likes of Dhoni :winky:). The 1/sqrt(n) scaling can give a simple estimate of 0.65 as the ratio then and now of a statistical outlier. Applying that to Bradman's average of 100 it would be an average of 65 today. This is when I've weighed the calculation heavily in today's favor. More realistically, there are only 30-40 international class batmen today if one looks at the batting line ups of Pakistan, West Indies, Zimbabwe, Bangladesh, and New Zealand with a bit more scrutiny. Taking the number to be 40, the ratio goes to 0.71 or implying an average of 71 for someone to be a statistical outlier like Bradman. For 30, it is around 81. Pick whichever one you like.

Full Marks for trying :hatsoff: but 1. You have included three batsmen from minnows like India and with example of Hazare being one of them. Would like to know what two other Indian batsman you would have selected from India. 2. There are many flaws in your statistical approach. I knew this approach but how you translate ratio of square root of sample sizes in ratio of averages is not very clear in your post, though I am not sure that it was the valid thing to do. 3. Don't limit your sample size just to the players who played international cricket. International players are just the representative of their countries. You have to take all cricket playing population in consideration. If you see, you have taken three data points from India for Bradman's time and 5 for Dhoni's time. This means that population of India which plays cricket has grown only by 40% from 30's to 40's. If you had done your analysis by taking 100 million and 1.5 billion numbers then it would have been closer to actual results. 4. Method of comparing sample size like this would be valid if they follow same probability distribution function (PDF). I don't think looking in increase of sample size of this order and with the spread of cricket to many parts of world - outside England and Australia, you can put them in same PDF.

[zen]

Appears as if at any given time, the cream of batting in that time avg around 50-60 .... it is as if for: - X standard of bowling and fielding ----> cream of batting avg 50-60 - Y standard of bowling and fielding ----> cream of batting avg 50-60 - Z standard of bowling and fielding ----> cream of batting avg 50-60 considering 50-60 as the base for the cream of batmanship, the outlier(s), imo, would be someone avg around 70-75 mark (avg of 100 is just phenomenal)

[The Outsider]

Raghav, ^^ It was only a rough estimate - someone has already done a detailed statistical study on it, published it and concluded an average of around 70-75. I have posted the link to it earlier on ICF as well. Not that it matters because the next time there is a discussion on this, the same people will parrot the same thing without bothering to read and understand what has already been addressed. But anyhow, as for batsmen from the top of my head I can recall Hazare, Merchant, Modi, Mushtaq Ali, Mankad from India. Nourse from South Africa, Headley from West Indies - even the 3 Ws might have had some overlap with Bradman or were were pretty close to his time anyway. Regarding choice of sample, the sample is already selected - the posed problem is that what is the chance of having a statistical outlier amongst international batsmen. If you are given a statistical problem to study the performance of managers in some company you don't go around including the population who ever applied for a MBA in your study. If you are given a problem to study the performance of students at Harvard you don't go around including everyone who took the SAT. The mean of the batting average has remained roughly the same from Bradman's time to today. If you plot a histogram of the number of batsmen and their batting average, you should get a similar distribution. It's also a fallacy that merely increasing the population pool will increase the quality - if that was the case how did the tiny island of Barbados produce more great cricketers than the entire Indian population? EDIT : Even from the modern era guys like Michael Jordan and Janahgir Khan have been much above their peers, so it's a fallacy that it's impossible to dominate a sport by a clear distance in modern times.

[Raghav_12]

Raghav, ^^ It was only a rough estimate - someone has already done a detailed statistical study on it, published it and concluded an average of around 70-75. I have posted the link to it earlier on ICF as well. Not that it matters because the next time there is a discussion on this, the same people will parrot the same thing without bothering to read and understand what has already been addressed. But anyhow, as for batsmen from the top of my head I can recall Hazare, Merchant, Modi, Mushtaq Ali, Mankad from India. Nourse from South Africa, Headley from West Indies - even the 3 Ws might have had some overlap with Bradman or were were pretty close to his time anyway. Regarding choice of sample, the sample is already selected - the posed problem is that what is the chance of having a statistical outlier amongst international batsmen. If you are given a statistical problem to study the performance of managers in some company you don't go around including the population who ever applied for a MBA in your study. If you are given a problem to study the performance of students at Harvard you don't go around including everyone who took the SAT. The mean of the batting average has remained roughly the same from Bradman's time to today. If you plot a histogram of the number of batsmen and their batting average, you should get a similar distribution. It's also a fallacy that merely increasing the population pool will increase the quality - if that was the case how did the tiny island of Barbados produce more great cricketers than the entire Indian population? EDIT : Even from the modern era guys like Michael Jordan and Janahgir Khan have been much above their peers, so it's aan fallacy that it's impossible to dominate a sport by a clear distance in modern times.

First and foremost thing the sample formula of square root n that you applied here is absurd in this context, sorry if I am sounding offensive. Division by square root of n is used to come up with the standard deviation of population mewhere you have calculated standard deviation using a limited sample sizes. http://en.wikipedia.org/wiki/Standard_deviation See section 6. How you have translated that into extrapolation of one data point from one sample to other sample is beyond my comprehension and also didn't care to explain, though I had asked for that in my previous post. I would like to see exactly how you have converted 99 in 65. Secondly there is a point of probability distribution which particular sample follows. The argument which completely fails your theory is that assume there were 15 international teams today and that mean you would have taken current sample size as 75. Now repeat your calculations and let me know what 99 now translates to. I am not sure, but would come somewhere around 40-50. So it concludes, if few more countries were given test status then Bradman would have been a lesser bat than Sachin. But, in current situation he isn't. Regarding your Harvard related example, you always consider all people who were aspiring to become part of that sample. All India JEE topper is not celebrated because he got 1st rank among 2-3 K selected, but celebrated because he got top rank from a pool of 200-300k aspirants. You don't say Usain Bolt was fastest among the 8 who ran the final, but he is considered fastest among all human beings who ever ran, not withstanding whether they represented country or not. Anyways, I had used that example to highlight that with larger population aspiring to become cricketers, competitiveness has increased multi-fold. You can not just it has increased from 20 to 50.

The mean of the batting average has remained roughly the same from Bradman's time to today. If you plot a histogram of the number of batsmen and their batting average, you should get a similar distribution.

Refer to the earlier link I mentioned in this post. In case of n being very small standard deviation will of mean would be high and hence reliability will be low. This is too when we agree that both the samples followed same distribution. Your other argument that population doesn't matter as you can see with example of WI, doesn't warrant a serious discussion. Population is not the only parameter which determines quality. If you gather 1000 monkeys and one human being, still you can't expect that atleast one monkey would be more intelligent than single human being - just because monkey's are coming from a sample size of 1000. I'll say again here I am not trying to prove that Sachin is better than Bradman. I am just saying that using the argument of difference from second best, you can't conclude that Bradman was definitely better than any of modern greats be it Sachin or Lara or Ponting or Dravid. If you ask my view on this debate I wouls say that you just can't compare them, they played different games where many rules happened to be the same.

[The Outsider]

First and foremost thing the sample formula of square root n that you applied here is absurd in this context, sorry if I am sounding offensive. Division by square root of n is used to come up with the standard deviation of population mewhere you have calculated standard deviation using a limited sample sizes. http://en.wikipedia.org/wiki/Standard_deviation See section 6. How you have translated that into extrapolation of one data point from one sample to other sample is beyond my comprehension and also didn't care to explain, though I had asked for that in my previous post. I would like to see exactly how you have converted 99 in 65. Secondly there is a point of probability distribution which particular sample follows.

I am using a method to identify statistical outliers, called DFFITS. You can read more about it to see how the 1/sqrt(n) term comes about in it. It's a commonly used and powerful technique to identify statistical outliers. How I calculated the numbers? DFFITS says a data point can be identified as a statistical outlier depending on the value of some constant, call it K divided by sqrt(n). The point does not have to be an outlier, but the above can be used to identify how far away it is from the sample. Say, there were 20 international batsmen at Bradman's time then his average of 100 will be equal to some constant divided by sqrt(n) : K/sqrt(20) = 100 For a data point to be at the same level in a sample of 50 international batsmen to the extent Bradman was in 20, say it's value has to be x. Then x = K/sqrt(50) = 100*sqrt(2/5) which is roughly 65. The only assumption which goes in here is that the underlying samples have similar distribution. You can plot a histogram of number of batsmen and their averages and check that - I am pretty sure they'll be similar enough. For DFFITS to apply, they don't have to be identical distributions.

The argument which completely fails your theory is that assume there were 15 international teams today and that mean you would have taken current sample size as 75. Now repeat your calculations and let me know what 99 now translates to. I am not sure, but would come somewhere around 40-50. So it concludes, if few more countries were given test status then Bradman would have been a lesser bat than Sachin. But, in current situation he isn't.

15 test teams? Anyhow, the number of test teams is not relevant here. What's relevant is the number of international class batsmen - you can keep a clear cut off for them at some reasonable average like 30-35 or so and as long as you do it the same way for both samples it doesn't matter. There would probably be some adjustments which would need to be done to weigh different cut offs for the two samples because a lot more cricket is played today.

Regarding your Harvard related example, you always consider all people who were aspiring to become part of that sample. All India JEE topper is not celebrated because he got 1st rank among 2-3 K selected, but celebrated because he got top rank from a pool of 200-300k aspirants. You don't say Usain Bolt was fastest among the 8 who ran the final, but he is considered fastest among all human beings who ever ran, not withstanding whether they represented country or not. Anyways, I had used that example to highlight that with larger population aspiring to become cricketers, competitiveness has increased multi-fold. You can not just it has increased from 20 to 50.

My turn to call something absurd? The point I made was not about the difficulty of getting into Harvard, but about the performance once the students are in Harvard. If you want to do a statistical study of students' performance while in Harvard (batsmen playing test cricket) will you include everyone who took their SAT (batsmen playing club cricket). To make the analogy clearer. Let's say Harvard admitted 1500 students 40 years back and someone scored a 95 on some relative scale, which let's say qualifies him for a special award. Today if Harvard admits 2000 students what should the score be to be an outlier of the same magnitude so that Harvard can identify if it should hand out that special award? This is a very well defined statistical problem, which can be solved. Want to give it a try?

Your other argument that population doesn't matter as you can see with example of WI, doesn't warrant a serious discussion. Population is not the only parameter which determines quality. If you gather 1000 monkeys and one human being, still you can't expect that atleast one monkey would be more intelligent than single human being - just because monkey's are coming from a sample size of 1000.

So, a cricket crazy country like India are monkeys that they have not been able to produce even as many great cricketers as the tiny island of Barbados? Well, in that case that's a self defeating argument because then the vast Indian population has made little impact on increasing the competitiveness in cricket. It's not just Barbados or West Indies take Australia if you want - they have a population the size of maybe the greater Delhi area or a bit more? New Zealand has a population the size of Lucknow, maybe? South Africa, Delhi and Bombay put together? On the flip side, Bangladesh has the third largest population among cricketing countries.

I'll say again here I am not trying to prove that Sachin is better than Bradman. I am just saying that using the argument of difference from second best, you can't conclude that Bradman was definitely better than any of modern greats be it Sachin or Lara or Ponting or Dravid. If you ask my view on this debate I wouls say that you just can't compare them, they played different games where many rules happened to be the same.

I am not bothering about the Bradman or Tendulkar argument anymore. My point is that argument that things have become so competitive that it's not possible to dominate peers anymore is wrong on two counts: 1. It is quantifiable. I've given a rough way to quantify it and you can check up the paper I was referring to for a peer reviewed and complete way. 2. Sportsmen like Jahangir Khan and Michael Jordan have shown that it is possible to dominate your peers by a very large degree even in the modern era.