What would a modern player need to beat Bradman?

[Coronis]

4.4 deviation for Bradman. Peak Bradman ofc even higher.

iirc his 40 inning peak was about 114 (post Bodyline-1948)

 

[Migara]

4.4 deviation for Bradman. Peak Bradman ofc even higher.

SD is a bad measure for deviation. The distribution is highly skewed. However the bowling averages are lot less skewed than batting averages. If you really want to have a better idea either you have to log transform and calculate SDs or fit a other distribution (like Poisson) and calculate SD using the variance. Some time back I posted these two histograms to show the differences.

If anyone has a csv / xls file of batting and bowling averages of all batsmen and bowlers I could redo it. Data mining is not my piece of cake TBH.

 

[Spark]

On the Ranking of Test Match Batsmen

Summary. Ranking sportsmen whose careers took place in different eras is often a contentious issue and the topic of much debate. We focus on cricket and ex

academic.oup.com

This is the only recent work I'm aware of which has attempted to do this properly - not assuming an analytically closed/known distribution and actually generating the posterior distribution using an MCMC scheme. The "most natural" distribution is obviously the geometric, which is why we use batting average, but it's equally obvious that using a geometric distribution naively is a pretty stupid idea - there is no single "player batting skill" or "bowling skill" number whereby a cosmic coin is flipped with some probability with each run scored until a dismissal.

You can see that the variance in these distributions is really massive:

Which is a big part of why I have really gone off the boil in judging x player as better than y player because they averaged 4 or 5 runs more or whatever, but even then Bradman is way, way out in front. If he was a particle then CERN would be calling a press conference saying that they'd
made some freak discovery that couldn't conceivably be statistical noise. Fig 6 shows this too.

 

[Migara]

On the Ranking of Test Match Batsmen

Summary. Ranking sportsmen whose careers took place in different eras is often a contentious issue and the topic of much debate. We focus on cricket and ex

academic.oup.com

This is the only recent work I'm aware of which has attempted to do this properly - not assuming an analytically closed/known distribution and actually generating the posterior distribution using an MCMC scheme. The "most natural" distribution is obviously the geometric, which is why we use batting average, but it's equally obvious that using a geometric distribution naively is a pretty stupid idea - there is no single "player batting skill" or "bowling skill" number whereby a cosmic coin is flipped with some probability with each run scored until a dismissal.

You can see that the variance in these distributions is really massive:

Which is a big part of why I have really gone off the boil in judging x player as better than y player because they averaged 4 or 5 runs more or whatever, but even then Bradman is way, way out in front. If he was a particle then CERN would be calling a press conference saying that they'd
made some freak discovery that couldn't conceivably be statistical noise. Fig 6 shows this too.

Actually 14.8, 12.8 and 11.7 standard deviations are not that different from each other. all are extreme values, unlike 2.0 and 4.4. (Variances are compared as ratios via a F distribution). So this explains that GA Headly was not that different from Bradman. Actually he was black Bradman.

Now if you do the same for bowling averages, it will actually show that it has a different distribution. Most likely averaging 19 with ball would be very much closer to averaging 90 with bat due to the extreme long tail in bowling averages. (zero inflation of batting averages vs infinity deflation in bowling averages)

 

[capt_Luffy]

Now if you do the same for bowling averages, it will actually show that it has a different distribution. Most likely averaging 19 with ball would be very much closer to averaging 90 with bat due to the extreme long tail in bowling averages. (zero inflation of batting averages vs infinity deflation in bowling averages)

More than a few bowlers like Sydney Barnes and Frank Tyson do averages under 19 with the ball..... No player who has played 2 Test matches averages almost 30 runs within Bradman's range. Simply put, a balling Bradman will average sub 15 atleast, to separate himself enough from his peers like the Don does. (Even then, George Lohman averages just above 10).

 

[ma1978]

I think people are missing how unique Bradman is. He is the ultimate statistical freak. And as you get further along the bell curve, he is multiple standard deviations better than the next person. I’m sorry Imran and Sobers don’t come close to the man. On a relative basis he is the most valuable athlete to have played any sport at any time in modern human history. And it’s not that close.

 

[capt_Luffy]

I think people are missing how unique Bradman is. He is the ultimate statistical freak. And as you get further along the bell curve, he is multiple standard deviations better than the next person. I’m sorry Imran and Sobers don’t come close to the man. On a relative basis he is the most valuable athlete to have played any sport at any time in modern human history. And it’s not that close.

Exactly my sentiment. Imran and Sobers are Great cricketers, but not peerless. Sobers' skills are definitely comparable to that of Jacques Kallis and to some extent Wally Hammond; while for Imran, the scope is even greater with comparison with Richard Hadlee, Keith Miller, Ian Botham and to an extent Malcolm Marshall and Wasim Akram. Don't get me wrong, they're still head and shoulders above their peers, but not like Bradman. That man averaged almost twice of that of Viv Richards..... Let that sink in. Sobers and Imran, no matter how great they're in one skill, are not Bradman's equal in that, and I don't think them being very average on the other is cutting the mark in anyway.

 

[ataraxia]

Just Zthing it over combined disciplines and let fielding be a tiebreaker when close.

I've read that Bradman was about 4 standard deviations better than average. Pulling figures out of my ass, peak Imran as a bowler might be 2-2.5 and as a batsman .5-1 so combined he is 3.5. still doesn't make it.

Does anyone care to get the real numbers?

Yeah this absolutely demonstrates why Bradman was an utter freak of a batter (which tends to get forgotten even here) but it isn't the best measure of cricketing impact. Using this sort of analysis you'd end up rating Marshall as, what, -1 standard deviation as a bat so he actually gets punished for his batting while Bradman doesn't get punished for his bowling, even though Marshall's batting >> Bradman's bowling.

 

[subshakerz]

Exactly my sentiment. Imran and Sobers are Great cricketers, but not peerless. Sobers' skills are definitely comparable to that of Jacques Kallis and to some extent Wally Hammond; while for Imran, the scope is even greater with comparison with Richard Hadlee, Keith Miller, Ian Botham and to an extent Malcolm Marshall and Wasim Akram. Don't get me wrong, they're still head and shoulders above their peers, but not like Bradman. That man averaged almost twice of that of Viv Richards..... Let that sink in. Sobers and Imran, no matter how great they're in one skill, are not Bradman's equal in that, and I don't think them being very average on the other is cutting the mark in anyway.

Yes. Bradman was two worldclass cricketers in one and no cricketer has approached that ever really.

Sobers and Imran are like 70% of Bradman IMO at best.

 

[TheJediBrah]

I think people are missing how unique Bradman is.

I don't think people are missing it, they're in denial. Which is understandable. If something is hard to believe it's natural to twist reality or perception in pursuit of an explanation

 

[capt_Luffy]

I don't think people are missing it, they're in denial. Which is understandable. If something is hard to believe it's natural to twist reality or perception in pursuit of an explanation

Exactly my thoughts. Like when I was younger (like 10) and used to day Dream about playing for my country, winning world cups, being highest run getter and wicket taker, retiring with 20000 runs, etc etc; even in my dreams I never imagined averaging 100, considering Sachin was averaging 50ish. It was even beyond my fantasy, and I think I am not alone.

 

[ataraxia]

I don't think people are missing it, they're in denial. Which is understandable. If something is hard to believe it's natural to twist reality or perception in pursuit of an explanation

He's unique as a batter. He's the greatest sportsperson of all time in excelling above the mean in one way.

But the influence of a single batter in cricket is overrated IMO. How often do we see teams losing matches despite their perceived strength when their fortunes rest on a couple of star batters (RCB, 00s WI)? How often does the same happen when the star is an all-rounder or bowler?

There are 8 batters-equivalent and 4.5 to 5 bowlers-equivalent in a test team. Individual bowlers are a lot more important than individual batters.

 

[RossTaylorsBox]

Actually 14.8, 12.8 and 11.7 standard deviations are not that different from each other. all are extreme values, unlike 2.0 and 4.4. (Variances are compared as ratios via a F distribution). So this explains that GA Headly was not that different from Bradman. Actually he was black Bradman

Those standard deviation values don't show that at all.

 

[Victor Ian]

I want them to do Bavuma using that method Spark listed. His posterior distribution would be huge!

 

[Victor Ian]

Yeah this absolutely demonstrates why Bradman was an utter freak of a batter (which tends to get forgotten even here) but it isn't the best measure of cricketing impact. Using this sort of analysis you'd end up rating Marshall as, what, -1 standard deviation as a bat so he actually gets punished for his batting while Bradman doesn't get punished for his bowling, even though Marshall's batting >> Bradman's bowling.

though i didnt say it, anything less than 0 in the se ondary discipline could just be taken as zero. you are not an all rounder.

anyways i like sparks link because it shows that Waugh is more than Tendulkar which I always knew deep in my heart.

 

[Migara]

More than a few bowlers like Sydney Barnes and Frank Tyson do averages under 19 with the ball..... No player who has played 2 Test matches averages almost 30 runs within Bradman's range. Simply put, a balling Bradman will average sub 15 atleast, to separate himself enough from his peers like the Don does. (Even then, George Lohman averages just above 10).

Barnes and Tyson played in an era batting average was much lower than the present. Infact Barnes' adjusted average was around 20-21. Still a legend, but with the pack, rather than away from it.

If there is time I would model the averages with number of opposition players played against to see whether variability reduces when facing greater number of opponents.

 

[Migara]

I think people are missing how unique Bradman is. He is the ultimate statistical freak. And as you get further along the bell curve, he is multiple standard deviations better than the next person. I’m sorry Imran and Sobers don’t come close to the man. On a relative basis he is the most valuable athlete to have played any sport at any time in modern human history. And it’s not that close.

There is no bell curve in batting averages. At least understand that. Once a tail is so long, difference between outliers become very less.

 

[Migara]

Those standard deviation values don't show that at all.

You must be joking.

 

[Migara]

Yes. Bradman was two worldclass cricketers in one and no cricketer has approached that ever really.

Sobers and Imran are like 70% of Bradman IMO at best.

Now this is the claim we have trouble with. Is he twice better as a 50 averaging batsman, say Tendulkar or Lara. Very difficult to say because there is no particular way to compare the quality of cricket they played, or the effect of increasing number of oppositions and conditions to play in. At best this is a subjective analysis.

May be some one use AI to model for all these variables and can come up with an answer.

 

[RossTaylorsBox]

You must be joking.

I'm not, maybe you can tell me how those standard deviations prove Headley was similar to Bradman.