Luckiest and Unluckiest batsmen

[Richard]

Yep, really ironic - given that I've never done anything of the sort.

 

[Richard]

How else will you find average luck when the sample of the whole population is so small.

And most sides play more than 4 or 5 specialist bats.

Yet few pick more than 4 or 5 consistently over a year.
To find a side settled on 6 batsmen who are performing consistently over a year is extremely rare.

 

[marc71178]

Explain why you have ignored half of each team from your sample.

 

[Richard]

Err - I explained that I didn't select anything on teams, simply on batsmen who played for the top sides in the three years 2000, 2001 and 2002.
I simply took a random sample of 15 players, regardless of who they played for.

 

[Camel56]

And a totally pointless sample at that. 15 players out of thousands is not statistically significant for starters and i very much doubt they were chosen randomly.
Did you pick the names out of a hat? NO

 

[Son Of Coco]

About 50.

2000-2002

Pretty darn sure - you get to know reporters and who's likely to accurately describe a chance and who's likely to call everything that touches the fingertip a chance.

My brother's mind doesn't work like that.

Ok, but 15 from 50 is not a large sample of a small population.

I don't think taking 3 years out of a players career sums it up adequately.

I can't believe you're using newspaper articles etc as quantifiable data. If you haven't seen the incident itself then it'd be very hard to say with 100% accuracy that a chance actually occured.

Everybody's mind works like that. You identify first and foremost with things you've heard of - they stand out.

 

[marc71178]

Ok, but 15 from 50 is not a large sample of a small population.

But it's not out of 50, in 3 years I think more than 50 people played Test Cricket!

 

[Richard]

And a totally pointless sample at that. 15 players out of thousands is not statistically significant

Nope, it's not - and given that there aren't remotely close to 1000 specialist-batsmen who've had decent length Test-careers in the last 40 years, you're wrong about that-'un again.

for starters and i very much doubt they were chosen randomly.
Did you pick the names out of a hat? NO

Nope, I got someone else to do a task which has an identical outcome.

 

[Richard]

Ok, but 15 from 50 is not a large sample of a small population.

It's nearly 1\3 - plenty.

I don't think taking 3 years out of a players career sums it up adequately.

It's not ideal, of course, but I don't think it's enormously misleading - what's happened for 3 years will more likely than not have happened for most of the rest of a lengthy career, too.

I can't believe you're using newspaper articles etc as quantifiable data. If you haven't seen the incident itself then it'd be very hard to say with 100% accuracy that a chance actually occured.

Everybody's mind works like that. You identify first and foremost with things you've heard of - they stand out.

I'm pretty good on my reporters - I know who tends to give good impressions of dropped catches and who doesn't. And of course it goes without saying that I've read 2 or 3 reports at least of every innings.
Anyway, for quite a few batsmen I'd seen plenty of their innings so I didn't need to read reports on half the games.

 

[Richard]

But it's not out of 50, in 3 years I think more than 50 people played Test Cricket!

Yet how many of them successfully scored runs?

 

[Son Of Coco]

It's nearly 1\3 - plenty.

It's not ideal, of course, but I don't think it's enormously misleading - what's happened for 3 years will more likely than not have happened for most of the rest of a lengthy career, too.

I'm pretty good on my reporters - I know who tends to give good impressions of dropped catches and who doesn't. And of course it goes without saying that I've read 2 or 3 reports at least of every innings.
Anyway, for quite a few batsmen I'd seen plenty of their innings so I didn't need to read reports on half the games.

Nearly 1/3 of a sample you could reasonably expect to be able to cover completely is not plenty.

Luck in cricket cannot be assumed to be a constant. You cannot reasonably say what happened one year (or three) will mirror what happens for the rest of their career, especially as far as luck is concerned. It's not controlled by anything else they are doing and is purely a random event.

I'm sure you know your reporters, but I've read quite a few articles recently that seemingly watched a completely different game to the one I was watching. If you're going to accurately make claims as far as luck etc is concerned relying on hearsay isn't the best way to do it.

 

[Son Of Coco]

Yet how many of them successfully scored runs?

Again, this is using personal bias to select your sample. I thought you were testing luck, not players who actually scored runs and were lucky. A player could get dropped 4 times on 0, and still score 0. He was lucky 4 times before he eventually ended up with that result though.

 

[Son Of Coco]

Nope, it's not - and given that there aren't remotely close to 1000 specialist-batsmen who've had decent length Test-careers in the last 40 years, you're wrong about that-'un again.

Nope, I got someone else to do a task which has an identical outcome.

It isn't scientifically random though Richard, no matter how you look at it. If you wanted to make it random, one way would have been to assign numbers to the players and then have them drawn out of the hat so that neither of you knew who was being chosen at the time. Thinking back on old stats classes I have a feeling that this still isn't 100% random but I can't think why at the moment.

Randomising something like that is not as simple as you believe it to be.

 

[marc71178]

It's not ideal, of course, but I don't think it's enormously misleading - what's happened for 3 years will more likely than not have happened for most of the rest of a lengthy career, too.

Except when it suits you, you start carving up careers into patterns - can't have it both ways.

 

[marc71178]

Again, this is using personal bias to select your sample. I thought you were testing luck, not players who actually scored runs and were lucky. A player could get dropped 4 times on 0, and still score 0. He was lucky 4 times before he eventually ended up with that result though.

Therefore every player should be tested, not a selection from an already handpicked group.

 

[Richard]

Nearly 1/3 of a sample you could reasonably expect to be able to cover completely is not plenty.

It took long enough as it was - if it'd taken 3 times as long I'd still be doing it!

Luck in cricket cannot be assumed to be a constant. You cannot reasonably say what happened one year (or three) will mirror what happens for the rest of their career, especially as far as luck is concerned. It's not controlled by anything else they are doing and is purely a random event.

Not sure about that.
Luck might be random in where it occurs but like most random things they do tend to form patterns eventually - if you have some amount of luck in one 3-year period you'll probably have about the same amount in another.

I'm sure you know your reporters, but I've read quite a few articles recently that seemingly watched a completely different game to the one I was watching. If you're going to accurately make claims as far as luck etc is concerned relying on hearsay isn't the best way to do it.

Exactly - you need to learn your stuff - once you know how to interpret certain reporters you're laughing.
And that doesn't just apply to those that are still writing.

 

[Richard]

Again, this is using personal bias to select your sample. I thought you were testing luck, not players who actually scored runs and were lucky. A player could get dropped 4 times on 0, and still score 0. He was lucky 4 times before he eventually ended up with that result though.

True.
But the thing is if you get dropped 4 times on 0 and still score 0 your f-c and scorebook score are the same.
I've not discounted those dropped catches just because they don't matter, but nonetheless they don't affect anything.

 

[Richard]

It isn't scientifically random though Richard, no matter how you look at it. If you wanted to make it random, one way would have been to assign numbers to the players and then have them drawn out of the hat so that neither of you knew who was being chosen at the time. Thinking back on old stats classes I have a feeling that this still isn't 100% random but I can't think why at the moment.

Randomising something like that is not as simple as you believe it to be.

Nothing's 100% random - it's almost impossible to achieve that.
The nearest you'll usually get is 80% IIRR (from my own memories of stats lessons) - so I presume that's all I got.
I'm not trying to pretend my survey was faultless in every way - same way anything else isn't.

 

[Richard]

Except when it suits you, you start carving up careers into patterns - can't have it both ways.

If you can see a clear pattern you divide it up.
If you can't, you don't.
Of course there'll be periods where you have more significant luck than in others (Vaughan in 2002, for instance), but if you take more than a handful of players the chances are you'll get some who're having lots of luck and some who're having none, so it'll roughly even itself out.

 

[Richard]

Therefore every player should be tested, not a selection from an already handpicked group.

And if anyone can be bothered to do that, I'd offer them my every encouragement.