How much can we justifiably use statistics

[flibbertyjibber]

In reply to the thread question, why would you want to do that?

To justify Sobers over Imran.

 

[kyear2]

That was the purpose for another thread. Just thought the stats were interesting and had never seen the ratings broken down like that and thought it was worth sharing.
My bad

 

[OMM!]

Cricket without stats is just rounders basically. And I ****ing hate rounders!

 

[smash84]

Yeah, I'm quite interested in how one would analyse players' performances in matches one did not see, or failing that, watch every match ever.

To assume one's cricketing knowledge is of such grandeur as to outweigh human fallibility of analysis, performance variance and natural bias.. and all while only watching a tiny portion of the cricket played at that.. is the height of arrogance. To profess to know better than the game itself because statistics don't always tell the full story reminds me a lot of the argument that it's better to use the human eye than HawkEye even if the latter is more accurate because it's not 100% accurate. Blind faith in our own flawed instincts whenever a measure, however much better, isn't perfect is a widespread ideal throughout humanity that I will never really grasp.

Anyone who reads match threads here will know that I watch as much current cricket as anyone and love nothing more than to discuss a batsman's technique, a bowler's contribution to the attack as a whole, the weight of pressure a good field-set can mount on a team and all those other "beyond averages" aspects of cricket. This is the part of cricket I find most interesting and while I like to use these judgments to form opinions on the potential of players still playing, once someone retires their potential means nothing and all we're left with is what they actually did. And as much as I'd love for my opinion on Matthew Hayden's head position post 2004 to have any relevance to how good he was at scoring runs throughout his career, it doesn't. Determining how good someone could be, might be or probably will be is an exciting and interesting cricketing topic for me because you can use what you see and form judgments from that, but comparing two players who have since retired though (or even a here-and-now comparison) should really come down to what they've put on the board more than anything else. I have a lot of respect for robelinda's position on the matter that it's just fundamentally boring because the more I delve into it, the more it is. Its boring nature doesn't change what it is, though, and no attempt to make it more interesting and achievable will produce more accurate results.

----

Oh, and ignoring all that and addressing the OP for a second, I'm faaaaar from convinced that ICC rankings are the best statistical measure around, let alone a uniform one that can be applied to each player and all his career intricacies. Good post anyway though.

To justify Sobers over Imran.

 

[sumantra]

not very much i guess...but at the same time can't say, that it's absolutely nothing...specially when it comes to judging a player from the past, to some extent it's important to see what his stats were (i am not saying that is the only criteria, but one of them) besides the stories u hear and the books u read, when u see a headley having a test average of 60 odd or a vijay merchant having a first class average of 70 odd after playing 150 matches, u got to take them with some kind of seriousness...the thing is, how far can u go with it, the problem for me comes, when u r trying to judge 2 players played in the whole different era, and trying to compare them in terms of stats...i can't take such attempts seriously, because u got to look at the other factors, the factors that have changed...not only in psyche or approach but in terms of actual playing condition as well...stats can be to some extent applied to players who played in the same era, faced same bowlers or batsman...but that too, only to some extent...how much can u apply is a subjective judgement...but any effort to wholeheartedly rely on stats is for me, not the right way of going about discussion on cricket...

 

[Sanz]

Still the process has to be explained, and bias is obvious when used in statistics. In subjective assesment, it's not apparent, and worshippers of it, says that they are not biased at all as well.

Says the guy who picked Moin Khan in his all time World XI.

 

[Arachnodouche]

I've never understood comparisons between eras. For all the statistics in the world, you instantly enter the realm of subjectivity as soon as you start conjecturing how Ponting would've faced up to Marshall or Bradman to a Steyn. The only sensible way to use stats as far as I understand is between contemporaries. And even then, you might have anomalies like Sangakkara having an average as good as Sachin or Lara having a lesser record than say Kallis. Kallis a better bat than Lara? Not over my dead body..subjective as hell, but Even amongst contemporaries, there are so many other variables and minutiae that factor in.

 

[Top_Cat]

All of the manipulations, calculations, transpositions and translations in the world won't change the fact that runs scored and wickets taken are very blunt measures of performance in a game of cricket. I once tried building a LMM taking into account country effects, big errors. Cricket stats are a vague guide only.

 

[Redbacks]

On a recent BBC series Prof. De Sauto used a good demonstration of the 'law of the crowd' and it's ability to predict the numerical quantity of an object with very good accuracy.

The example was a jar of jellybeans and he asked roughly 30 (iirc) people in an office to guess how beans in the jar. There was a few reasonable guesses 5000, 3000 etc. but then one person who you would think was blind said 500. Not to be outdone one lady answered 50,000! What would seem such a ridiculous answer and sure to stuff up the estimate.

In the end the averages of the guesses was roughly (no I cant remember exact numbers) 4652.18, and the correct answer was 4626 jellybeans in the jar. Fricken amazing! It would suggest that the outrageous estimates of some people are compensated for by others who will be more cautious in guessing and that this principles infact allows for greater accuracy that possible by one expert alone.

This has certainly made me more in favour of using polls of many people to determine ATG lists in the hope that the overlap of each 'guess' leads to a cancelling out and a pretty decent list at the end of it.

 

[Flem274*]

It all comes down to what you value more when rating a player. objective logic, or subjective opinion.

I am not a stats person eighty percent of the time, but in spite of that when someone asks who was better at what, I know my opinion holds no weight unless I can back it up with evidence.

I love watching the game, and watching Shane Bond or Ross Taylor destroy opposition means more to me than where they rank in world cricket. But at the same time, I can't go around saying Peter Ingram is a better batsman than Mark Richardson because it simply isn't true.

Statistics can and are used in intellectually dishonest ways, but that is the fault of the human being behind it. Their argument holds no more weight than me saying Brendon McCullum is a better bowler than Richard Hadlee. If used in an objective manner with a sound method, statistical analysis works.

When it comes to judging players of similar statistical records however (and I'm taking aim at the all time great debates here), quibbling over small differences seems a bit tedious. International crickets are the top 1% in the bell curve, and the all time greats are a much smaller number than that. The only all time great to be in a different ball park to the rest is Bradman. You could pick any of the top thirty batsmen or bowlers ever and you would have a side similar in quality to a team picked from those you left out.

But the paragraph above does not exonerate those who say "who cares about the stats? The opinions of other people hold more weight." This is false, unless the past greats can show with objective logic the player they think is the best, is in fact the best. "He's the best because I watched him play" is not an argument.

As an aside, taking the opinions of past players as gospel makes me uneasy. It's something people do in other areas of life as well (I'm not going to name anything), and it can have disastrous results. Just because a legend says something, it does not mean his opinion cannot be questioned by someone who has never played international cricket.

 

[smash84]

On a recent BBC series Prof. De Sauto used a good demonstration of the 'law of the crowd' and it's ability to predict the numerical quantity of an object with very good accuracy.

The example was a jar of jellybeans and he asked roughly 30 (iirc) people in an office to guess how beans in the jar. There was a few reasonable guesses 5000, 3000 etc. but then one person who you would think was blind said 500. Not to be outdone one lady answered 50,000! What would seem such a ridiculous answer and sure to stuff up the estimate.

In the end the averages of the guesses was roughly (no I cant remember exact numbers) 4652.18, and the correct answer was 4626 jellybeans in the jar. Fricken amazing! It would suggest that the outrageous estimates of some people are compensated for by others who will be more cautious in guessing and that this principles infact allows for greater accuracy that possible by one expert alone.

This has certainly made me more in favour of using polls of many people to determine ATG lists in the hope that the overlap of each 'guess' leads to a cancelling out and a pretty decent list at the end of it.

Wow.....this is awesome

 

[kiwiviktor81]

The jellybean example is interesting, but it doesn't give much credence to specialist knowledge.

For example, if you're sick, would you value the opinions of twenty random people over that of a doctor?

In much the same way, a panel of professional cricket commentators pick Wasim Akram in an AT Test XI, but I doubt most cricket fans agree.

 

[weldone]

Says the guy who picked Moin Khan in his all time World XI.

Ha ha, did he?

 

[Smudge]

I know my opinion holds no weight

Completely AWTA.

 

[smash84]

Ha ha, did he?

I doubt anyone could be that stupid

 

[sumantra]

On a recent BBC series Prof. De Sauto used a good demonstration of the 'law of the crowd' and it's ability to predict the numerical quantity of an object with very good accuracy.

The example was a jar of jellybeans and he asked roughly 30 (iirc) people in an office to guess how beans in the jar. There was a few reasonable guesses 5000, 3000 etc. but then one person who you would think was blind said 500. Not to be outdone one lady answered 50,000! What would seem such a ridiculous answer and sure to stuff up the estimate.

In the end the averages of the guesses was roughly (no I cant remember exact numbers) 4652.18, and the correct answer was 4626 jellybeans in the jar. Fricken amazing! It would suggest that the outrageous estimates of some people are compensated for by others who will be more cautious in guessing and that this principles infact allows for greater accuracy that possible by one expert alone.

This has certainly made me more in favour of using polls of many people to determine ATG lists in the hope that the overlap of each 'guess' leads to a cancelling out and a pretty decent list at the end of it.

the success of this on polls rely heavily on the variety of samples gathered, age is one such factors among many...

 

[weldone]

the success of this on polls rely heavily on the variety of samples gathered, age is one such factors among many...

Also on

1. Whether one voter can influence decisions of some of the others.

Example: Tommorrow if an 'Imran vs Wasim as bowlers' poll comes up in CW, and SJS is absent the result will be different than if he's present and writes one elaborate post describing the qualities as a bowler that Imran possessed and Wasim lacked.

2. Whether people are particularly knowledgeable about the poll contestants in question.

Example: Tommorrow if a Trumper vs Hayden comes up in CW, there'll be plenty who'll vote for Hayden without knowing/caring who Trumper is/was.

 

[JBMAC]

also on

1. Whether one voter can influence decisions of some of the others.

Example: Tommorrow if an 'imran vs wasim as bowlers' poll comes up in cw, and sjs is absent the result will be different than if he's present and writes one elaborate post describing the qualities as a bowler that imran possessed and wasim lacked.

2. Whether people are particularly knowledgeable about the poll contestants in question.

Example: Tommorrow if a trumper vs hayden comes up in cw, there'll be plenty who'll vote for hayden without knowing/caring who trumper is/was.

awta

 

[Redbacks]

Also on

1. Whether one voter can influence decisions of some of the others.

Example: Tommorrow if an 'Imran vs Wasim as bowlers' poll comes up in CW, and SJS is absent the result will be different than if he's present and writes one elaborate post describing the qualities as a bowler that Imran possessed and Wasim lacked.

2. Whether people are particularly knowledgeable about the poll contestants in question.

Example: Tommorrow if a Trumper vs Hayden comes up in CW, there'll be plenty who'll vote for Hayden without knowing/caring who Trumper is/was.

The CW50 avoided the first problem as the vote was done via e-mail and both of your comparisons fell in the direction you advocate. Imran was ranked higher than Wasim and Trumper made the list whilst Hayden didn't feature. In the case of a cricket poll, people will need a good knowledge of the game and its history to provide a reasonable answer. CW did just that I think.

 

[G.I.Joe]

CW voted Bradman at #8 FFS.