Wisden Ranks All-Time Best

[Gotchya]

I promise i'll go through all that and try to make sense(). I know my stats but thats a bit greek to me


And T_C after posting all that and knowing that you made everyone feel like a jack*** why dont you post those discrepencies to WIsden eh ?

 

[royGilchrist]

Sounds good TC, is it possible to do a quick list with maybe 10 different types of players and see how it works out.

For example, use Walsh, Andy Roberts, Abdul Qadir, Ambrose, Marshall, Sydney Barnes, Srinath, Botham, Warne, Harbhajan Singh.

Lets see what your list produces, and if it is in agreement with what our general perception is about these players.

Also, in my opinion we need to ask ourselves why we think one player is better than the other. For exmaple you said Andy Roberts should be higher than Walsh. I agree with you, but can u list the reasons why you think so. And then maybe we can translate these reasons into statistical categories.

So for exmaple Walsh had more longevity but Roberts' peak was better than Walsh's peak, so maybe to incorporate this we can only look at the best 30 tests (or best 50 tests) for both. Infact, it will be better if most weight is given to the best 10 tests, then less weight to the next ten and so on.

There is obvioulsy the age old debate of whether being part of strong attack helps or hurts a bowler? The hadlee/WI bowlers' phenomena. Would hadlee have fared better if he was one of the WI players, or would Holding/Roberts have done better if they were in NZ instead of Hadlee? I think it all depends on what you look at. The average will be better if you are part of a strong lineup but you will obviously end up taking less wickets.

 

[Top_Cat]

Okay, a few posts to respond to:

PS: Absolute zero is -273.15 degrees celcius, to be precise...

Hey, you're talking to a chemist here (don't ask how I ended up in stats) so don't push me, boy.

But how would you use those methods?

A quick example:

You could come up with a null hypothesis like "Andy Roberts bowled better in a team where there were strong players around him" and try to reject that hypothesis in a t-test. The rejection of the null hypothesis doesn't directly prove that he bowled better in a team with weak players around him but just rejects the notion that he bowled better in a stronger team, for exmaple.

Like anything, statistical analysis is a tool and can be used to misrepresent an argument so you have to be careful in talking about what your results actually prove.

And T_C after posting all that and knowing that you made everyone feel like a jack*** why dont you post those discrepencies to WIsden eh ?

Hey, if I did a full-on analysis of the list, I'd want some bloomin' money for it!

Sounds good TC, is it possible to do a quick list with maybe 10 different types of players and see how it works out.

For example, use Walsh, Andy Roberts, Abdul Qadir, Ambrose, Marshall, Sydney Barnes, Srinath, Botham, Warne, Harbhajan Singh.[/quote[

That's actually a good choice for a representative sample but to make it more thorough, you might want to pick players from other countries who are in high on the list and pick players who are very low as well as players who were great in a weak team as well as a string team etc. To be honest, I doubt you'd be able to come up with a representative sample with a list of only ten and make it a credible analysis. Yours is a REALLY good attempt but I think the only thing wrong with your list is that it might need to be longer.

Also, in my opinion we need to ask ourselves why we think one player is better than the other. For exmaple you said Andy Roberts should be higher than Walsh. I agree with you, but can u list the reasons why you think so. And then maybe we can translate these reasons into statistical categories.

This maybe where the problem is. I know Andy had more variation but to me, his greater ability was more about how talented he 'looked' rather than anything else. He just looked like a more talented bowler. He was more accurate too but you're right, criteria needs to be invented for what makes a good fast bowler, which is where the REAL arguments start.

So for exmaple Walsh had more longevity but Roberts' peak was better than Walsh's peak, so maybe to incorporate this we can only look at the best 30 tests (or best 50 tests) for both. Infact, it will be better if most weight is given to the best 10 tests, then less weight to the next ten and so on.

That's where your multi-variate analysis comes into it. Also if you pick the best Tests for the bowlers 'statistically', you start getting into the murky area of basing a bowler's stand purely on performance. And what constitutes a great Test for those players? Best average per wicket? Best strike-rate? Number of wickets? Again, this criteria would have to be defined and declared before the Test so as to ensure that some basic assumptions are clear.

Still, picking the best 20 Tests might be the way to go because even though it's performance-based, it's certainly probable that in those 20 Tests, at least some of the performances would have resulted because of their overall ability as a bowler and not just blind luck.

There is obvioulsy the age old debate of whether being part of strong attack helps or hurts a bowler? The hadlee/WI bowlers' phenomena. Would hadlee have fared better if he was one of the WI players, or would Holding/Roberts have done better if they were in NZ instead of Hadlee? I think it all depends on what you look at. The average will be better if you are part of a strong lineup but you will obviously end up taking less wickets.

Exactly right which means that you couldn't take one or the other statistical factor as indicative of a great bowler alone. More variables need to be looked at all together, hence multi-variate analysis. This can get tricky in trying to 'prove' something, though. I'll give it my best shot, though!

Thanks for the comments guys. I'll put that on the 'to do' list, right next to the Cricketweb song. Seriously, I'll give it a go in the new year and declare the criteria to you guys so you can tell me I'm an idiot and should include something here and a stat there etc.

 

[marc71178]

You could come up with a null hypothesis like "Andy Roberts bowled better in a team where there were strong players around him" and try to reject that hypothesis in a t-test. The rejection of the null hypothesis doesn't directly prove that he bowled better in a team with weak players around him but just rejects the notion that he bowled better in a stronger team, for exmaple.

But surely deciding whether he was bowling in a strong or weak team is a matter of opinion in a lot of cases?

 

[Mr Mxyzptlk]

Although its hard to argue with a team that had such a fine bowling attack. It may be opinion-based but surely alot of people would consider that West Indies team rather strong. He still performed creditably. It wasn't his fault that he was playing on the best team in the history of the game.

 

[Slats4ever]

What abouut Gunaratne
Nissanka can't be far outta there either