Relationships within the stock market

[rynman]

Hi guys

I think this is my first (or 2nd?) post, but hopefully one of many in this community. I've recently decided to start educating myself on stock markets and have been doing a lot of reading.

I've got a statistical background and was wondering if analysts have found any relationships between different sets of data within the stock market? i.e. if one industry is trending down, another tends to improve....things like that. There's a wealth of data there to do historical analysis, and surely people have run models to try and generate something predictive.

Is this a case of people possibly doing exactly this, and keeping it under wraps for their own benefit? Or does this type of thing happen all the time?

Or is it generally know that this is just a waste of time....you can't use the movement or performance of one or many indicators to accurately predict another.

I don't expect a detailed answer, just hoping to generate some discussion.

Thanks guys!

Ryan

 

[Gringotts Bank]

Google "pairs trading".

 

[skc]

The broad term to search for is quantitative methods in trading. Quants (i.e. maths geeks working in Wall Street) use some of the most complicated strategies on the market to observse relationships between everything. Note however these guys have a bit more than just a statistical "background" - they are more like multiple PhDs or quantum physics professors etc. The field is extrememly wide.

But sometimes simple things may work as well - this is not a bad place to dip your toes.

http://quantifiableedges.blogspot.com.au/

 

[sinner]

What you're looking for is factor decomposition, here is an example

http://quantumfinancier.wordpress.com/2011/01/09/predictorfactor-selection/
http://quantumfinancier.wordpress.com/2011/01/29/sp-500-sectors-analysis/