How would you pick the stocks that are less highly correlated? High correlation refers to when a stock becomes more correlated with the index in a falling market.
Example:
Stock ABC has a correlation of 0.4, and stock DCE has also 0.4 correlation coefficient with index X.
The market fall more than 10% or so in a given period of time and here the correlation of ABC is 0.6, and of DCE 0.86
Indeed ABC is more desirable than DCE as it protects better the portfolio.
Anyone knows about a model that allows us to pick the right stock in such situations?

Region

=correl(return stream 1, return stream 2)

Right, but remember OP that historical correlation does not necessarily imply a future relationship. Traders have been notoriously burned by the market not behaving as expected.

"There are three ways to make a living in this business: be first, be smarter, or cheat."
Best Response
Sandhurst:
=correl(return stream 1, return stream 2)

Right, but remember OP that historical correlation does not necessarily imply a future relationship. Traders have been notoriously burned by the market not behaving as expected.

Agreed. To further the point, if you are doing one stock vs. another, it's almost entirely useless. If you have a large group of stocks and are minimizing correlation amongst those stocks, it can be useful (assuming you are dealing with stocks that are actively traded). To be fair, even if you assume that the correlations remain stable (which is a ballsy assumption), you need to make sure you take into account the variance as well if you are trying to minimize overall volatility or do something similar.
=correl(return stream 1, return stream 2)

What do you mean by return stream 1 and 2?

Fundamentals build statistics not the other way around.

Generally, companies with stable and more predictable cash flows from operations and/or earnings will be that ABC type of equity.

Example of this are companies from utility industry, AKA defensive stocks. You just can not stop using their services/products, reduce perhaps.

I like to use statistics over CFO and Earnings to define which one from a group of defensive stocks is more stable, but you still need to analyze about companies future and threats to this stream of revenue.

The DCE stock would be more a typical commodity/consume related stock.

• 1
rafaelgirotto:
Fundamentals build statistics not the other way around.

Generally, companies with stable and more predictable cash flows from operations and/or earnings will be that ABC type of equity.

Example of this are companies from utility industry, AKA defensive stocks. You just can not stop using their services/products, reduce perhaps.

I like to use statistics over CFO and Earnings to define which one from a group of defensive stocks is more stable, but you still need to analyze about companies future and threats to this stream of revenue.

The DCE stock would be more a typical commodity/consume related stock.

You're talking statistical models used in conjuction with fundamental analysis, which is a sound method, this way you can avoid spurious correlation between stocks, however the very question I posted in the begining is about those statisticals tools. Basically ABC and DCE were picked from a universe of stocks and were easily screened by just plotting their returns against the main index and just using the Correl function provided by any spreadsheet to get the correlation coefficient, simple and straightforward, is there a similar way for finding the stocks with less volatile correlation?

Well this is just an idea but if you are only using returns and their statistical measures(probably to create an algorithm?), i would try to:

1 - calculate stocks average monthly or weekly betas , considering them as population samples (30 samples at least. btw thats why is so difficult to use annual betas which would be better, i think).

2 - rank stocks with low correlation to index and low beta stdev from previous step.