Correlated Stocks

F

FaderTrader

Question regarding trading correlated stocks (this is NOT a pair trading question):

Do you think it's better to watch stocks that are highly correlated to one another or do you think that it is better to watch stocks that have the greatest weighting in the index of which your stock is a part.
 
When talking about correlation, you have to define the time scale you are referring to. Moving intraday correlation? Daily closing price correlation?

Usually things that are weighted together in an index are highly correlated, so I'd say your answer is 'both'.

Quote from FaderTrader:

Question regarding trading correlated stocks (this is NOT a pair trading question):

Do you think it's better to watch stocks that are highly correlated to one another or do you think that it is better to watch stocks that have the greatest weighting in the index of which your stock is a part.
More...
 
Well, I trade SUN, which is in the XOI - which is price-weighted.

Ironically, SUN is more highly correlated to stocks in the OSX. So, if I were to go with a pure correllation strategy - then I'd be looking at a mix of both indexes.

My intuition tells me to first rank all stocks in the XOI end-of-day, then do my regular end-of-day calculations for each stock, and then sort them in descending order according to price (and hence, thier rank in the XOI) So, if I trade SUN, I'd be looking at the top 3 most highly weighted components of the XOI the next day.

Would you reccomend adding another step to this by saying ignore anything that is less than +.60 correlated to my stock?

Thanks.
 
I wouldn't advocate putting any hard cutoff points in your calculations (like 0.60), because by doing that you are introducing bias beforehand with no real theoretical justification, and also no way to indicate to your model the level of uncertainty of your 0.60 value.

I don't have any experience with daily correlation measures so I can't comment on what correlation 'looks' like at this scale.

I'd suggest reading up on arbitrage pricing theory. It won't give you any directly applicable formulas but it will lead you into the direction of understanding certain behaviors. E.g. a large amount of the correlation between stocks is actually common market movement, so by isolating this movement and then looking for the correlations among the stock-specific affects..




Quote from FaderTrader:

Well, I trade SUN, which is in the XOI - which is price-weighted.

Ironically, SUN is more highly correlated to stocks in the OSX. So, if I were to go with a pure correllation strategy - then I'd be looking at a mix of both indexes.

My intuition tells me to first rank all stocks in the XOI end-of-day, then do my regular end-of-day calculations for each stock, and then sort them in descending order according to price (and hence, thier rank in the XOI) So, if I trade SUN, I'd be looking at the top 3 most highly weighted components of the XOI the next day.

Would you reccomend adding another step to this by saying ignore anything that is less than +.60 correlated to my stock?

Thanks.
More...
 
Quote from stephencrowley:

I wouldn't advocate putting any hard cutoff points in your calculations (like 0.60), because by doing that you are introducing bias beforehand with no real theoretical justification, and also no way to indicate to your model the level of uncertainty of your 0.60 value.

I don't have any experience with daily correlation measures so I can't comment on what correlation 'looks' like at this scale.

I'd suggest reading up on arbitrage pricing theory. It won't give you any directly applicable formulas but it will lead you into the direction of understanding certain behaviors. E.g. a large amount of the correlation between stocks is actually common market movement, so by isolating this movement and then looking for the correlations among the stock-specific affects..
More...

Thanks for the feedback. I don't really like to execute strategies based upon specific stock correlations as I don't consider myself to be well-enough versed in the theoretical aspect of it to trust that I'd ask all the necessary questions, but - based on what you're telling me, I'm concluding that following those stocks which are most highly weighted to my respective index is the way to go (since most correlation is due to market movement as you say). At the very least, I feel it makes for a more tolerable risk/reward opportunity.

Regards,
Fader
 
Right, you might also look into taking offseting positions in the index vs. the components you are holding so as to hedge against sector specific risks, although you'd still be exposed to market wide drift.

Quote from FaderTrader:

Thanks for the feedback. I don't really like to execute strategies based upon specific stock correlations as I don't consider myself to be well-enough versed in the theoretical aspect of it to trust that I'd ask all the necessary questions, but - based on what you're telling me, I'm concluding that following those stocks which are most highly weighted to my respective index is the way to go (since most correlation is due to market movement as you say). At the very least, I feel it makes for a more tolerable risk/reward opportunity.

Regards,
Fader
More...
 
i've run a little bit of daily correlation analisys though i dont use it in my trading. you also need to decide whether you're interested in correlation of two price series, or two returns series.

the correlation of returns will tell you whether two series tend to do the same things at the same time, in my experience these tend to be low unless two symbols are directly related, in which case i guess these should be avoided for non pairs trading.

choosing symbols with high correlation of prices in my experience generally results in a list of symbols that are all trending in the same direction.
 
You must log in or register to reply here.
Back
Top