Two securities are highly correlated. Is there a way to profit knowing just this fact?

Depends a little bit. What you could do is check for a first to default quote. Unfortunately this only works if you are an institution... if you are a retail investor you have virtually no chance to profit from that knowledge.

Correlation and changes in correlation is often used to 'observe' market sentiment and changes in market sentiment. An honours student, Manson, a few years back devised a strategy whereby he would invest in three industries that had the highest increase in correlation, and that passed some previous returns criteria. Essentially it is based in Barberis and Schleifer's style investing. I think it had a Shrpe ratio of about 0.29 for a period where the market was about 0.15. So there is some scope.Sometimes the correlation can be represented with a deeper motion of causality, but not always (see Wikipedia article on correlation implies causality (false logic)). There is some evidence that ECNs lead Nasdaq on the same stocks, and some stocks like BHP can lead similar stocks like RIO.All the above however, come with a big conditional!

two securities are very highly correllated until they aren't

Vote for McCain!

thanks that was very useful... your post is most helpful....fyi responding to taglines may not be such a good plan.....

historical correlation are very subjective, okay, lets assume that correlation number you have is correct, but even then correlation does not imply cause-effect relationship., as pointed by joshblak corrrectly.also if only correaltion figure is given and no inputs about individual spot price of 2 stocks, then i don't think, one can use it to make some profit(devise a strategy to profit). correlation is a "symmetric" function of 2 return series and principal of profit is "assymetry". ----------amit7ul

mdlm, cross return predictability is a topic that hasn't shown proven consistency or validity. A lecturer at my university has done some work on it, see <here>. Unfortunately the paper doesn't reference other works (as perhaps there are none?).There may be pockets of this type of inefficiency, but as always it is elusive and will most likely not be proven to be statistically significant through time.

Correlation trading happens all the time. Example: pairs trading, or even convertible arb can be considered correlation trading (long convert and short equity as there is a positive correlation between the two, *usually*). The thing to keep in mind is what Wilmott said in his book: "Correlation is even worse than volatility when it comes to estimating it" and what ex-Goldman and now-Citigroup huncho Robert Rubin said: "In a bad market, all correlations go out the window and supposedly negatively correlated assets all go down at the same time". So do people make a living doing correlation trading? Yes, everyday. Do people lose their shirts doing it? Yes, again everyday.

I'd fully agree with cougar's point about the reversals of correlations. Perhaps the greatest example (as commonly alluded to) is LTCM.Nonetheless, the question of whether to or not to execute on a correlation eventually comes down to balancing risks and your risk-appetite. In the event you face large threats from credit risk but low threats from market risk, a short term position via the use of correlation may be well justified. Nonetheless, poor allocation of recources to cover your hide (capital in the case of credit risk and stop-orders in the case of market risks) will prove highly harmful in the event things don't go your way.

May be a good research topic. Obviously ability to profit will result from a sudden variation in one security's price without an underlying reason. Just wondering if it is a 'variant' of mean reversion.