I have been looking for edges that make logical sense to me and are tolerable psychologically and I believe I have just found something that is very suitable and profitable. I got into pairs trading recently simply because I was looking for something that was market neutral and that didn't require you to decide whether a market was trending or not. Pairs trading seems to fit the bill perfectly. I'm using tradestation to chart correlated pairs (I don't quite understand everything about covariance and standard deviation yet so I just looked online for a list of pairs that are historically correlated, KO vs PEP, MAR vs HOT, etc) and I'm simply using a spread ratio indicator with bollinger bands on it. My entries are when the ratio deviates past 2 standard deviations from the 50 period mean on a 5 min chart and then exiting when it crosses back over the mean. This has tested quite well for me but it's tough to code in EasyLanguage. If there's anyone out there that knows how to code this using Data1 and Data2, hit me up, I would be more than willing to pay for something I can automate. Thanks!
Have you given any thought to betting on greater divergence, rather than convergence of related pairs? One might, for example, buy a good stock (where you find one is another topic, but for starters you could go to IBD's top 100, I guess) and short a poor performing stock (and there are lists of shorting candidates out there as well). That way, you're not betting against the existing tendency of your correlated stocks to diverge. That's what freaks me out about pairs trading. Investopedia uses GM and Ford as the pairs in their example of how to put on a pairs trade in their article of the same name. Well, we all know how that trade would have worked out if you tried it recently: Going long GM and short Ford, banking on convergence between of the two would have backfired, as they haven't and never will converge. Of course, one cuts one's losses, but still, why buck the trend when you can go with it. By going long a winner and short a loser, you're hedged, yet trading with the trend. What do you think?
Have a look at the Pairs Trading Strategy Journal thread by jonnysharpe. Guys use a software called Pairs Trade Finder.
Pairs trading is an approach that is defeatist in nature. You are basically throwing up your hands and saying "I give up--I can't figure markets out". And by the way, you can lose just as much or more with pairs trading. --Ishmael
Thanks to ET posters, yes I've heard about this and I've taken a look at it. I'll definitely do the free trial eventually. Tradestation doesn't have a correlation indicator built-in but Thinkorswim's free platform does, so I might end up just using thinkorswim to find more correlated pairs, I'm just working with 10 now.
I don't like the idea of trading pairs below $5-$10 share so I've avoided the automaker pairs like GM and Ford. I also don't like the idea of betting on further divergences. Aren't you just tossing away the whole idea that they are correlated if you do that? Mean reversion makes more sense to me, if you're betting on diversion, you're just saying that one company is stronger than the other.
Excellent post... many folks have lost their "shirts" bettiing on reversion to mean between two pairs, when in fact it's better to simply long the strong stock and short the weak stock as a hedge. Many employ a layering strategy, which works until it DOESN'T!!! And when it fails, you lose big due to the adverse risk:reward ratio. Walt
What do you mean by layering? If you mean averaging down on a pair that's diverging further than your initial entry, I agree that it's a weak strategy.