Relative value + pair trades

Hey all, I’m trying to learn more about relative value trading, specifically long/short pairs trading. Let’s take for example a company such as Coach (ticker: COH).

Obviously no company is a perfect match, but if your view was to long Coach (or whatever else you wanted to pick), what would you choose as a short and why?

I’m not an expert in Coach, but the name just came to mind as I was thinking about a belt I own that is made by Coach. To me, assuming that I thought Coach was going to have a great couple quarters and I also wanted to short something else, some risks I might be concerned about would include the cost of leather, fashion risk, and price point. So, maybe I choose to short a higher-end luxury brand such as Tiffany (TIF), or something that I didn’t expect to perform as well from a fashion standpoint such as Michael Kors (KORS), or even something in a different part of the value chain but generally sells higher ticket items that I don’t think will sell as well in a continued challenging economy, such as Saks (SKS).

However, no other company is exactly like Coach, and I don’t have much insight into which would make the best pair. Can anyone here please opine? I’d value any ideas, or recommended reading material for pairs trading. Thank you.

Also, can any investment professionals here please comment on the effectiveness of pairing a short ETF versus short equity as a means of hedging sector noise? I’m thinking that perhaps for a mid-cap company in some arbitrary sector that doesn’t have meaningful publicly-traded competitors, perhaps you’re better off shorting an ETF than to try to find some other publicly traded company in a different part of the value chain (e.g. an input manufacturer), because at least you can then hedge out the sector noise and still get beta and company-level idiosyncratic exposure.

Is one (i.e. stock vs. ETF) better than the other, and how would I evaluate or forecast performance of how successful a particular pair trade would be? Feel free to critique my line of thinking if this doesn’t make sense, or if there’s a better way to look at it. As you may know, I have experience in long/short equity but generally don’t think about relative value arbitrage, yet I’ve developed a recent curiosity for it.

I think basis risk (if thats what its still called when hedging not using futures) would be your primary concern. My approach would be to screen quantitatively for closely correlated equities in the same narrow industry. Then from that list pick an equity that was fundamentally inferior to short. My biggest fear would be to be right in picking an inferior company to short but still lose out because the price relationahip between it and my long is not firmly established, of course thise types of relationships don’t alwaays hold.

I think a classic pairs trade is something like long toyota short gm with the primary consideration being the relative strength of each companies franchise. My fear is that if you look for a short somewhere else down the value chain or in a less closely related industry the correlation between the two positiOns won’t be as strong.

Alternatively you could not put too much thought into it and just short LV because you think their brand is cheapened by all the rap videos. Heh

In a real pair trade, you think that one company is going to outperform another company based on some analysis, so the pairing is based on a thesis about the two companies, and not by liking one company and then searching for another company to short against it. If you just think the company is good, you can use a future or an ETF for the short end if you want to try to neutralize your market risk, though you will get hurt if the correlations or betas change, because your hedge ratio will go off.

Statistical arbitrage reduces the analysis to a historical correlation and a belief that any deviation from that historical correlation will tend to mean revert.

lf you have only two companies, you then have to decide whether to be dollar neutral or beta neutral. If you have more than two companies in the long+short portfolios, you can try to optimize such that it is market neutral in both dollar terms and beta terms, which tends to reflect a profit potential more purely related to your analysis.

Thanks for the thoughts bchadwick and bodhisattva. The types of pair trades I had in mind would be something such as long Target and short Wal-Mart, long GM and short Ford, long Pfizer and short Merck, etc. – you get the idea. In addition to the counter-directional investment theses for both companies, I would also be looking to exploit relative undervaluation of one stock versus overvaluation of the other – let’s say for example that industry comps are 18x P/E, but the undervalued stock is at 15x P/E and the overvalued stock is at 21x P/E.

I also realize the importance of understanding your trade objectives, e.g. whether you want to be beta neutral or dollar neutral. Assuming you wanted to be dollar neutral, what types of analysis would you want to do to predict the success of your trade? Knowing that past stock movements don’t exactly dictate future ones, what types of analysis would you want to do to evaluate whether you had chosen an ideal pairing? For example, would something such as the Sharpe ratio be useful, or is that too much of a retrospective measure?

stocks within the industry tend to move in lockstep a lot so your time horizon must be appropriate to deal with this…