Pairs trading helps to eliminate market risk, and absent any significant fundamental reason for the ratio between pairs to change, the expectation is that the mean ratio will be obtained again in short order. If there's a fundamental reason for a change in ratio, then all bets are off. RD/SC is a good example. The market was paying RD a premium versus SC due to greater liquidity and its inclusion in the S&P 500. The premium disappeared quickly. No way to predict it. If you were on the wrong side of the trade (as I was), you lost. I was on the right side of an AA/AL trade when AL was removed from the S&P 500. In both cases, just luck. I think these anomalies are the greatest risk in pairs trading. I have a high percentage of winning trades but the few loosers can be big.
oh yes. thanks to el perro verde for showing me it! are you trading it? babak, sorry i don't have a copy. i just remembered reading it. i don't see anything too enticing about it, although it warrants further investigation. perhaps waiting until the pair actually starts to converge again and then entering, setting a stop just beyond the recent divergence high would work. but then why stop out? if the pair continues to diverge it'll just appear more and more attractive. in fact, that is precisely how a trader i know lost a big chunk of his equity. to get this to work, i think there needs to be a whole lot more to it than that. otherwise it's about as reliable as blindly continuing to sell naked premium. (the goings great for a while, then BOOM) i'd be interested to see some more work on playing divergences instead. it's something i've got on the agenda, but it'll be a while before i do any serious study on it.