New related paper to #12 – Pairs Trading with Stocks

"There are several questions to be noted in previous literature regarding pairs trading. One is the paradox noted by Do and Faff (2011) that points out to the fact that we are looking for the closest pairs in the past and expecting that they are the ones with the highest probability of drifting and then converging in the future. The stability of the price relation and the probability of divergence (followed by convergence) by uninformed demand are two distinct characteristics. This idea suggests that there is space for improvement in the way the pairs are ranked and chosen, and not only regarding the price relation stability but also including a second dimension (maybe a second form of ranking) that would take into account the probability of a pair diverging. Another interesting and related matter is that the methodology proposed by GGR (as noted by the authors themselves) doesn’t exclude the possibility of a pair being chosen to trade with a negative maximum expected return after costs (for pairs with a short historical standard deviation measure that is used as a trigger). This leaves the need to establish such a filter that would leave out some pairs that are by definition destroying value or in the best possible scenario not adding (if they do not trade). "

"Between the pair’s training period and the opening of a position the relation between the assets that form the pair can dramatically change. In an attempt to monitor events that could lead to such a change one can monitor the demand for that asset and try to find abnormal changes (increases) that could signify such an event. One way to do that is through the volume data. Engelberg, Gao and Jagannathan (2009) test the same rationale with news data and find that news affecting just one of the assets decreases the profitability and news that affect both assets increases profitability (they argue that the increased profitability could be explained by differences in the speed information is incorporated). It’s expected that the volume data yields the same result with the advantage that is far easier to be implemented in a trading strategy (the information is more accurate then the number and importance of news and is widely available). So it’s expected that if a common shock exists then a volume increase will occur in both assets and if that shock only affects one of the assets then the volume increase will be confined to an increase in the respective assets volume. With that in mind the first hypothesis comes: Are pairs that open with single sided shocks less profitable and is volume data a good proxy for shocks."

"A pair opens if the relative price of the assets diverges by more than they usually did in the recent past (the 250 trading days period). This divergence is expected to be temporary and the product of market friction and inefficiencies and the new market equilibrium is expected to arrive in a short matter of time when participants acknowledge the mispricing. If that expected equilibrium (the price convergence) fails to come in a reasonable amount of time then maybe something fundamental has change in the assets that form the pair and the new equilibrium is already set (the price convergence is no longer expected). If a pair remains open for a long time we can start to assume that the excess value we thought it had isn’t there and the risk of holding it isn’t worth taking. In this situation the hedge we thought we had is lost and there is no rationale left for holding the position and bearing the risk. The second hypothesis looks to answer that question: Is it a good strategy to limit to the time a pair is open?"

"The limit introduced on the maximum days a pair is open improves the risk return characteristics no matter what limit is imposed which appears to support the rationale that a convergence must be observed in a short period of time. We can also conclude that is best to avoid pairs that trigger around abnormal volume changes in one of the assets."

Are you looking for more strategies to read about? Check

Do you want to see performance of trading systems we described? Check

Do you want to know more about us? Check

Share onRefer to a friend

Subscribe for Newsletter

Be first to know, when we publish new content

    The Encyclopedia of Quantitative Trading Strategies

    Log in

    We boasts a total prize pool of $15,000
    Quantpedia Days Bring 1+1 Special Offer
    Quantpedia Days
    - bi-weekly research insights -
    - tips on new trading strategies -
    - notifications about offers & promos -