Abstract by Dean Sobczak
Implementing the Pairs Trading Algorithm
The author presents the methods and results of a simulation study he used to investigate the effectiveness of the pairs trading algorithm. This algorithm is an investment strategy in which investors alternate ownership positions between two positively correlated stocks with the assumption that any significant deviation from the historical average ratio between the two stocks is transitory (i.e., pairs traders believe the price ratio will eventually return to the mean over time).
The author finds that the optimal number of standard deviations from the mean to signal stock 1’s sale is 0.1. At this threshold, the average profit and probability of profit appear to decrease with increasing correlation between the stocks. Furthermore, the author shows that applying this algorithm to a real stock pair yields a profit that is greater than what an investor would have realized by simply selling the first stock after 10 years. Thus, the pairs trading algorithm appears to be viable investment strategy. Shortening the moving average window for the stock price ratios and varying the proportions of stock 1 sold at each selling point are two possible areas of further research.