eThesis - Kauppakorkeakoulun sähköiset gradut

Objective of the study is to further investigate pairs trading strategy on the U.S. equity markets and the possible explanations for the strategy profits. The main contribution of the study to the existing literature is to provide a thorough empirical study of pairs trading profits and to examine the unexplored link between pairs trading profit and short-term return reversals. In addition, this thesis studies pairs trading profits in different market conditions and the exposure of the pairs trading profits to known market risk factors.

The data used in the empirical study comprises of daily stock market return data for the S&P 500 equity index constituents from December 1967 to December 2009. The data set is extracted from Center for Research in Security Price (CRSP) database, consisting of observations for 10824 days and 1331 unique S&P 500 constituents.

The main findings of the thesis show that the pairs trading strategy yields markedly higher risk adjusted returns compared to the S&P 500 benchmark index, which are comparable in size to the earlier studies on the topic. According to the return analysis, the strategy on average yields small but positive returns for majority of the periods, but the returns are at least to some extent driven by small number of high return periods.

I also find that pairs trading profits persist over time, contrary to earlier studies, and profits to be dependent on the overall market conditions. Although the returns show some time variation, the returns persist also to this day. The pairs trading profits are highly exposed to funding liquidity risk as well as overall equity market volatility, which imply that arbitrageurs are constrained to enforce relative pricing discrepancies. The exposure the overall market conditions might prove trading according to the pairs trading strategy difficult in practice.

The main finding and largest contribution to the previous literature is that pairs trading profits are independent of short-term reversals and therefore independent of liquidity provision. The profits from pairs trading are statistically highly significant and are not explained by the relevant factors included in the factor model. I find pairs trading profits to have a statistically significant negative correlation with market excess return and Fama/French Small Minus Big factor and positively correlated with medium-term momentum profits. So far, pairs trading remains as a unique anomaly, unexplained by the academic finance community.