COINTEGRATION IN PAIRS TRADING UTILIZING JOHANSEN AND ENGLE-GRANGER METHODOLOGY

Abstract

The study aims to test profitability of pairs trading applying Johansen and Engle- Granger methodology. The main issues of pairs trading strategy are specified and described in the thesis: explained rules how to choose trading pairs, identifying the statistical equilibrium and when the portfolio is away from it, moreover, how and when the trading position should be opened and closed. This study introduces qualitative method of time series selection, ranking the sample under BICS sub-sector specification. Empirical results show that cointegration exists between assets, but relations are not stable and thus it is difficult to evaluate any in-sample long run relation that holds in the out of sample basis. It is interesting to mention that Johansen approach does not give the advantage in the investments. Comparing the results of Engle-Granger approach, that shows 16 positively performing subsectors, the Johansen methodology demonstrates positive performance of only 2 subsectors. In any case, the majority of positive results of Engle-Granger method for pairs trading are around zero. The evidence suggests that there is an influence of some common factors that decline drastically the profitability. Thus in this framework arbitrageurs are exposed to the firm specific volatility, synchronization risk and publish research effects. Another possible explanation is that with the classical econometrics methods, it is not feasible to filter the noise from the cointegration and this cause the effect of not stable returns in the long run.