Pairs Trading

Pair trading is a statistical arbitrage strategy that relies on the short-term fluctuations between two cointegrated stocks over the long-term equilibrium. According to Caldeira and Moura, the desired characteristics of this type of strategy is market neutrality, allowing firms to profit regardless of the market sentiments. As such, this strategy relies on two cointegrated stock pairs that share a long term equilibrium relationship. Therefore, it is presumed that short term deviations in price will ultimately level out and allow traders to profit when there are deviations from the long-term equilibrium. The trade is opened by purchasing the undervalued stock and shorting the overvalued stock by the same amount, hence yielding a net position of zero. The strategy relies on finding cointegrated pairs with p-value of less than 0.05. Once we determine stock pairs via the Engle-Granger Procedure for testing cointegration, we perform linear regression on the two series to determine if they are indeed stationary. If they are indeed stationary, we extract the linear regression coefficients to use to create the stationary linear combination. Then, we calculate the z-score of the stationary time series and trade according to the strategies.