About this question
Hard · Statistics & Regression · Quant Trader interview question · statistics, cointegration, time-series, engle-granger, stationarity
You are analyzing two stock prices, Stock A and Stock B. Both Stock A and Stock B are found to be non-stationary when individually assessed using a standard unit root test (Augmented Dickey-Fuller test). Specifically, they are integrated of order 1, denoted as I(1). However, when you create a linear combination of these two stocks, defined as the spread: $S_t = A_t - \beta B_t$, where $\beta$ is a constant, you discover that the spread $S_t$ is stationary, meaning it is integrated of order 0,