OLS Regression on an Alternating Series

Reported in interviews at Two Sigma

Problem statement

A time series $X_t$ alternates deterministically between $-1$ and $+1$. Consider two scenarios:

(i) $X_t = 1, -1, 1, -1, 1, -1, \ldots$ (starts at $+1$)

(ii) $X_t = -1, 1, -1, 1, -1, 1, \ldots$ (starts at $-1$)

For each scenario, if you run an OLS regression of $X_t$ on time $t$ -- that is, fit the model $X_t = a + bt + \varepsilon_t$ -- what are the signs of the estimated slope $\hat{b}$ and intercept $\hat{a}$? Justify your answer.

Show hint

What does the OLS slope formula reduce to when the response has a known, simple mean?

Relevant roles: Data Scientist, Applied Scientist and Research Scientist.

Related free problems