Range of R-Squared in Multiple Regression

Reported in interviews at D. E. Shaw Tower Research Capital Two Sigma The Voleon Group

Problem statement

Let $x_1$, $x_2$, and $y$ be random variables. The simple regressions $y \sim x_1$ and $y \sim x_2$ each produce $R^2 = 0.01$.

What is the possible range of $R^2$ when you regress $y$ on both $x_1$ and $x_2$ simultaneously? State the assumptions under which the minimum and maximum are attained.

Show hint

How does the joint coefficient of determination depend on the three pairwise correlations among the two predictors and the response?

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

Related free problems