Spurious correlation: two drifting stock prices

Question

Two stocks in unrelated businesses both drifted upward over a decade. Their daily price levels correlate at $r = 0.98$ . A colleague says such tightly correlated names must be a great pair trade.

Why is the 0.98 misleading, and what should you correlate instead?

Accepted Answer

The flaw: correlated price levels are almost meaningless for trending stocks. Both stocks went up over ten years, so their price levels rise together and correlate near 1, even though the businesses are unrelated. Price levels are nonstationary (they behave roughly like random walks with drift), and any two upward-drifting series correlate strongly. The 0.98 reflects a shared upward drift, not co-movement that a pair trade could exploit. What a pair trade actually needs. Pairs trading bets that a spread between two names mean-reverts. That requires: 1. Correlated returns, not correlated price…

Spurious correlation: two drifting stock prices

Your answer

More Statistics questions