Concept · example: BTC, week of 24 April 2023

Correlation

When two assets move together hour by hour, what does that number actually tell a trader holding both?

In plain words

Correlation measures how closely the hour-to-hour price changes of two assets track each other. A high positive value means they tend to move in the same direction; a value near zero means their moves are unrelated; a negative value means they tend to move in opposite directions.

Why it matters

Holding two highly correlated assets gives far less protection than holding two unrelated ones, because they tend to fall together. Correlation also says nothing about which asset performs better over the week, only about the similarity of their short-term rhythm.

In a real week: BTC, week of 24 April 2023

In the week of 24 April 2023, BTC and ETH hourly returns were correlated at 0.96, close to moving in lockstep, yet ETH underperformed BTC by 5.52 percentage points over the week. BTC gained 5.95% and closed at 29233.21. The ETH-BTC ratio changed by -5.21%, so the two assets tracked each other hour by hour while ETH lost ground against BTC across the week.

A common mistake

Reading high correlation as meaning the two assets perform the same. They can move together every hour yet one can drift steadily below the other all week.

Related ideas

betadiversificationcovariance

Research notes on real market weeks, not investment advice. Sources: Binance spot BTCUSDT and ETHUSDT 1-hour candles and USD-M perpetual funding rates; weekly figures computed by WSTiger.

RISCON Labsrisconlabs.com