Mean reversion is an important property of a stationary time series.
When a stationary time series deviates too much from the mean, massive correction takes place to bring it to the mean.
Interestingly almost all financial time series are non stationary.
But it has been found that by adding or subtracting a few non stationary series, we can obtain a stationary time series.
This is known as Cointegration. A cointegrating time series is in essence stationary.
Read this post in which I explain how to trade correlated pairs using this concept of mean reversion.
GBPNZD and GBPJPY should be correlated. It is obvious. Both are GBP pairs.
First we calculate the correlation which is high so we can use the mean reversion strategy.
We look for an opportunity when the difference between the two GBPNZD and GBPJPY closing prices exceeds 2 standard deviations.