Over the last few weeks I have been looking into creating a currency strength indicator as input to a Nonlinear autoregressive exogenous model. This has involved a fair bit of online research and I have to say that compared to other technical analysis indicators there seems to be a paucity of pages devoted to the methodology of creating such an indicator. Apart from the above linked Wikipedia page I was only really able to find some discussion threads on some forex forums, mostly devoted to the Metatrader platform, and a few of the more enlightening threads are here, here and here. Another website I found, although not exactly what I was looking for, is marketsmadeclear.com and in particular their Currency Strength Matrix.
In the end I decided to create my own relative currency strength indicator, based on the RSI, by making the length of the indicator adaptive to the measured dominant cycle. The optimal theoretical length for the RSI is half the cycle period, and the price series are smoothed in a [ 1 2 2 1 ] FIR filter prior to the RSI calculations. I used the code in my earlier post to calculate the dominant cycle period, the reason being that since I wrote that post I have watched/listened to a podcast in which John Ehlers recommended using this calculation method for dominant cycle measurement.
The screenshots below are of the currency strength indicator applied to approx. 200 daily bars of the EURUSD forex pair; first the price chart,
next, the indicator,
and finally an oscillator derived from the two separate currency strength lines.
I think the utility of this indicator is quite obvious from these simple charts. Crossovers of the strength lines ( or equivalently, zero line crossings of the oscillator ) clearly indicate major directional changes, and additionally changes in the slope of the oscillator provide an early warning of impending price direction changes.
I will now start to test this indicator and write about these tests and results in due course.