Recurrence Plots represent a new and relatively easy way to detect patterns in chaotic environments. The principle is to determine a phase space representation and lay it out in such a way that salient features of the series appear as regular patterns and invariants on a 2D graph. Such picture can then be analysed graphically or analytically.
An interesting site on RPs is : http://www.recurrence-plot.tk/ (reading references below)
I have therefore implemented RPs in a VC++ program using the API provided by the toolset allowing extracting data and calculations in an external program. (Some minor updates were necessary, so a new version of the library itself will be soon released).
I here give you a preview of what RPs look like on a 60mins EURUSD chart, both using price (z-score) and a MTFS indicator. The price RP uses a delay based phase representation whereas the MTFS uses the built-in separation in 3 views of the stochastic indicator.
Z-Score Price RecPlot
MTFS RecPlot
We will now record, analyse first visually then with the addition of neural networks to better detect salient turning points in the market.
For more reading, see the links page on the above site. I particularly recommend reading Norbert Marwan publications.
