Intraday Spectrum Analysis - EURUSD 1m & 5m

In his review of the recent Hurst manual approach David has stressed the fact that spectrum analysis is really not the holy grail and an approximation is very workable and provides very similar results. As a matter of fact, spectrum analysis will average results of say a 20 and 24min cycle and create a broader 22min cycle. Just for interest and to help with visual understanding, I will post a set of 1m and 5m Intraday spectrum analysis for EURUSD to highlight variation over time and in particular how an impulse (the mathematical/control engineering term) impacts the spectrum. Note in particular that conventional spectrum analysis will not provide any information on phasing. Hopefully this will assist with cycle understanding.

Sorry people - cannot upload a table so you will have to work with the individual images, download these and then line them up if you like.

22/3/2017 21:00



Just a small insight: say take the 1m/31m cycle (based on 30m nominal cycle). The upper and lower boundaries are 27 and 33min respectively - slightly lob sided. If you add these FLDs you get a band of sort which often is a little more forgiving that the hardcore FLD line:

Very nice, keep those charts coming Lex.

Hey Alain, beautiful charts - but can you please clarify what I am looking at? Clearly envelopes (with fixed vertical displacement?) but not seeing the spectrum analysis linkage. What am I missing?

1 Like

23/3/2017 21:05

** The spectrum analysis only takes into account the most recent 512 bars. MESA can work with a much smaller sample that FFT.



Hello Lex,
Can you explain me more about your spectrum strategy and where you find these kind of charts?


Luther - 512 bars ? it is roughly 2 years of daily data or 2 days of 1 minute data on EURONEXT

The Envelopes gives you the Maximum Amplitude of the Cycle and an estimation of the Underlying trend

Spectral analysis gives yo uinformations on the probable or average length of the cycles

You can also check this length on the Upper and Lower Boundaries of the Enveloppes.

When Price Amplitude meets with Time Length , Gan said that Time and Price “squared” and that it is Time for a Directional Change

Hi Peks, to be clear this is not a strategy - but really about information that should further provide insight into wave theory as detailed so thoroughly by DH in the FLD strategy course and many of his postings since. It is a mathematical approach to determining underlying cycles on limited data. You don’t want too much data as cycles continue to shift. Google MESA (Max entrhopy spectral analysis for details).

To generate the spectrum I use eSignal with an EFS script by Divergence software.

My reason for publishing some of these charts if to help colleagues understand how the cycle periods do continuously expand and contract - and how harmonics create noise you need to remain cognizant off.

Hope that helps, Lex



Intro-day 1 min EURUSD cycle picture at the glance seems to be stable. Further testing is needed.
Following are out of sample tests. 60 minutes

and 30 minutes

and 10 minute

Sorry folks - had a few days off :wink:

29/3/2017 20:00

29/3/2017 20:00

30/3/2017 22:30

30/3/2017 22:30




Last one for the time being.

1m EURUSD 3/4/17 21:00

1m EURUSD 3/4/17 21:00

It is absolutely fascination to see how these spectra change over time. Planning to do a proper detailed study in due course. Clearly Intraday spectrum changes significantly - as David highlighted in his Hurst Signals videos. In particular the MESA approach is quite sensitive to inter-day gaps potentially creating harmonics that do not really exist.

I hope these spectra were of some use to colleagues by providing cycle insights from a different perspective.

Thank you for the comments and expertise and other ways of looking at the charts above.

Cheers, Lex

1 Like

Hey Lex,
What software you use to draw this kind of bars ? I would like to try …

Best Regards,

Hi Peks, I already responded to your question when you first asked? Scroll up!

Sorry just saw it :slight_smile:

OK, thank you for your response…