Some time ago I spent 'a lot' of time looking at digital signal processing and time series data. There is a good discussion on Tradestation forum the subject is there is anything in phase, amp, or frequency that can be used to trade. The consensus was no. Previous phase, amp, or freq has nothing to do with the next bar, only the past. Too bad, it makes great charts. You can look at 'Rocket Science for Traders', get all of the ELDs from Tradestation.com and see for yourself. A better set of C++ tools is available on Numerical Recipes site http://www.nr.com/. Between the Tradestation ELDs and the NR code you can look for yourself to see if there is anything helpful. Or, study price action by itself which is a better use of ones time. The written work by J.M. Hurst is much more helpful that the more recent computer tools. (At least for me...) Has anyone found anything useful with signal processing???? Good trading next week. Should be interesting!!!
I use John Ehlers' Hilbert Sine Wave oscillator and find it very useful in 2 ways: 1. Identifying whether the market is in a trending mode or cyclical mode. 2. Identifying high probability turning points. I find the trick is to use multiple time frames. In trading Emini's I track daily, 135 minute, 81 minute and 45 minute charts. I recently wrote an article about the sine wave having 10 out of 10 winning long trades in the last up run in the Emini from July to December. You can check out the article here: http://emini-watch.com/emini-daily-update-cycle-indicator-10-out-of-10/68/
thanks for the input. i have been receiving e-mails from david gurwitz and his cycles analysis. has anyone heard of him? thanks again, bgp
Conventional DSP (Fourier) doesn't depict the series well because market data is non-stationary and plagued by impulse events. Those more experienced in signal processing might say I'm incorrect, but I don't think so. Take a look at wavelet analysis instead. I'd say it's much more tenable for market data. Neither, though, is well suited for predication, but rather for classification instead.