Ah, yes, the venereable FIR filter! Don't let the closely-guarded secret of the efficacy of the single iteration filter out! Do you tune yours to handle 20 point death spikes? The joy of anything other than a Kalman is that you know it is a solution to a multitude of practical problems. You just don't know if it is the solution to YOUR problem.
Yes, trading is competitive. Having said that, we are actually providing jacksmith a great service here (as we have attempted to provide to others on many an occasion).
Sorry, I am still embarassed by my youthful excesses trying to optimize two filter parameters simultaneously. Oh, the times better spent sleeping! But I have suffered worse embarrassments chasing the DSP will-o-the-wisp. The more powerful the tools, the worse the predictions diverge from reality.
If you want a VERY rudimentary broad and comprehensive coverage of all these types of analysis , try: Trading Systems and Methods, Kaufman He makes a great effort to provide a detailed encyclopedia of methods that have been attempted towards markets (DSP, markov chains, kalman, etc..), even dedicating a chapter to the more esoteric psuedo-astrology type stuff. Keep in mind it is very a rudimentary survey, and most ideas are not explored much further. But if you are looking for ideas to explore (even some laymen accessible concepts on time-series, ARMA, etc..), there are plenty covered here. http://www.amazon.com/Trading-Syste...=sr_1_6?ie=UTF8&s=books&qid=1235550133&sr=8-6 You'll come across so many ideas it will make your head spin. ----------------------------------------------- There are of course, plenty of more rigorous texts that require a higher level of math, if that's your thing.
In my experience the most useful aspects of "dsp" algorithms is their use for noise reduction. Most people use simple or exponential moving averages which both have terrible spectral characteristics. I suggest you start looking at some of the good and simple filtering techniques, such as Kalman. I have had limited success with spectral estimation techniques, but that's just me.