If you’d allow a suggestion: Just put the indi on your screen, play with the settings and adapt it to your needs.
As it is a wave predictor, it is a “new dimension” indicator. Which gives one more edge in addition to those with trend lines, s/d, fibbs, pps, price itself, regression lines, etc. 
Here is the prediction from yesterday with the actual outcome of prices:
As one can see it went a little more down than expected, but with a 15 pip sl below the last swing low not an issue. After that it took longer to reach the target, but it hit the projected tp. This is the m15 chart as stated. The white line is the analysis line, the blue line is the projection from yesterday and the chart is the actual chart from right now.
Just to add my opinion regarding the lag of the outcome and the lower prices before. Imho that’s related to stop hunting activity, which for sure can’t always be predicted well. It depends on how many stops are below the actual swing low (in this case) or high. If there are many stops, the stop hunters may target them. If there are not many stops, they will probably leave them in place.
So that was imho responsible for the delay in the outcome. After stop hunting however the prediction was still very accurate. It predicted the upmove and the rather harsh downmove afterwards. As this prediction was made several hours ago, it’s imho still amazing what one can do with frequency analysis of charts. For a successful strat it just needs to be looked at the stop and target requirements, but that is valid for every strat. It’s not uncommon that stops are placed below a swing pivot. Sometimes it will be hit, sometimes not.
Shroomhead the markets aren’t sinusoidal, it’s more the idea that you can build any function out of a bunch of sinusoids, of varying amplitudes and phases. And multiplying them all together. You can build any function out of this, to sufficient accuracy, given enough of them. So the market doesn’t have to be inherently sinusoidal for this kind of analysis to work. There are some more subtle things that could invalidate this approach, though.
As for anyone uploading a data file here, I am not aware of it? But then I miss a lot.
I wasn’t able to get good results out of this indicator, nothing like what Buckscoder has been able to, but maybe I am misapplying it badly. That, and still trying for trade consistency more than analysis right now, so I was very distracted. The impulse~response argument is pretty good, though. We see this with news events. It’s like the market was struck with a hammer and it reacts. Or maybe ten thousand small hammers at once… I guess it really does come down to individual orders.
From looking at the results so far… my guess is that a day’s worth of data would reveal clear peaks on a spectral analysis… just like a chart of the past day shows exactly where price has been. But would it be useful? Not after the fact, in isolation… any more than yesterday’s chart is useful today. Each day would be different.
Only when taken together, collectively, would we see if “most days did this” or “most days did that” … or “Monday London open does this X% of the time, by Y hour.” The actually reliable statistical edges may be quite small. And perhaps there are time periods smaller than days where this is useful ~ hours, maybe. Plus, how much weight to give preceding data, and how far back to go? Another question.
Of course I may be missing something important. I do not yet understand why/how/if this could work on a minute by minute basis. Maybe it can. I don’t have trade experience or knowledge to say one way or the other.
What package are you guys using that is captured in the screen shot?
Rich
I wasn’t aware that Fourier series was also being used for the Quinn-Fernandes algorithm (seems so obvious now lol). It explains a few gaps in my understanding, thanks for clearing that up.
I use my own indis with my own settings. No package. Some indis are from the net, tho. There ya go:
MQL4 Code Base: expert advisors, custom indicators, scripts and libraries
I used MatLab several years ago when I worked at a unversity research lab. Octave is a GNU Open Source package very similar to MatLab except for the price (FREE!).
I’m working on getting daily index data or pairs if possible. I’m curious what the spectral analysis would reveal if anything.
Rich
That would be wonderful if you could put that together!
Perhaps what we should most easily expect to find, is confirmation of what a lot of the professional traders say. Such as daily highs and lows more toward the first four hours of trading, and stuff like that reflected in the data. Plus the volatility of the different sessions should stand out very strongly. Perhaps some of that won’t be useful trade information. But it would serve to calibrate that we were seeing the market correctly.
Momentum would be really interesting, too. For instance Gasanvill’s strategy uses momentum, or basically almost any trend system does. My guess is that we’d need ‘price rate of change’ or ‘price slope’ in order to look at momentum, then analyse that. Something any data analysis package could do. Just a bit of fussing with it really.
Another interesting one is VSA… though adding volume as a third variable, I’m not quite sure how to do. Perhaps trying to analyse that with these methods would just make it harder than it really is. Or maybe this whole thread is making it harder than it really is But I’m still curious. Some people have stats for an entire sports league and would memorise that stuff. I’d love to see as many ‘stats’ for forex as I can get my hands on… not so much that I think there’s a ton of money to be made that way, but more as an intellectual pursuit and perhaps a way to gain deeper insight.
Maybe there are some unseen correlations that nobody’s paid much attention to before? Or maybe not. It could also possibly ‘rule out’ a lot of bad trading strategies too. Providing that past data is useful going forward, which… well, that’s a big ‘if’…
Another update; I saw this in the news or something and am thinking of doing some data mining using EURUSD just to see what it turns up: Home | Maximal Information-based Nonparametric Exploration
Seems to be a free ‘run at home’ java thing developed by some professors, and good at sniffing out data correlations. Not exactly fourier analysis per se, but it would be very interesting to see what it turns up. I’m going to be playing with it soon as I get a chance, and if I come up with anything good I’ll post here.
If anyone beats me to it and miraculously squeezes every bit of market advantage out before I do, well, then just send me a pizza…
Hmmm, did I come to this party too late…?
I’ve looking into Fourier analysis as I’m trying to fully direct my trading to 100% algorithmic. Unfortunately, my math skills are limited (freshman level), but I have looked into statistical modelling as well as a few other concepts on essentially finding a predictable edge over the market. Give me a PM DS, maybe we can bounce some ideas.
Also, just something you could look at as a logic sequence to “quantify” things, look into “piece-wise functions”.
Regards,
Clark