10 real pips for me and it was helpful on the m1/5/15. Running particular on m1 like on rails. 
Bought at 13489 and sold at 135. Here is my chart:
Just look at the white/blue line in there. It was running as predicted, just faster.
Now it goes up a little further.
Well, that doesnāt tell anything about the long term success, but I have a feeling this indi is one of the best Iāve seen so far. And Iāve seen a lot. 
Thatās how it looks now:
Pretty much the same than before, just some minor bumps are different. Still running like on rails. After topping off at ~ 13510 it corrects to the south.
Well, theoretically at least, if you have a really simple wave with a few well defined frequency components, you can turn a fourier spectral analysis into a predictive curve like this.
Example: almost anyone could take a good guess at the continuation of the curve in the upper left, here:
But the fit would have to be phenomenally good, in order to get away with that. Not just a few questionable looking peaks and valleys fading in and out of the noise, which is what Iād vaguely expect to find.
Assuming we used a filter that ignored frequency components longer than one day, and weighted only recent data for a curve fit, I suppose this is possibleā¦ but Iām still trying to wrap my head around a rationale as to why this would, or even could, work so well.
IIRC thereās a member named Simba over at forex-tsd whose done quite a lot of work on spectral analysis. Not sure how much of his stuff heās made public but Iāve always found him to be knowledgable and helpful.
I really have no clue how they coded the analysing. Anyways, to filter out the noise is easy. You can do that with any ma. Thatās what most traders do already.
It should work in any quiet sessions where no āignitionā like news or anything is likely. Plus as I said, that it works right now so well doesnāt mean it works all the time. One has to look for other indis / pa / whatever as well. Anyways, how it works right now is amazing. Iāve never seen anything like that before.
Here is the actual chart. Still running like on rails. Now wait for the upmove. I hope it works, lol.
Hmmm I wonder if we could get that person to come over and tell us why, in fact, this might work so well?
The implications would be astounding, though. It would be as if the top mathematics wonks at all the major banks used a garden variety algorithm to trade withā¦ itās justā¦ I donāt know, it strikes me as too glaringly obvious to be even remotely possible.
About on par with the secret code to the US nuclear arsenal being ā12345āā¦
If Buckscoder is onto something here, weāll have a shockingly clean~looking fourier spectral graph with well defined peaks and valleys, at least if using proper filtering over the past few hours or days. This canāt be ātooā rightā¦ if it is, we are all rich. At least until Western civilisation collapses three weeks later because everyone is now a billionaire.
Quite a headscratcher, this. I really want to feed the data into a maths program now. That will say quite a lot with regard to how good this is, or ever was.
Now it runs out of order:
The top of the predicted upmove is now 10 pips lower. Went also 10 pips lower than one would have expected. Plus then itās the m1 chart. I just took it to have some development quick. If some stop hunters are going for the stops thatās hard for any fa indi to predict. On the weekly or so timeframe it might be more accurate.
On the m5 chart the prediction was all the time the downmove. Well, nothing works all the time. Iāll probably take this indi as āidea indicatorā where price could go to. If it then is in confluence with some other indis like the MD sequential what I used for that last trade it is imho worth looking for a trade. Good thing is that the indi corrects the prediction after any candle.
One of the common tests in physics is to actually apply a small impulse to a physical system, to see how it reactsā¦ the reaction says volumes about the system itself. A bit like striking a bell with a tiny hammer. The sound you hear will say a lot about the size and shape of the bell. Or drum, or string, or jello (no sound in that case). Perturbation theory stuff.
For instance if a news event ārings the bellā but then the response damps out to nothing within minutes, the forces controlling the market greatly outweigh the news impulse. But if it goes waaay off suddenly after that, youāve got a system thatās very weak, and responds dramatically to news. Not that we can really ācalibrateā news, but, thatās vaguely how it works. A news event would have a certain spectral signature and an attenuation rate, or maybe an amplification effect, if the news were to be āMoney Tree discovered in Amazonian Jungle.ā
Iām afraid from saying that, but itās far from that, lol. I mean, itās imho a really nice indicator, but you have anyways to build a system around it and combine it with other tools. The āfizzleā in successful trading lies not in predicting every move with accuracy of 100%. This indi will also give false predictions.
Itās really just that itās a nice tool. One has to know when to use it or leave it in the toolbox and better use something else or stay at the sidelines. Plus then somebody needs a feeling for the market anyways. Even with this indi most ppl would probably still buy at the top and sell at the bottom. 
It might have to do with time of day, as wellā¦ Iād expect a look ahead indicator to āfailā with regularity no matter how good it is, when the banks changed sessions, &c. Unless that were somehow taken into account. Would be interesting to see how it does during say, London or asian sessions, comparatively speaking.
I think Iām going to download it and watch it for a few weeks.
Exactly. It needs to get a feeling when to use it and with what settings. Also look at the white line. If the white line is on par with the candles then it is a good indication that itās in sync with the waves. If the price oscillates around the white line it is not in sync. Forgot the originial colors, the white is the past data for calculation and the blue is the prediction.
Aww man I canāt believe I was away whilst this thread got underway. Fourier Analysis, filtering, signal synthesis, it all reminds me of the 2nd year of my degree, fun stuff
My memory is a little fuzzy on how exactly FFT works though, I think Iāll have to dig up my old lecture slides and MATLAB files lol. Iāll keep reading until I possibly have something to contribute.
Maybe you can look at the source of this indi and if it can be improved or what settings might be best.
At the site they say itās done by the Quinn-Fernandes algorithm.
This is what I mean with confluence of indis:
A pretty good buy at 13480 (institution level as well) and good at least for some nice 40 pips. If not more.
Iāve been lurking on this thread from the beginning. Interesting stuff.
I wonāt tell you how long ago it was that I studied Fourier analysis in school. And, Iāve never had the need to use it professionally. Consequently, whatever understanding I ever had of that subject ā it has just faded away (along with the French I studied).
So, unfortunately, I canāt contribute anything here.
But, this is a fascinating conversation. I hope you guys will continue it. And, if you donāt mind, Iāll go on lurking.
A confluence of indicatorsā¦ is that like a murmuration of starlings, a congress of baboons, or a murder of crows? Though, yes, itās good when everything points in the same direction!
Iāve had my eye on the fourier āpredictiveā indicator for a fair bit of the London open now, and, wellā¦ bleah. It certainly has no shortage of predictive error at the moment.
Possibly the most critically lacking bit with it is an āestimation of errorā ~ another science thing, if you are going to predict the future in any way, itās important to know how good or bad a prediction youāve got. Thatās the good thing about spectral analysis, you can see how large, sharp and well defined the peaks are. If itās a fuzzy, barely there mess, you know itās not very goodā¦ even if it would produce the same predictive result as something 100 times more reliable.
As for continuing the conversation ~ my main problem is that Iām good with questions, not answersā¦ if I had the answers this would be a very different thread indeed!
Yes I have some background with scientific analysis, processing the data isnāt hard once itās in hand, but there are some deeper questions with regard to āhow relevant is it.ā One, do markets really work in a way that this makes sense. Markets are a āpeopleā thing not a natural process. And two, even if this does apply, will it continue to do so. In the same way. It also might āsort of workā in the way that mathematics can describe population numbers of rabbits and foxes, without having to know everything about each rabbit and fox. I donāt have the answers to these.
What I would love to see, though, is statistical signatures ~ such as peaks on a graph ~ indicating where ābest oddsā are, and when. I think such things must exist. Consider ICTās killzones, Gasanvillās long term trend tradesā¦ VSAā¦ these things are real, they do work for some people. Unless we write off all the successful traders here as just ālucky for yearsā there are mathematically measurable, predictive advantages. Perhaps slight, perhaps intermittent, but quite real.
Of course it may be a lot easier to simply understand the fundamentals and the trading simply as described by the people doing it. But I really want a look at the āwhole elephantā ~ Iād like to see peaks on a graph that roughly correspond to some various successful methods. Of course it may not be quite that simple. But knowing that the graphs are created easily enough, wouldnāt it be nice to take a look?
The great thing about a forum is serendipitous input ~ Iām eager to hear what others have to say. There are some very educated and experienced people here, letās just hope someone has enough time on their hands to convert and process the data. Eventually I could get around to it myself, but it may be a while.
I canāt find the post now, but I received an e-mail that someone had uploaded a data file of one of the indices. Is it deleted, or am I not looking hard enough?
I wish my coding or knowledge of this stuff as that good, unfortunatley 1 year of MATLAB doesnāt really cut it 
Iām not really sure how Quinn works mathematically (Itās a bit out of my depth) but supposidly the outcome is designed to āestimate the frequency of sinusiod in the presence of noiseā. Do markets move in Sinusiodal patterns, I canāt really see that it does in reality?
Lets for a moment work on the basis that ideally the market price, in the absence of noise, accellerates and deccellerates between supply and demand levels, as per my beautiful diagram, and forms a consistent wave as orders are met.
You could assume quite happly that the average price in a market stalls at supply or demend and accellerates to the next level once all the available orders are filled. Which would leave us with a somewhat sinusiodal wave, depending on how quickly the orders are filled.
The whole idea behind Quinn-Fern, is to predict the frequency of a wave with a relativley small dataset. So for trading you might be able to use the equation to predict after what time the wave will peak, it would simply be a case of buying after a certain amount of time as the wave should reverse. But that seems pretty counter intuative as to how markets work and certainly doesnāt predict future moves with any amplitude accuracy (so I have no idea how the indicator youāre using tries to predict the future!).
The real problem comes with filtering. How do we filter out the current market noise with only a small dataset (we would have to assume that the level of noise the market was relativly constant)? For a filter we would need the noise coefficient, Īµ(t), to perform the calculation accuratley. The standard Quinn-Fern equation assumes Gaussian noise distribution (I think), but I have no idea if the ārandomā flucations in price around the consistent movement consistently adheres to a Gaussian distribution or not.
And how do we know how long oscillations will last? At what point do we need to calculate a potential new frequency? If we are continually calculating the expected frequency then any noise filtering means that we might miss/ignore important data which doesnāt match up with our expecting turning point, etc, etc. There are so many problems with applying this to a ārandomlyā fluctating market.
I might have gotten the totally wrong end of the stick lol, but for sound, electronic or mechanical waves, where you can be assured of more consistency in any oscillation and noise, Iād say this is probably a pretty accurate technique. But markets are too chaotic to apply such mathematical constants to, in my opinion!
What Iād be more interested in is looking at spectal analysis on tick charts or volume to use as a chop filter, high frequency oscillations indicating that a lot of transactions are taking place (suggesting a supply/demand level or a ranging market), whilst a lower frequency would suggest a steady trend. Once we noticed the frequency start to decrease we would see a breakout happen as the orders would be filled up and the price would have to move to a new level. 
Or perhaps using a filter to remove market noise so that we could observe an accurate average price, then simply look at the gradient of the filtered price to watch when supply or demand has been met.
Well I am more interested in the practical approach. I built my first radio with 12 and bought my first oscilloscope with 18. Built a transistor graph circuit for it. I know in principle how frequencies look like if they are mixed.
Given the markets are just peoples trading behavior one can see that there are waves in the market. Itās however not that easy, because those waves decrease in power over time. Then a new āignitionā comes around and it all goes on again. Like one throwing stones into a lake. Or some throwing stones into a lake.
Given the m15 from yesterday the prediction was rather accurate again:
Sure, it went down a little further (~ 10 pips), but given that the prediction is a predicted ma and not the price itself those 10 pips on the m15 is just minor noise. After that it went up as predicted. For a m15 chart which is rather noisy, that is already amazing. Particular if you take into consideration that I do this just with a random pick. I did not select the best of some. At least it went up to the predicted target of the fibb 62. If you look at the ma 40 which is the red line it was in convergence with it more or less most of the time. Sometimes oscillating around the ma. So, itās a pretty good ma predictor imho. The more timeframes you look at, the more reliable the prediction might become. Like with every indicator or even looking at naked price.
As I said DS, you need to play with the setting. Standard settings will probably not work. It has to be adapted to the tf and currency. Even then it will for sure not do it all the time. One has also take into consideration the general market state. If somebody throws new stones into the water, the old waves are not anymore that relevant.
