Edge effect on the way to the GRAAL - page 7
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
The system doesn't predict = you can't say that with such and such probability the price will reach a certain level, but it is possible to say that after a certain number of trades the probability of a certain level of AMOUNT of profit will be so-and-so.
mql4com, I certainly respect your point of view as much as any other. It (point of view) has a right to exist, the question is how interesting it is to me... >> that's what I'm trying to figure out now.
Gentlemen, good afternoon.
I am working on a reverse MTS. Expert Advisor gives BUY and SELL signals and the system reverses trades
depending on these signals.
SELL appears at the maximum of the indicator, BUY at the minimum. This indicator is a proprietary one, based on
on several sliders, which represents the momentum of price changes (similar to RSI, but not it).
The indicator itself is noisy, so tried to use low-pass filters and wavelets for smoothing.
Smoothing by wavelets turned out the most successful, but the edge effect (distortion at the extreme points) spoils everything.
I made an experiment: approximation of an indicator by wavelets from 2000 to 2008 on EURUSD on a one-minute timeframe.
The system gives on average 2000-4000 points per month without any MM or other additives. Purely advisor.
The graph of balance growth is smooth and almost straight.
Of course there is nothing to rejoice about, because in real online trading the edge effect gives only minuses.
Question: has anyone tried other methods of approximating functions or finding extrema?
Or is it a futile case?
Or is it still possible to come up with something?
Dear Desperado! How are you getting on with the edge effect?
My experiments with the Meyer wavelet have led to a couple of interesting ideas. For example, a remarkable property of a wavelet approximation by a Meyer wavelet is that it is smooth and differentiable. Maybe this can be used to determine the strength of a trend? For example, if we take high order approximations and look for inflection points, it might suggest that the trend is getting weaker.
It seems to me that the wavelet transform by itself cannot give any predictions. But it seems like a great tool to pre-process data at the input of any predictive algorithms. For noise filtering and long/short term trend decomposition, I think wavelet analysis is very good. For example, wavelet approximation, unlike MA, doesn't lag. The only problem is the edge effect. Probably one should think of a proper padding method, e.g. - continue high order linear approximation.
This is what my experiments are now. In the figure below:
In the upper graph:
- the thin grey line is the price
- The thick red line is the Meyer wavelet level 5 approximation. Asterisks indicate max/min.
- thick green - approximation. level 3. Asterisks indicate max/min.
On the middle graph:
- thin red - 1st derivative of the 3rd level approximation
- thin blue - 2nd derivative of the 3rd level approximation
In the lower graph:
- thin red - 1st derivative from level 5 approximation
- thin blue - 2nd derivative of the 5th level approximation
I suggest to pay attention to the following:
1. the long-term and short-term trends are very nicely defined. The lines are unbiased and smooth
2. You can determine min/max points (red lines on the lower charts) on the 1st derivative - for example, on samples 1850-2000 there is a clear channel. It is very easy to think of its application in a trading system for a channel case. The main difference from "classic" is unbiased, but again - possible edge effects
3. You can use the 2nd derivative to identify inflection points of a long-term or short-term trend (blue lines on lower charts). Note the blue line on the lowest chart - the crossover with zero occurs when the pre-long term trend slows down. For example, somewhere around 1620 an up trend started, which slowed down around 1670. It may also be used as an additional signal.
Colleagues,
I would be grateful for any information on Meyer wavelet implementation (links, books, code). I searched all over Internet, but didn't find anything but mentions. If you don't mind, please share.
So the edge effect is defeated by someone or not.
How can it be defeated if it is a fundamental property of the same wavelets.
You have to use some other methods.
Colleagues,
I would be grateful for any information on Meyer wavelet implementation (links, books, code). I searched all over Internet, but didn't find anything but mentions. If you don't mind, please share.
strangely, my google found it fine, here http://zhurnal.gpi.ru/articles/2007/093.pdf
How can you defeat it if it's a fundamental property of the same wavelets?
You have to use some other methods.
For example...
I would like to know (c)
In fact, it's worth reflecting on the very formulation of the question.
Whether prices need to be researched so close to the right-hand edge at all. EVERYTHING.
You are just at the beginning of your journey.
To make a prediction you have to answer the question - do you have a stable equation or not? If it is not stable, you cannot make a forecast.
To answer the question about the stability, we should analyze the residue between your indicator and the quotient. There is information there and this information may answer about the stability of your equation.
You have the beginning of success, because in my opinion the stationarity of the residual can be achieved using wavelets, and the stationary residual will give the stationary prediction error.
You are just at the beginning of your journey.
To make a prediction you have to answer the question - do you have a stable equation or not? If it is not stable, you cannot make a forecast.
To answer the question about the stability, we should analyze the residue between your indicator and the quotient. There is information there and this information may answer about the stability of your equation.
You have the beginning of success, because in my opinion the stationarity of the residual can be achieved using wavelets, and the stationary residual will give the stationary forecast error.