Floating market parameters - page 2
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
And you are trying to apply methods from one area to another.
And what methods can be applied? List them, if you don't mind.
The methods are (simplified) - you have to think about who may have entered by how much, why, and when they will leave. And from that, you dance around. It has to do with the market, for sure. For example, the volatility exists in the market, it changes during the day, it's an undeniable fact, it can be calculated. So that's the thing to work with. Or develop such a property of the market, as the degree of reversion to its average price (strange as it may seem, but in a simplified form it is MA)). Such a reversion exists in the bazaar, and it's also a fact... Or seasonality... Well what I've listed is a priori relevant to the market. Perhaps, even astromethods have something to do with the market - the phases of the moon, the mood of the masses, etc... But how do wavelets relate to it? Or quantum mechanics? Wavelets have to do with harmonic oscillations, for example music. If a DJ asked me about them, it would be logical. Quantum mechanics would interest nuclear scientists. And so on. But what does the market have to do with it? In fact, maybe there is a connection, but first of all, it must be justified (i.e. you have to tell yourself that groups of traders will behave like atoms in the nucleus because so and so. And then the programmer should go to the head and tail of this area. Otherwise it is just a waste of time. Imho.
Or develop a market property such as the degree of return to its average price ( oddly enough, it is MA in simplified form :)).
This is what I am doing, but I need to shift the MA (or something else) half a period back.) I want to use the Predictor for this purpose. First, I'll see what errors may occur and then either develop the method or look for another one.
But how do wavelets apply there? Or quantum mechanics? Wavelets have to do with harmonic oscillations, e.g. music. If a DJ would ask me about them, it would be logical. Quantum mechanics would interest nuclear scientists. And so on. But what does the market have to do with it?
For example, the method of maximum entropy came to radio engineering from geological prospecting, regression is used in radio engineering and economics, the price goes from level to level, which is not physics. I think we should experiment more.
what can a wavelet do?
The answer to this question can be found in any book that covers the basics of the wavelet transform - unlike the Fourier transform, which gives a frequency domain representation of the signal, the wavelet transform gives a frequency-time representation of the signal. If the signal is a sum of sinusoids with varying periods, the wavelet transform will show how those periods vary over time (unlike the Fourier transform, which simply gives a blurred spectrum).
Is it possible to modify the Fourier series decomposition? Take a limited number of periods for each frequency, e.g. only 2. Then the data would be less relevant compared to the usual decomposition. Then phase, frequency and amplitude changes could be monitored for modulation, e.g. by feeding them to the neural network input. Has anyone tried this?
People here are basically correct about the applicability of time series analysis methods.
One should first note this or that pattern or property of BP, and then select an adequate apparatus of matanalysis for the given series. As long as the cart is ahead of the mare, this is a waste of time and effort (maybe money). Our task, with all variety of methods and their complexity, comes down to forecasting the sign of expected price movement (Buy or Sell). If not to fall into mysticism, the prediction comes to the analysis of initial BP (history) or to the analysis of its environment (other tools). In the first direction all kinds of regression models are used, or if there is no power to formalize the task, neural networks are used. The second line of approach uses cross-correlation analysis with all its mappings.
People here are basically correct about the applicability of time series analysis methods.
One should first note a particular pattern or property of BP, and then select an adequate apparatus of matanalysis for the given series.
You could say I have found a pattern - fluctuations around a "fair price", now I am selecting a suitable method.
I searched for information on wavelets and it looks pretty good.
Advantages and disadvantages of wavelet transforms:
-Wavelet transforms have almost all the advantages of Fourier transforms.
-Wavelet bases can be well localised in both frequency and time. When picking out well localized multi-scale processes in signals, only those scales of decomposition that are of interest can be considered.
-Wavelet bases, unlike the Fourier transform, have quite a variety of basis functions whose properties are oriented towards solving different problems. Baseline wavelets can have both finite and infinite carriers, implemented by functions of varying smoothness.
-The disadvantage of wavelet transforms is their relative complexity.
I.e. there are no minuses as such.
Especially liked the results on the application here (Andre69 28.06.2007 20:43). There are quite specific frequency and time dependencies - somewhat stationarity.
In the archive there are files on market applications of wavelets and a comparison with Fourier.
IgorM, would you mind sharing the library?