Econometrics: one step ahead forecast - page 118
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
We should start with a review of the results in this field. Burg wrote his dissertation Maximum entropy spectral analysis in 1975. Ehler wrote his book Rocket Science for traders also about 30 years ago. His book MESA and Trading Marcket Cycles was also published in 1993. There are programs and indicators that implement these ideas. So before you reinvent the wheel, you should just network and read the books and get to the level that is available.
I'm writing this generally for everybody, in an attempt to protect people from your illusions. I don't have any illusions concerning the applicability of DSP in trading.
Dissertations are written to show off in front of others. They write them on all kinds of nonsense. I have a dozen dissertations to work on. But it's a pity to waste time on it.
Alexander, the less nonsense you read, the less closed-minded you are. You have a wider view.
============
Now for the digital processing.... I don't understand what you mean by that? In a general sense, how do you imagine signal processing on a computer? It's all digital itself! There's no other way to do it. Or you have to stop processing anything at all. Just because it's done on a computer.
If by DSP you mean past processing, I've already explained that I don't do it. I have BIH filters. They don't know the past. They are real filters. Unlike FIR filters, which can only be applied and implemented in a computer. Of course, the implementation of IIR filters is digital. Because it's in a computer :-)
Dissertations are written to show off in front of others. They write on all sorts of nonsense. I've got a dozen dissertations to work on. But it's a pity to waste time on it.
Alexander, the less nonsense you read, the less closed-minded you are. A wider view.
Vadim! I don't insist on anything. I have spent my whole life engaged in design work and I know how it is done and how it is not done. I'll only be glad if I get a successful acquaintance with a broad mind. Whether you learn something from me or not is your choice.
Once again. I'm not imposing anything on anyone. Moreover, I opened a thread to hear other people's opinions on the issues that interest me. I am not interested in DSP as I know exactly its place in econometrics, where to apply it and for what.
Dissertations are written to show off in front of others. They write on all sorts of nonsense. I've got a dozen dissertations to work on. But it's a pity to waste time on it.
Alexander, the less nonsense you read, the less closed-minded you are. You have a wider view.
============
Now for the digital processing.... I don't understand what you mean by that? In a general sense, how do you imagine signal processing on a computer? It's all digital itself! There's no other way to do it. Or you have to stop processing anything at all. Just because it's done on a computer.
If by DSP you mean past processing, I've already explained that I don't do it. I have BIH filters. They don't know the past. They are real filters. Unlike FIR filters, which can only be applied and implemented in a computer. Of course, the implementation of IIR filters is digital. Because it's in a computer :-)
The whole topic is richer than the last post you commented on. The question of variable significance has been dealt with many times. Accumulation of prediction error is a medical fact, as one takes the previous prediction value for the next prediction due to lack of fact. If a fact is taken, it is a prediction one step ahead.
But these are minor and technical issues.
The use of increments was. Nothing works, because in the increments there is no trend, but there is a forecasted trend. and here is the main question of the topic: what properties of the model give a guarantee of predictability? A whole set of such properties for an ordinary regression model has been suggested. What you are commenting on is a breakout model and there are other models here that I don't understand.
I would be grateful for your comment on any of the many points in this thread.
This is just a comment to the numerous provisions of the topic and the disagreement with them is this:
A trend is an increment on a sample of values by a certain lag, to previous lags and in such lags there can be more than one step. How to calculate this increment and assume a dependent increment for the next lag is a prediction model. At the same time methods of determination of significance of variables, just use step forward as the criterion, but not at all on lag - I wonder why with such common practice someone suddenly expects to get any guarantees of accuracy of forecast of exactly trend. Friendship with such "medical fact" is a straight carpet to a specialist psychotherapist... It goes without saying that the error accumulation will grow together with the lag size, but it does not mean reduction of forecast accuracy - for this measure is relative and is set by the correlation quality estimation, not the error size... Therefore, the choice of a model and its parameters is only a secondary problem, solved (and easily) after determining the size and properties of the sample of dependent variables...
This is just a commentary on the many provisions of the topic and the disagreement with them is this:
A trend is an increment on a sample of values by a certain lag, to previous lags and in such lags there can be more than one step. How to calculate this increment and assume a dependent increment for the next lag is a prediction model. At the same time methods of determination of significance of variables, just use step forward as the criterion, but not at all on lag - I wonder why with such common practice someone suddenly expects to get any guarantees of accuracy of forecast of exactly trend. Friendship with such a "medical fact" is a straight carpet to a specialist psychotherapist... It goes without saying that the error accumulation will grow together with the lag size, but it does not mean reduction of forecast accuracy - for this measure is relative and is set by the correlation quality estimation, not the error size... Therefore, the choice of a model and its parameters is only a secondary problem, solved (with ease) after determining the size and properties of the sample of dependent variables...
Holy crap! 2009... It's been almost three years.
I answered there. I posted a picture of my filter. It's only 22 frequencies out of 45. There's even a sum of the gold line. Again almost no one thought to use it. This is the closest answer to your question in the whole thread. This is the quasi-stationary market picture. All frequencies have an unvarying period. There is an unstable amplitude. But it also changes smoothly. The unsteady modulation frequency is also harmonic. Yes, it does not matter. You can apply this function to each line a few more times. The lines continue into the future smoothly without jumps. One bar can always be predicted with very high accuracy.
Everything that has been told (problems and perspectives) in our conversation can be seen in this picture.
Box and Jenkins also use similar solutions in some of their models, but by defining the spectrum of only the nearest low frequency subcarrier and using it as a moving average parameter, and using autocorrelation coefficients as the high frequency subcarrier. In fact your approach is more complete with respect to the frequency spectrum and thus possibly more accurate... on the other hand their approach probably has better adaptive properties, but this is not fully articulated in the publications for obvious reasons...
How to find the key to determining the sample size? Maybe go down the path of minimising the RMS from the regression equation?
Well, yes, just like you know the exact place of information entropy in econometrics. It doesn't seem to be provided there?
Holy crap! 2009... It's been almost three years.
I answered there. I posted a picture of my filter. It's only 22 frequencies out of 45. There's even a sum of the gold line. Again almost no one thought to use it. This is the closest answer to your question in the whole thread. This is the quasi-stationary market picture. All frequencies have an unvarying period. There is an unstable amplitude. But it also changes smoothly. The unsteady modulation frequency is also harmonic. Yes, it does not matter. You can apply this function to each line a few more times. The lines continue into the future smoothly without jumps. One bar can always be predicted with very high accuracy.
Everything that has been told (problems and perspectives) in our conversation can be seen in this picture.