Econometrics: bibliography - page 6
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
Well the books are awesome - but there's no time to go through them all - choose a decent algorithm for prediction - and we'll write it down in code and check it out. And we can talk about it endlessly.
If you're specific about the example.
By differentiating you have removed the trend. Your example tells a different story. The trend that you have removed, can be predicted as the forecasting error will be stationary (mo and sko approximately constants). This is the market now, but six months ago a second differential was required.
I wrote that I did a unit root test before that. The hypothesis that the TS series is rejected, therefore the series is not TS, and from what I know I can say that the DS series. I don't hear much, I don't know much. It is clear that it is difficult to measure something accurately with linear relationships if the nature of the process is non-linear, so I am now asking the head of the department to tell me about wavelets. There is some non-linearity (fractality) there.
And data with such estimation of ACF and CHAFC cannot be modeled, within those models that I know. That's all, if you tell me what model to apply for them I'll be grateful to you and include in my course work.
To be specific on the example.
By differentiating you have removed the trend. Your example tells a different story. The trend you removed can be predicted because the prediction error will be stationary (mo and sko are approximately constants). This is the market now, but six months ago a second differential was required.
This is in case it is TS, then yes the task is simple, the trend is linear and all is well. We work, we predict all is well. But here DS (I checked it, you can do it yourself, above post) that's why we switched to differences, the trend here is stochastic, and therefore the error is stationary relative to the stochastic trend. That's what this proves, and the price increment in pips is stationary too, and not the fact that it's random, it just looks like random. I think that's why sometimes you can earn on Forex by chance, by getting into algorithms, etc. by doing a system you start to get away from chance and then you lose your depo.
I recently read Shiryaev's book, I don't remember the volume, I think it was 2. So the example has a function that is deterministic, but at some values it behaves almost like a random one (white noise in the example). And it is explained that stochastic is not the same as chaos, but they sometimes have a high degree of similarity and can be difficult to distinguish. I mean that in practice traders trade differently (thousands of algorithms of strategies), it generates chaos in the market, and in this case randomness is closer than algorithms, but at the same time there is no pure randomness but pseudo-randomness. I've decided for myself to work with non-linear dynamics etc. in the near future.
If I am stating something wrong, please correct me, only correctly, so that I can assimilate.
2orb. There is a logistic function that generates mathematical chaos. It is perfectly approximated by a neural network, because it is a FUNCTION. And try to approximate the price series, I think I can go on.
If memory serves right:
x_(n + 1) = x_n * ( a - x_n )
P.S. Almost got it right (https://ru.wikipedia.org/wiki/%D2%E5%EE%F0%E8%FF_%F5%E0%EE%F1%E0) - see figure on the right. The links are not inserted for some reason.
This is in case it is TS, then yes the task is simple, the trend is linear and all is well. We work, we predict all is well. But here DS
TS and DS is a Russian dissertation invention.
The problem is different. My view. we extract the deterministic component from the quotient and look at the residual. If the residual is stationary, the deterministic component can be extrapolated. If not, then extract the deterministic component from the residual .... Is it possible to get a working system that way? Not in the general case, I have no proof of it. But in the attached attachments it is argued that everything will work fine if there were no kinks in the trend. But some suggestion is made for this case to overcome this nuisance as well.
Judging by the posts, the team doesn't understand what a "breakpoint" is. The model is fitted. At each new bar we re-fit and the new one matches the previous one. And then the new model by its parameters does not coincide with the previous one. It means that the kotier inside the sample has changed in such a way that the model parameters have changed. If the parameters are good, we can adjust them and hope that everything will be all right on the next bar. But sometimes the quotient changes so that the functional form has to be changed. Besides, the break is most probably not diagnosed after the arrival of one bar but it needs several bars, i.e. we have suffered a loss and here starts the song about SL.
Here is an attach about this problem. As I see it - this is the main problem in trading - this breakdown.
2orb. There is a logistic function that generates mathematical chaos. It is perfectly approximated by a neural network, because it is a FUNCTION. And try to approximate the price series, I think I can go on.