Econometrics: one step ahead forecast - page 5
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
it is possible to predict, only the prediction is meaningless, although the prediction error will be normally distributed.
I completely agree, but a quotient is not a random wandering, it can be seen with the naked eye, you can see quite specific trends. So we extrapolate them. And we only trust extrapolation if the residual is stationary (mo constant and variance) and not randomly distributed.
Yes, but stationarity of the residuals is not enough. It's about more useful properties that already imply stationarity of the residuals. The return on the residuals should be to the predicted value. I don't know if there are such tests in econometrics. For example Hurst index or h-volatility are among them, but they check the reversion of the series itself, but not to the predicted values. Although every TS based on returns is such a test in itself :)
I don't know the link. It's in the terwer textbooks. And it follows from the fact that the Wiener process is a martingale.
Sufficient conditions for adequate prediction have not yet been invented. The necessary ones are plentiful. So we pick up new bricks at random, i.e. necessary conditions, in the hope that someday their set will be sufficient.
This seems to be the whole point of econometrics.
As a communist communist - I support you: either a bright future or nothing.
I don't see how one can take seriously a model that takes into account only the last few values - even if for several currencies. Here "a few" is one or two.
Why the latest. Above, it was 1 +2, then 4 +2 and got better. Factor in 100, who's stopping you. But econometrics with the help of tests you do not like will prove 100 or 99 or 101 or 1 or not possible at all in this area kotir.
I don't know the reference. It's in the terver textbooks. And it follows from the fact that the Wiener process is a martingale.
No question for martingale, even made a TS for them. But it's still profitable on some plots and unprofitable on others. Above I called martingale a leverage.
Yes, but the stationarity of the residuals is not enough. It's about more useful properties that already imply stationarity of the residuals. The return to the predicted value. I don't know if there are such tests in econometrics. For example Hurst index or h-volatility are among them, but they check the reversion of the series itself, but not to the predicted values. Although every TS based on the return is such a test in itself :)
The stationarity of the residual means that I can continue not only the deterministic component but also the error - it will not jump. Mathemat argues that this is not enough. Maybe he's right. But I'm not a philosopher, I'm an artisan. Stupid. We take the cotier. It's not predicted as there's some other one in the next plot. We try to decompose it into components. We see a trend and a shift - we single it out and look at the residue. What is it? Is there a trend left in it? Is the variance bouncing? If there is still a trend, we delete it and ask the same two questions again. If the variance jumps, then model these jumps and remove another type of uncertainty in the original quotient.
Where is Hirst and his Nile crocodiles here? On what sample did he base his indicator? Private observation.
Where is Hirst and his Nile crocodiles here? On what sample did he base his figure? Private observation.
You lack a theoretical basis.
I get the impression that you also equate econometrics with the instructions for the miracle program you use.
The stationarity of the residual means that I can continue not only the deterministic component but also the error - it will not jump. Mathemat argues that this is not enough. Maybe he's right. But I'm not a philosopher, I'm an artisan. Stupid. We take the cotier. It's not predicted as there's some other one in the next plot. We try to decompose it into components. We see a trend and a shift - we single it out and look at the residue. What is it? Is there a trend left in it? Is the variance bouncing? If there is still a trend, we delete it and ask the same two questions again. If the variance jumps, we model those jumps and remove another type of uncertainty of the initial quotient.
Where is Hirst and his Nile crocodiles here? On what sample did he base his figure? Private observation.
Clearly. Wouldn't it be more logical to build a profitable TS and check the distribution of equity increments (not the balance) for normality as a selection criterion? Otherwise the cart was ahead of the horse :)
P.S. Perfect equity is a random walk with an upward drift. The distribution of increments is normal and the lower the dispersion, the more grail)))
P.S2 And it is important that equity drawdowns do not have thick tails (one of the characteristics of NR) and let them be upwards.) For example, trend-following systems will have such spreads. The Sharpe ratio estimates quality of equity on the basis of dispersion, while Sortino takes into account only dispersion of equity movements downwards
Don't confuse martingale with martingale, colleague. You should be uncomfortable with such confusion...
I see. Wouldn't it be more logical to build a profitable TS and check the distribution of equity increments (not the balance) for normality as a selection criterion?
We have checked it and what if something is wrong? Where to look for it and by what means?
I see. Wouldn't it be more logical to build a profitable TS and check the distribution of equity increments (not the balance) for normality as a selection criterion?
We have checked it and what if something is wrong? Where to look for it and by what means?
How to check the series of equity increments for normality of the distribution? You seem to have done this quite scientifically in your articles.
And searching for TC with a tester and including logic.
How do you check the series of equity increments for normality of the distribution? You seem to have done this quite scientifically in your articles.
And searching for TS with a tester and including logic.
I mean the TS itself. After all the bad equity is buried there. My main complaint about TA is that you can't tell what's bad inside the TS if it's bad.