Econometrics: one step ahead forecast - page 22
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
If you knew, you'd be living in Sochi....
It is very difficult to judge about forecasts for the future. Although I had a system based on the opening or closing of banks, one bank hands over to another one to one rate. It agrees of course, but time is a bit variable plus or minus 5-30 minutes. Here is the problem
The problem is different. A prediction has been made. How trustworthy is it? On what basis can it be trusted?
There is prediction error, which is something, but by no means everything. My articles show that the model by which we forecast must be stable. Not the market, but the model. By here on the forum I can't get to that stage, because unexpectedly the model I'm using predicts correctly. But the questions still remain, as time success in forecasting is not proof of the success of the model.
The problem is different. A prediction has been made. How trustworthy is it? On what basis can it be trusted?
There is a prediction error - it's something, but not everything. In my articles it is shown that the model we are predicting by should be stable. Not the market, but the model. By here on the forum I can't get to that stage, because unexpectedly the model I'm using predicts correctly. But the questions still remain, as time success in forecasting is not proof of the success of the model.
I see. But what about testing it on the quotes history. For example, in FoxPro?
For example - estimate the probability estimate of the prediction? If it is greater than 0.5, the advisor can already write
I see. How about testing it on the quote history. In FoxPro, for example?
For example - to estimate the probability estimate of a forecast? If it's more than 0.5, we can already write
What does FoxPro have to do with it?
There is testing available. The code for the EA is in the appendix to the article. I don't think it's the right time. We need to build the model correctly.
Now I get it. It's not really important to me at the moment. I have other problems to solve.
You don't want to see the problem caused by the KP used... But this is very important.
Look: each output point of your filter is calculated taking into account not only the previous values, but also captures, with respect to the calculation period, the future value. That is why it lays on BP so nicely. But this future value is not available at the end point of the series - and such a "future" point is necessary for calculating the KP filter's output. How to be in this case? - think the creators of this miracle. I suspect, that they took the path of least trouble, and in calculation of an end point they use the last actual value of BP as the "future" value. i.e. they set X(t+1):=X(t)
To see the problem more clearly, do calculations of two simple wizards -- one ordinary, and the other using "future" value, i.e. shifted one step ahead. Put them on BP -- that would be visual.
You don't want to see the problem caused by the KP used... But this is very important.
Look: each output point of your filter is calculated taking into account not only the previous values, but also capturing, with respect to the calculation period, the future value. That is why it lays on BP so nicely. But this future value is not available at the end point of the series - and such a "future" point is necessary for calculating the KP filter's output. How to be in this case? - think the creators of this miracle. I suspect, that they took the path of least trouble, and in calculation of an end point they use the last actual value of BP as the "future" value. i.e. they set X(t+1):=X(t)
To see the problem more clearly, do calculations of two simple wizards -- one ordinary, and the other using "future" value, i.e. shifted one step ahead. Put them on BP -- it'll be visual.
and captures, in relation to the calculation period, the future value
How do you put a future value in a formula?
Leave HP alone. The filter has been widely used in economics for years, no one has noticed and it's just you. Doesn't that bother you? It just makes me tense.
Once again. I only need the filter to get a quality residue. Nothing more than that. The quality of the forecast is buried in the residual after removing the trend. At the moment the last 4 candles are taken. Forecast. Shift. 4 candles. Forecast.....
How is the "low quality" HP related to the forecast? Through a substandard residual?
Well... keep fighting him...
By the way, on the subject of
... применяется в экономике много лет, никто не заметил и только ты
this problem has been around for engineers for a long time but economists don't listen to it and have been fiddling with it for years :)))
This problem has been known to technicians for a long time, but economists do not listen to it and have been dealing with it for years :)))
I am not interested in technicians' opinions on economics.
Let's focus on the state space.
I'm not interested in technicians' opinions on economics.
Let's focus on the state space.
I'm not interested in technicians' opinions on economics.
Let's focus on the state space.
What are you talking about -- economics... economics...
Bullshit! Where's the economics here.... You handle BP -- that's it! Don't you get it yet?
You shouldn't have done that... if you don't want to hear the same thing.
It's not for nothing. It's a worldview. That's what they taught me at the institute. Everyone has their own vegetable garden. That's what I wrote about in your post.
Show me the state space. We'll discuss.