Not Mashka's business! - page 9
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
Further, when I ask you to provide a forecast one bar ahead, you should be aware that for you it will result in a 1 bar forecast + the value of your MA's group lag.
What does lag have to do with it? Neither Burg's method, NS nor your method knows anything about MA lag. For these methods it's just BP and nothing more. And then if you accurately calculate the future value of MA, you will also accurately calculate the future price value. What does this have to do with any lag??????
You know. And that should be enough!
Suppose we have an LPF with an averaging window of N bars. We look for the current MA value using the scheme s[j]=SUM{Open[j-i]}/N, where i runs from 0 to N. In this formulation, the group delay is N/2. So if you want to look ahead one bar of the price series you should compensate the lag and make a forecast vector N/2+1 bars ahead of the current MA value. That would be the prediction bar forward (relative to the price series).
Further, when I ask you to provide a forecast one bar ahead, you should be aware that for you it will result in a 1 bar forecast + the group lag value of your MA.
What does lag have to do with it? Neither Burg's method, NS nor your method knows anything about MA lag. For these methods it's just BP and nothing more. And then if you accurately calculate the future value of MA, you will also accurately calculate the future price value. What does this have to do with any lag??????
You know. And that should be enough!
Suppose we have an LPF with an averaging window of N bars. We look for the current MA value using the scheme s[j]=SUM{Open[j-i]}/N, where i runs from 0 to N. In this formulation, the group delay is N/2. So if you want to look ahead one bar of the price series you should compensate the lag and make a forecast vector N/2+1 bars ahead of the current MA value. That will be the forecast for the bar ahead (relative to the price series).
Seryoga, either I don't understand something or you've made a mess of the solution. Frankly speaking, I did not understand much about the diffraction and its connection to the possibility of forecasting (I hadn't found such a relation in any source), and I absolutely did not understand anything about the N/2+1 vector :o)
I guess that maybe my poor knowledge of mathematics doesn't help me to understand the problem in depth, but if it's about forecastingprice 1 bar ahead using lagged MA, I'm doing it very simply. Instead of tedious explanations I give a picture (CONNECTED):
You need to go from MA to price and it's easy, the picture shows everything and the solution comes down to an elementary equation. Similarly, moving into the future, you can reconstruct the series.
Seryoga, I'm definitely dumb, why do you need this N / 2 1 vector? Yes, and do not forget to praise the highly artistic creativity :o)
By the way, Seryoga, how did you simply determine that the price range is all over the place? Do you have proof of this humble fact?
I have nowhere claimed that the price range is differentiable.
Forget everything I said about N/2+1. As you correctly and VERY nicely portrayed, we have a generated Close and we need to make a prediction for the next one, and it doesn't matter which way or which algorithm is involved.
Explain that there is a Y in the bottom figure? And why is it included in the sum. Or Y is a price series and the correct way is:
(1/N)*(x2+x3+...+xN+Y[i-1])=M
Y[i]=M*N-(x2+x3+...+xN)
Is it like this?
I have nowhere claimed that the price range is differentiable.
Forget everything I said about N/2+1. As you correctly and VERY nicely portrayed, we have a generated Close and we need to make a prediction for the next one, and it doesn't matter which way or which algorithm is involved.
Explain that there is a Y in the bottom figure? And why is it included in the sum. Or Y is a price series and the correct way is:
(1/N)*(x2+x3+...+xN+Y[i-1])=M
Y[i]=M*N-(x2+x3+...+xN)
Is it like this?
Not so. I guess I will not become Leo Tolstoy or at least one of Kukrynikovs. And my expressive and artistic wordplay is clearly lame :o( The value of "Y" symbolizes the future price (for the bar ahead) that we're looking for, the vertical bold line is now. I assumed that by MA we mean the average, I think that's what we wrote earlier:
For the current moment (for now) and for the past we know these mu's. By predicting a chain of these mu's (more literate to say MA with some fixed window N) we take their historical values and predict the future (by the way, if we predict price, only this chain, everything else simply cannot be predicted). Now look at the formula above - you have a known left term, you don't know only one value of the sum on the right, it's future price value.
Remember, we have a fixed window, so we took this window and shifted it to the countdown ahead and calculated the forecast average for this shifted window. The definition of a predicted average is no different from a "non-predicted" average - it is the sum of the series divided by its length. But for a forecast mu it is the sum of the known series except for one price value (future value), which should appear in the future, but which would "not break" the equality for the forecast mu. In other words - what would be the future price so that the average of a series of length N equals the predicted mu, this is it:
Seryoga, I don't know how else to explain it, it's a very trivial thing. But if you don't understand it, I'll explain more, I'm a famous explainer :o)
Ahhhh! Now you're going to be in my shoes too - you're going to explain.
Now, I'll read again what you're trying to say and ask a question or two :-)
Um... so far I'm just sensing that you think you're very perceptive...
one more time.
So, it hasn't been half an hour... Question: how do you predict the window average for the countdown ahead?
1. it is early to the previous value;
2. first option + linear term;
3. second option + quadratic;
4. something else.
Bingo! Played and guessed it all!!! YOU'RE KIDDING ME????
Так, не прошло и получаса... Вопросик: а как ты прогнозиоуешь среднее по окну на отсчёт вперёд? варианты ответов:
1. it's early to the previous value;
2. first option + linear term;
3. second variant + quadratic;
4. something else.
"Serega, I have been telling you about it from the very beginning - I make MA forecasts by Bourg's method . I told you how I do it, I have drawn charts and explained it in detail - just read it once again. Either you or I'm getting the hang of it, or you've got it all figured out and now you're just mocking me. Where did I write about it, point the finger?
Hmmm... so far I've only sensed that you think you're very astute...
I thought you would understand what I am writing about. Seryoga, I try to assess myself as objectively as possible, although there are kinks. You're kidding me!!! Fine, fine.
Neutron wrote (a):This is a serious philosophical question. Seryoga, it would take a lot of beer and time, gathered in one place with us, for me to answer it.
A friend of mine called a friend, a former classmate. They talked for about 20 minutes, and then one thing led to another, it turned out that the phone number had been dialled wrong...:)
(it's not a story, it's true).
I thought you would understand what I was writing about. Seryoga, I try to assess myself as objectively as possible, although there are kinks. YOU'VE GOT TO BE KIDDING ME!!! All right, you're welcome.
Sergei, relax! - You're being taken off :-)-)
So, when are we going to evaluate the results of the forecast?
I thought you would understand what I was writing about. Seryoga, I try to assess myself as objectively as possible, although there are kinks. YOU'VE GOT TO BE KIDDING ME!!! All right, you're welcome.
Sergei, relax! - You're being taken off :-)-)
So, when are we going to evaluate the results of the forecast?
Roughly speaking, if running today, my suggestions are as follows:
Now (I mean quickly) I can give out:
Clarify, are you sure you want 1 bar? I'm going to count for a day for 1000 counts and I'd rather not waste my time. OK, I'll indicate the sign of the first predicted bar in the general "predictive mess", and then I'll select the data by it.
With that setup I can run the calculation overnight tonight or/and tomorrow.