MACHINE period with minus value - page 27
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
:-)
I posted the indicators in the codebase last year... and it's still there... :-)
All red tape and "everyone's gone to" mcl5. :-)
And then there's the hang-up to check topics... :-) Especially since you never know what can grow out of even such a theme.... It may be something really naughty, but it is possible...
Like, I wanted a MA with a minus, ended up with a Kalman filter and am overjoyed... got the parameters right, made some dough! Why not...
I doubt it. There's no filter, just an image.
The main thing is that the start has been made...
There's no way...
The main thing is that the beginning has been made...
There, it doesn't matter...
The beginning was made more than once in the previous years and continued in the new one, but with substantiated and interesting data:
https://www.mql5.com/ru/forum/124401 Sorry, I don't know how to make links yet! How easy it is to do that!
Caesar, is that the kind of waving you want?
No, I don't. A line drawn at closing prices is just as good. How do you trade with that?
No, don't. A line drawn at closing prices is just as good. How do you trade that?
He already replied to my comment on it:
alsu 07.01.2013 00:21
Caesar is in a ban again, he will not answer you now. But we understand very well that it is impossible to open a position successfully without determining the trend. This Masha with period 1 or 2 will give the same result, if the entry condition is made by the last two Open() or Close().
Like, I wanted a MA with a minus, ended up with a Kalman filter and I'm happy about it... ...and I'm happy about it, I've got the parameters, I've made money! Why not...
No, don't. A line drawn at closing prices is just as good. How do you trade with it?
It's the simplest variant, you can add noise preprocessing and get a smooth and lag-free (almost all matrixes are good) picture.
For example: blue one is SMA(5), degree of smoothing is comparable, but lag is much bigger. What am I talking about, it's long known stuff, right?
Well, how to trade it is a tenth of a question:)
1. Kalman doesn't have to opt for anything, it opts for itself))))
2. so this is the simplest variant, you can add noise preprocessing, so you get a smoother and lag-free (almost, where matrices are good) mashup.
For example: blue one is SMA(5), degree of smoothing is comparable, but lag is much bigger. What am I talking about, it's long known stuff, right?1. oh so much the more!!! :-)
2. would you be so kind as to take a look, and in your IMHO opinion, offer these matrices...
You can also post the code... and pictures of how to use this filter...
Not familiar with Kalman, haven't googled the subject yet... Things not yet known, in general.
1. oh! all the more so!!! :-)
2. be kind enough to have a look, and for your IMHO, suggest these matrices...
You can also post the code... and pictures of how to use this filter...
Not familiar with Kalman, haven't googled the subject yet... Things not yet known, in general.
About Kalman here is very clear, with an example. The code... it's almost all in my DLL... in addition, the rascal itself is just by the given formulas (yes, in fact, exactly the same as in the figure in the article), nothing complicated. Alglib has a nice matrix library, which saves most of the fuss.
2. Once again, the filter calculation itself is not a big deal; the only difficulty is generating correct matrices F and B (in the notation as shown here). This work is 90% theoretical since it implies formulating the fundamental model and translating it into the language of matrices. By and large, we initially know nothing about either matrix F (but that is half the trouble) or matrix B. The second is much more serious - one should not only speculate HOW the price moves, but also what is the reason for the movement itself, and even guess statistical characteristics and form of external action.
I'm sorry, I won't lay out the code of matrix fitting, but the idea please (and I've done it here more than once). In the used model the market is presented as a quasi-linear system of 4-6 orders in time and 2-3 in steering impact, "quasi-linear" means that in each moment we approximately consider the model linear, but its parameters are estimated anew each time. In Kalman's language, this means that the matrices F and B are time dependent. The control action is considered a complex Poisson flow ( see Tikhonov V.I. and Mironov M.A., Markov Processes (1977), p.421). To make it stand out from the noise, statistical distinctions between the noise and the actual intended control are used(maximum likelihood method).
Well, that's pretty much it... I leave the formulas and rationale for assumptions and calculations to those who suffers as a term paper)))