Machine learning in trading: theory, models, practice and algo-trading - page 622
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
Very much so). I have an intraday.
If you take a day or more, you don't need either glass or ribbon at all. For 10 trades a day there is nothing without it.
I don't know how to do it on Forex.)
You can't, Yuri. Because the probability density functions are such that you have to work in Forex with more than 10 000 sample volumes. In fact - on diaries. Distribution formed by increments is stable and infinitely divisible. That's why many people, when they see the market fractality, rush to trade on minutes and leave the market poorer than a church mouse.
And on the daily market - yes, there won't be many deals, although your method is absolutely correct, I finally got sure of it today (do you understand what I'm saying? - I'm not going to tell you about it in details.)
NEVER, Yuri. For the probability density functions are such that you have to work in Forex with sample volumes over 10,000. In fact - on diaries. Distribution which is formed by increments is stable and infinitely divisible. That's why many people, when they see the market fractality, rush to trade on minutes and leave the market poorer than a church mouse.
And on the daily market - yes, there won't be many deals, although your method is absolutely correct, I finally got sure of it today (do you understand what I'm saying? - I'm not going to tell you about it in details - you may as well do it yourself).
I, so short term (days and more) is a dead end. And there is even a poorly formulated justification for this).
I use only minutes, and only intraday interspersed with scalping.
This is for the stock market. I cannot say anything clear about Forex. But people successfully pipsing, it seems. This is my experience in scalping strategy on Forex.
I wonder why not Python? Probably the same R? I don't get it.
I will not speak for Michael, he is able to argue his choice, but I can guess probably because python and R interpreting high-level languages, they are like matlab or "math" rather an interface to the set of libraries, like a command line than the languages themselves, for the familiarization is very cool, but in production when you need to fight with the exclusive algorithms, no chance, it is like a scooter to beat formula 1
Indeed. Since there is such a craving for knowledge, you take the ranking and from the top study what seems to fit.
One thing is the number of student publications on githab and articles by novice scientists, another thing is on what the machine learning algorithms are coded by serious investment firms (banks, hedge funds, etc.), the sample of "top" is biased towards the former, as the latter are not very advertise what and how they do, except at the top or even as a distraction and lead to an elegant dead end.
I won't speak for Michael, he is able to argue his choice, but I can guess probably because python and R are interpreting high-level languages, they are like matlab or "math" rather an interface to the set of libraries, like a command line, than the languages themselves, for the familiarization is very cool, but in production when you need to fight with exclusive algorithms, no chance, it's like winning formula 1 on a scooter
I forget, isn't it about Jave? Well, Java is also interpreted language which isn't compiled to machine code and runs on JVM.
Python, like Java, is also compiled and, similarly, runs on a virtual machine. Python has an advantage in that it's like a scripting language, all the libraries are already in compiled C++. And it's the vast libraries that are valuable in Python, not the Python itself. R is also a scripting language and all the main work is done by packages in compiled C++.
In general, Java is quite slow. It's not at all like Formula 1). I don't know what he's into it). To me, if the strategy is not scalping pips, then there is no need to beat anyone, the performance does not play a role.
As for Michael, it is his choice. We are sort of discussing languages).
I've seen somewhere a comparison of performance of MO algorithms for different languages. Java and Python are close to each other.
And now can I get some links or explanations? :) I can do it for forex
I am interested in concrete results, not assumptions.
I've done everything I could find on the Internet about the trading of quote tools.
I should look for VAR, VECM, the vars package and, as always, the links in it. Entry Boundaries on SETAR
Hooked it up, but google for help - literature and specific applications ....
If you get over the spread, maybe share. I can't take my eyes away from such a beauty without spread.
Look for VAR, VECM, the vars package, and in it, as always, references. Entry Boundaries by SETAR
Hooked it up, but google to help - literature and specific applications ....
If you get over the spread, maybe share. Without the spread, it's just so beautiful, you can't take your eyes off it.
Thanks, I'll take a look. Something strange - spread for pair trading seems to have never been an issue
AAA... it's autoregressive+multivariate regression... sort of... interesting, you can do it yourself. I'll get into it later.
I, so short-term (days and more) consider a dead end option. And there is even a poorly formulated justification for this).
Only minutes, and only intraday interspersed with scalping.
This is for the stock market. I cannot say anything clear about Forex. But people successfully pipsing, it seems. And, as far as I know, most of the working strategies are scalping or close methods both on stock and forex.
The right chart is "memory". It only disappears in increments >=17 pips. And the confidence interval for +-17 pips is about 99.5% of this distribution or 28.900 ticks. This is the volume at which it should be traded. All other cases are just special cases of this giant distribution and lead to a loss.
Once again look at the distribution of increments for the USDJPY:
The right graph is "memory". It disappears only in increments >=17 pips. And the confidence interval for +-17 pips is about 99.5% of this distribution or 28.900 ticks. This is the volume at which it should be traded. All other cases are just special cases of this huge distribution and lead to the defeat.
I will not speak for Michael, he is able to argue his choice himself, but I can assume probably because python and R interpreting high-level languages, they are like matlab or "math" rather an interface to a set of libraries, like a command line, than the languages themselves, for familiarization is very cool, but in production, when you need to fight with exclusive algorithms, no chance, it is like a scooter to beat formula 1
One thing is the number of student publications on githab and beginner scientists' articles, another thing is what serious investment firms (banks, hedge funds, etc.) use to code their machine learning algorithms; the "top" selection is biased towards the former, since the latter do not advertise what and how they do it, except at the top or even to distract and lead to an elegant dead end.
You always have to look at the root, as the classics taught us.
And the speed of R is a big question.
The interpreter is on the surface. A classy thing when researching. In my experience a debugger is not needed at all. But the speed seems to want better. But it's only appearance - let's get to the bottom of it.
Speed nuances.
1. If you take any computationally intensive package R, it is written in R only to access the libraries, which are used Fortran and C libraries. The choice was made for maximum speed and it is doubtful that this can be surpassed.
2. Further, for those transitioning from C. R is a matrix language that lacks the concept of scalar. The Intel library is behind vector and matrix operations. Besides, the code in R itself is very capacious because of this.
3. Loading all the cores and processors of the computer is the norm with the corresponding toolkit.
4. Developed in R works perfectly on the "server-client" scheme.
5. You can write a very stable program in R which is capable of handling any exceptional condition. The notion of na standard with the corresponding tools.
6. R and Cpp get along very well, the interaction is well documented, so if you manage to write a large program in R (which is very difficult because it is full of ready-made code) you can always rewrite all or some narrow parts in Cpp