From theory to practice - page 884
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
In fact, the TS is already established and if it doesn't produce a result in the contest, I will be extremely surprised and disappointed.
However, for that case, there is an alternative version of another theory - Gann or Elliott or whatever. I would be happy to have separate threads devoted to other market theories.
Alas, no one has the courage to run such branches... Very sorry about that....
In fact, the TC has already been set up and if it doesn't produce a result in the competition, I will be extremely surprised and disappointed.
It can't be because it can never be. (с) ))
Alexander wants to create a system:
In what specific time will Person A cover a distance of 500 metres?
Known numbers:
Statistically, 100 people have traversed between 4 and 10 minutes, 10+4/2=7, average time 7 minutes.
The statistics have been gathered in the past.
Let's run a real test with person A and thus be sure that he will walk in 7 minutes.
Conclusion: All calculations and statistics have nothing to do with reality, maybe as soon as he left the starting point, it immediately started raining.
As for the market, it is the same: big money came into the market today and did not come tomorrow. There are many reasons, maybe someone even coughs and does not trade, or someone had a bad dream and lost his position.
You didn't specify the parameters of person A. Who is an athlete, or maybe a grandmother of 80. And there are many other conditions, so it's not a good example.
You did not specify the parameters of person A. Who is an athlete, or maybe a grandmother of 80 years. Yes and many other conditions, so the example is not good.
It's all summed up by this:
Forum on trading, automated trading systems and trading strategy testing
From Theory to Practice
Vitaly Muzichenko, 2019.01.29 17:55
Alexander wants to create a system:
In what specific time will person A cover a distance of 500 meters?
Known numbers:
Statistically, 100 people have traversed from 4 to 10 minutes, 10+4/2=7, average time 7 minutes.
Statistics collected in the past.
Let's run a real test with person A and thereby make sure that he will pass in 7 minutes.
Conclusion: All calculations and statistics have nothing to do with reality, maybe as soon as he left the starting point, it immediately started raining.
As for the market, it is the same: big money came into the market today and did not come tomorrow. There are many reasons, maybe someone even coughs and does not trade, or someone had a bad dream and lost his position.
Here's the interesting thing about the sliding window of observation. Because of the different approach, you get different data.
1. You can take a sliding window W = n and there will be a shift at each step.
2. You can take a reference point and add n at each step.
You can take a reference point shifted to the beginning of yesterday and add n at each step, i.e. at the beginning of the day W = 1440 minutes, and by 12:00 W = 2160 minutes. This results in a dynamic sliding window.
In point 3 lies the riddle and the answer at the same time, which explains the signal shift to the left.
I used to think the signal lag was due to averaging.
Oh noooo... that's not it.
;)
You were saying something above about windows and forever with them...
Well, your windows and statistics are no way in hell to analyse real signals. And even on history.
There is a great deal of literature on windows and signal analysis. The Internet is large, you can find it.)
In fact, it is not the first time I've written it).
Yuri, so how do I make a predictor so that it doesn't change its properties over time? You can't get away with polynomial regressions.
Because the forum is deserted, only rabid newcomers pile up from topic to topic
The most interesting thing is that there is a movement of a certain type in the market, which repeats explicitly or not so much from time to time and the most interesting thing is that it always happens regardless of the year of the month or the quote. the only conclusion I came to is that this movement is the only way for the instrument to dump such a number of speculators to remain random and at the same time be trending in a certain sense.
And the movement is a reverse correction relative to the global trend and further continuation of the movement. Of course it is strange, but it happens 70-85% of the time. And even if it is flat, the risk is minimal.
But to detect it programmatically...))
Here you can solve the problem of objective analysis and search for ways to detect a trend in time, and so on).
Comrade Che! Marxist, poet...
I read you and my soul is glad - you understand and know a lot. However, I am puzzled - you said yourself that you widely use both entropy and the Hirst coefficient. And about asymmetry and kurtosis of distribution of increments you cut a chip. You should have no questions - what motion is random and what is not.
What makes it so?
Yuri, how do you prepare a perdictor so that it doesn't change its properties over time? You can't get away with polynomial regressions.
Because the forum is deserted, only rabid newcomers rushing from topic to topic.
There are no predictors, they don't exist in nature.) To be clear, one predictor means exactly nothing - an empty shell. Any search for meaningful ones on their own - it's all empty. It is like finding something on a plane or in a volume, running along the x-axis).
It is not excluded, but I don't claim that a pair of predictors is also empty.
To really find something with non-zero probability one needs a set of N orthogonal, or at least linearly independent, predictors.
That is, before we build predictors, we need to decide with N-dimensional space, in which we will build it all. I can't help you with this - it's too big a topic.)
And there are no predictors, they don't exist in nature). To be clear, one predictor means absolutely nothing - empty space. Any search for meaningful ones on their own is nothing.
It is not excluded, but I do not claim that a pair of predictors is also empty.
In order to really find something with non-zero probability one needs a set of N orthogonal, or at least linearly independent, predictors.
That is, before we build predictors, we need to decide with N-dimensional space, in which we will build it all. I can't help you with this - it's too big a topic.)
Of course we mean a predictor in Hilbert space, not some one-dimensional crawler
Comrade Che! Marxist, poet...
I read you and my soul rejoices - you understand and know a lot. However, I'm puzzled - you said yourself that you widely use both entropy and the Hurst coefficient. And about asymmetry and kurtosis of distribution of increments you cut a chip. You should not have questions - which movement is random and which is not.
Why?
The man is studying the price movement.
I searched through several previous pages, but did not find this post.
I will say the following.
1. The pair may flop when everything is normal there, i.e., those suffering a profit, will not get it, or someone is around the spread before the profit.
2. 2. The pair may start trending (or may make a spike) with decent risk for profit seekers and with a guarantee of increasing risk (someone is building a network).