From theory to practice - page 1138
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 remember all the information so the distribution view
it's kind of like
- Erlang.
- Logistic.
- Laplace.
but I don't remember exactly...
I need it now...
and formulas if you can)
Well, if you're interested in isovershaped distributions - something similar to a logistic distribution at the centre.
https://en.wikipedia.org/wiki/Logistic_distribution
but with the tails trimmed :)))
It can be approximated by a triangular distribution.
It's only a matter of time before we get to cosines, arctangens (aka fischer) and logarithms. The question is, when?
The flow of ticks is already a natural primary non-linearity, smooth it out smma(for example) with a fixed period of 2/3(200-300) of the average number per minute, fix the result by minutes and enjoy(analyse). Tics here tuzhe someone has collected, and mt5 already contains tics.
I am NOT interested in ticks. For one simple reason: their quantity differs from broker to broker. Same as OPEN M1, M5,... because of the inappropriateness of uniform sampling in the market.
I am interested in EVENTS, i.e. quotation flow synchronized between different brokerage companies. This can be observed on the second TF.
It's only a matter of time before we get to cosines, arctangens (aka fischer) and logarithms. The question is, when?
The flow of ticks is already a natural primary non-linearity, smooth it out smma(for example) with a fixed period of 2/3(200-300) of the average number per minute, fix the result by minutes and enjoy(analyse). Somebody has collected ticks here too, and mt5 already contains ticks.
I would put it this way - the flow of events is already a natural primary non-linearity. However, I disagree with the further text. This is exactly where you don't need to average anything, but just use this non-linearity directly. I do this when calculating the variance of a process.
I am NOT interested in tics. For one simple reason: different brokers have different numbers of them. As well as OPEN M1, M5,... because of the inappropriateness of uniform sampling in the market.
I am interested in EVENTS, i.e. quotation flow synchronized between different brokerage companies. This can be observed on the seconds TF.
From tick streams you will get 1 smoothed value per minute, better than m1. In daytime with ~8 gmt most (from)large +/- the same. Between different DCs by seconds you can not synchronize due to signal transmission delays (light speed, delays in equipment) - the order of 50-100ms from moscow to europe + delays in servers and terminal. The m5 after hours is enough for me.
I would put it this way - the flow of events is already a natural primary non-linearity. However, I disagree with the following text. This is where you don't need to average anything, but simply use this non-linearity directly. I do this when calculating the variance of the process.
You have the kernel itself - MA - linear, as you said sliding window.
So your probability of finding a non-random event is equal to the probability of indexing a non-linear process by a sliding window, i.e. 50/50.
Your kernel itself - MA - is linear, as you said sliding window.
So your probability of finding a non-random event is equal to the probability of indexing a non-linear process by a sliding window, i.e. 50/50.
No. My sliding window is now non-linear in time - equal to a certain number of events (but not ticks!!!). However, along with the number of ticks, the normal linear time is also involved in the variance calculation. And we will look at the result at the end of the coming week.
Actually, at some point in time I wondered - this is what exactly (as Bas likes to put it) distinguishes real VR from an artificial random process like SB, Brownian motion, etc.
Yes, the time intervals between events!!! If in SB these are even single points of reference N, N+1,... i.e. a conditional number of steps, and the Wiener process is a model of a continuous process which corresponds to chaotic Brownian motion, when collisions between particles occur in infinitely small intervals of time and in this case the uniform discretisation is valid, then the time intervals between events have a magical meaning of their own in the market.
And not taking this non-linearity in time into account is the worst mistake in all calculations.
No. My sliding window is now non-linear in time - equal to a certain number of events (but not ticks!!!). However, along with the number of ticks, the normal linear time is also involved in the variance calculation. And we will look at the result at the end of the coming week.
Actually, at some point in time I wondered - this is what exactly (as Bas likes to put it) distinguishes real VR from an artificial random process like SB, Brownian motion, etc.
Yes, the time intervals between events!!! If in SB these are even single points of reference N, N+1,... i.e. a conditional number of steps, and the Wiener process is a model of a continuous process which corresponds to chaotic Brownian motion, when collisions between particles occur in infinitely small intervals of time and in this case the uniform discretisation is valid, then the time intervals between events have a magical meaning of their own in the market.
And not taking this non-linearity in time into account is the worst mistake in all calculations.
Get rid of the sliding window altogether.
Unless, of course, you want to get results.
You have no idea how hard it would be to analyse the market if there were no robots working in the markets.
Now it is 100500 times easier to analyse than 20 years ago.
I'll show you an example with standard EURJPY settings from several TFs. I've been reproached that I select variants for pretty pictures. But it is not so.
Here is the H4 levels aligned by the robots.
If we look closer at H1
M30 has its own order of robot behaviour
To be confident in walking with the market we need to understand the behaviour of market robots. And then you will have a grail.
Actually, at a certain point in time I wondered - this is what exactly (as Bas likes to put it) distinguishes real BP from an artificial random process like SB, Brownian motion etc.
Yes, the time intervals between events!!! If in SB these are even single points of reference N, N+1,... i.e. a conditional number of steps, and the Wiener process is a model of a continuous process which corresponds to chaotic Brownian motion, when collisions between particles occur in infinitely small intervals of time and in this case the uniform discretisation is valid, then the time intervals between events have their own, magical meaning in the market.
And not taking this non-linearity in time into account is a terrible mistake in all calculations.
In principle, nothing. SB exists in a closed system - everything is equilibrium, homogeneous and decent. In an open (non-closed) or non-equilibrium system, classical SB cannot exist. Our as-is SB is by definition in an open system.
As for continuity, replace the process with discrete observations and the SB remains in place and nothing changes in the Wiener model. By the way, in a discontinuous environment, SB events will not occur at equal intervals, and nothing will be infinitesimal. There will be something like shot noise, which was also studied by Wiener and is of the same nature as SB.