From theory to practice - page 66
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
Welcome to reality, happy baptism to you :)
It turns out that you have to work with ticks - you can't get away from them. There are archives for them, etc. etc. And this difference in tick flows is what makes us all rivals, not friends. So that's it... I see...
There's something wrong with you. If you are working with ticks and need 5000 then with M1 it should be 80-240 bars.
You're catching the same readings just on generalisation.
Yes, it will be rougher, the high-frequency component will go away, but not so much as to splash out the essence.
If the market changes the nature of the movement, from randomly wandering in a corridor to a unidirectional spike, your system should identify it.
5000 ticks is half an hour and you have one or two trades a day on this window. The same two deals per day should be obtained on M1, because at this level of market watching, the market gives you the opportunity to make 1-2 deals a day. If more often it will be pipsing or scalping.
... People will be looking longingly at the monitor waiting for 1 deal a month... No - that's not our way... And working with OPEN turns Forex into a boredom... No, this method of data reception is not suitable! Boring!
...Conclusion - it's impossible to come up with one concept as a team! We speak different tick languages. That's the Forex uncertainty (see Heisenberg's uncertainty principle)... Now it's clear to me.
The problem is surmountable. If you want to make all traders on this forum happy with your solution, there is a solution. The model is developed and traded in a benchmark brokerage company. If the model shows a stable positive yield and if the IRR is at least 3-5 times the spread, then:
The bottom line, Nikolai, is this - read https://ru.wikipedia.org/wiki/Принцип_неопределённости carefully.
This is the most fundamental law and this is what we encountered in Forex. It was only tonight that I realized this.
We cannot accurately measure the state of price at a particular point in time. Different tick flows. Well, it just catches my eye and why didn't I think of it before???
But, it doesn't mean that we should give up - we need to work with large sample sizes and that's it.
And what happens - take 5,000 OPEN values - it's exactly 5,000 concrete values, not some abstract population. How much is that in hours? About 4 days! That is, we work (I think that's how the big banks do it) in general on days. But the banks see the large amplitudes, and have nowhere to rush, and with their market invasions they literally destroy small traders who work on small timeframes ...
How to fight?
Well, everything I've described in this thread is true. We need to deal with probability theory and mathematical statistics with setting stop-losses.
But, here, turns out very unpleasant thing - EVERYBODY IS FOR HIMself because of Heisenberg's uncertainty principle.
The maximum that can be done - this is within one DC, a group of comrades combine their resources into one and also crush small traders with big lots.
Holy shit... I have to think...
Market crashes, trader in terror runs to analyst...
T- what happened?
Calm down, I'll explain everything.
T- I can explain everything, just tell me what is going on.
I can explain how banks withdraw money from the market, and I helped you because you tried to analyze the market using statistical methods (which I am not good at).
I have long ago posted that the market has algorithmic rather than statistical laws, this is IMHO, but I'm not going to interfere in your quest.
Well, I think the man knows what he's doing, so why bother.
And by the way, since you need a closed system where all participants are defined and the price is also defined, you should go to the stock exchange.
There are many of them, choose whichever one you like, either Russian or bourgeois.
There's a tape of deals and a market of prices and open interest.
The bottom line, Nikolai, is this - read https://ru.wikipedia.org/wiki/Принцип_неопределённости carefully.
This is the most fundamental law and this is what we encountered in Forex. I realized this only tonight.
We cannot accurately measure the state of the price at a certain point in time. Different tick flows. Well, it just catches my eye and how did I not think of this before???
Alexander, forex is an over-the-counter market, so there are no unified benchmark quotes. But no one is forcing us to use it as a source of quotes. Let's move on to currency futures - they are traded on the exchange, there is a single quote for everyone at any given time, regardless of the dealer and woe to the dealer who tries to change it even one iota. There is also a unified database of currency (and any other) futures quotes. Your model can be developed based on futures quotes, and trading signals can be executed on the same futures, as well as on forex. There is a tight linear relationship between the price of currency futures and the forex price.
However, there is a small problem here - stock quotes are paid, but to build and test the model you can take a demo, where quotes have a small delay. I see the point in minimizing labour and time costs and first making sure that your model works with at least one broker on one currency pair. This is the most important and difficult task. Everything else is a technical matter.
Have I moved off topic? I need to think about it. After all, if you start working with OPEN for me it is boring - by the nature of my character I am bored to work with this value. And to work with tick-flow - do you remember how you yourself described difficulties of processing archived tick-flow data? Can you repeat it again? Maybe I have not understood something.
I'll duplicate it.
In order to organize a RELIABLE system of ticks saving/reading (as well as the history of a cup, the person asked in one of threads), it is necessary
to write software to save ticks which are broadcasted by DTs, there are options here...
It should be noted that these ways are implemented quite simply in MT5, but poorly in MT4. For MT4, the ticks collector is suitable, directly from the market environment broadcasted by brokerage company. It is unreliable, but in principle it is possible.
Further ... All this infrastructure should be deployed on a rented VPS, and there we should launch the data compression system.
And the trading client connects to the UPU, takes the accumulated data in the right quantities and works with it.
But all of the above is necessary for a battle bot working with real money.
Investigations can also be carried out on history pulled from somewhere.
So not to pick up all this hassle, enough to go to M1 history on which there is at any time and in very large quantities. I went through several brokerage companies yesterday (I chose a pallet) and they all have M1 history about 2 km. I looked at 1.7 and 2.5, but on average I got two million bars. I do not know what to do with them.
This is approximately 2011-2013 years, for five years and the market has already changed ten times, and the patterns have changed a hundred times, so for the implementation of the combat bot history of M1 is enough. IMHO
Both in theory and in practice, this leads to this. If you work with a sample of 500, it is about 96% of the Laplace distribution. But here's the trouble - it's also about 96% of the t2-distribution, and 96% of the Cauchy distribution... We don't see CROWNS!!! And that's what we have to work with! We need large sample sizes. And working with OPEN turns Forex into a bore... No, this method of data reception is not suitable! Boring!
I originally wanted to describe a universal algorithm for all. But, precisely because of this difference in tick flows, it is almost impossible to do so.
The conclusion is that it is impossible for one team to come up with one concept! We speak different ticking languages. That is the uncertainty of Forex (see Heisenberg uncertainty principle)... Now it makes sense to me.
If trades last for several hours, the difference in units of pips at different brokers becomes absolutely insignificant. It's not like when you measure the length of two bricks, you don't have to do it to the nearest nanometer.
And the difference for brokers is only in the frequency of ticks, it is not a problem at all. Well read the ticks in increments of a few seconds and that difference will disappear even at the tick level, let alone the level of a few hours.
And the price value at the same moment at different brokers is the same with the accuracy of the spread, if it were different - arbitrageurs would go broke brokers. I measured it personally)
You have made up some mythical problem and elevated it to an absolute and insurmountable. In fact there is no problem at all. Have a think about it at your leisure.
The question is, how often do you have transactions taking place? According to my calculations, it should not happen more than 1-2 times a day. Am I right?