From theory to practice. Part 2 - page 117
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 there was a maths thread here, it would be interesting to chat there.)
So let's do it, no big deal))
https://www.mql5.com/ru/forum/368720Thanks mate, that's interesting!
I understand that mostly fundamental analysis is used, a stream of fresh data. Technical analysis of chart history is not particularly considered. Have I got it right?
I have downloaded the brochure and I`m reading it.
I cannot cope with the search, I cannot cope with scribd, I do not like it since a long time.
Well yes, tell me that we should move to FPGA with collocation in the exchange )))
Have you counted the costs of such goodies? Cost of equipment, cost of direct feeds, cost of collocationetc.
Not to mention the cost of an FPGA specialist if you don't know anything about the technology yourself.
It's very expensive to maintain such a structure.
Of course, they are covered by the market, but the initial threshold of skills and technical support is very high.
Therefore, ordinary ordinary ordinary people are more or less understand where the fish is, using old tried and tested approaches.
Let it not huge profits, but it is there and stable. The point is not how you caught the fish, but that you caught it.
As far as satellite data is concerned, I studied this issue once. So the speed of this data is lower than Fibre optics.
Therefore satellite data is not suitable for HFT. For this reason, faster those on a short leg, or in the dark fiber.
On the alternative data, also thought about this subject, but did not look for an approach implementation, it is necessary to understand what we are looking for and from what.
This is a difficult task of data analysis, which again requires skilfulness and not small. How many people have them? I doubt it.
Although there is an article on hubra as an example, on alternative data for understanding the process.
FPGA, optical fibers , etc. is towards ultra-HFT, this area of activity has changed significantly over the past decade, to the detriment of the toxicity for markets , and what remains tightly occupied and you can not squeeze into it, and it is not interesting there, this is not about data and models with machine learning, and about hardware,routers and the speed of light)) To be honest, I've never met anyone who does it professionally, where I worked, the " HFT" is a couple hundred trades per instrument during a trading session, closing positions overnight . Colocation is the norm and is not expensive , pure technical experts, they are also not expensive, cheaper by about five quarters, the threshold of skills is high on average, but there are many organizations, there are not enough superstars for all, people are convertible, following the capital The people are moving with the funds and accordingly the size of the bonuses, with the growth of skills, places are freeing up for the less experienced but daring and clever ones .
The average philistine doesn't know where the fish are, that's the point. The average man in the street, if he is a fan of this craft and if he is lucky not to get stuck in many dead ends, after 5-7 years of trading and intensive following, absorbs a lot of information from a couple of university specialties in engineering and the humanities , stepping on almost all the rake and getting almost all the bumps, turning into a quantum . Quantum differs from an average man, because he trusts numbers, but not his own expectations. Quantum, for example, doesn't endlessly search through all indicators in the hope to find the Grail, with machine learning algorithms, with optimization of parameters in the backtester; these all things a person should understand himself in time, and not wait until someone writes about it officially in some book/article/blog.
There are many different traps when collecting, preparing data and analysis, impoverishing them is that the more accessible data, the less "fish" in them, well you can not get a good result on publicly available classic data such as prices themselves, and when they get "acurasi >60%" or the grail with SR3 it is an obvious sign that somewhere there is an error in the model or trading and analytical system , most likely in the preparation of data, the quantum will look for an error, the philistine plans to buy Lambo and no matter what acurasi completely diverges from the backtest, the quantum will rummage around for two months withdata preparation , will try many options and will find the error, the philistine will do the same for years without trying to understand why the same does not work, and then start selling these "strategies" in the near market and lead trading courses.
Alternative raw data from satellites, etc. is not about HFT, or rather not about ultra HFT, we are talking not about microseconds, but minutes to tens of minutes, the other thing is that resellers supply them refined but already expired,they are no more useful than prices. They have to buy crude themselves, and it's very expensive and not even every fund can come to an agreement, and analyzing them also requires a lot of work. For example you trade on grain futures, you get an image of agricultural fields every hour of 10-20% of the planet, from them you have to determine, first of all, the amount of various crops, degree of productivity, weather impact on the harvest and hundreds of other parameters and so on. This is the work of data analysts for man-year approximately, that is, about $ 150-200k per data stream, and the more data streams the more the costs and the data are very expensive, if you can even get them. Yes, it is expensive, but there are many cheaper data streams, and the most interesting ones are the ones that nobody has paid attention to yet)))
FPGA, fibre optics , etc. is towards ultra-HFT, this line of business has undergone significant changes in the last decade, to the detriment of the market toxicity , and what remains is tightly occupied and you can not squeeze in, and it's not interesting there, this is not about data and machine learning models, but about hardware,routers and light speed)) To be honest, I've never met anyone who does it professionally, where I worked, the " HFT" is a couple hundred trades per instrument during a trading session, closing positions overnight . Colocation is the norm and is not expensive , pure technical experts, they are also not expensive, cheaper by about five quarters, the threshold of skills is high on average, but there are many organizations, there are not enough superstars for all, people are convertible, following the capital The people are moving with the funds and accordingly the size of the bonuses, with the growth of skills, places are freeing up for the less experienced but daring and clever ones .
The average philistine doesn't know where the fish are, that's the point. The average man in the street, if he is a fan of this craft and if he is lucky not to get stuck in a lot of non-obvious dead ends, after 5-7 years of trading and intensive following he absorbs a lot of information of a couple of university specialties in engineering and the humanities , stepping on almost every rake and getting almost every bump, turning into a quantum . Quantum differs from an average man, because he trusts numbers, but not his own expectations. Quantum, for example, doesn't endlessly search through all indicators in the hope to find the Grail, with machine learning algorithms, with optimization of parameters in the backtester; these all things a person should understand himself in time, and not wait until someone writes about it officially in some book/article/blog.
There are many different traps when collecting, preparing data and analysis, impoverishing them is that the more accessible data, the less "fish" in them, well you can not get a good result on publicly available classic data such as prices themselves, and when they get "acurasi >60%" or the grail with SR3 it is an obvious sign that somewhere there is an error in the model or trading and analytical system , most likely in the preparation of data, the quantum will look for an error, the philistine plans to buy Lambo and no matter what acurasi completely diverges from the backtest, the quantum will rummage around for two months withdata preparation , will try many options and will find the error, the philistine will do the same for years without trying to understand why it does not work, and then start selling these "strategies" in the near market and lead trading courses.
Alternative raw data from satellites, etc. is not about HFT, or rather not about ultra HFT, we are talking not about microseconds, but minutes to tens of minutes, the other thing is that resellers supply them refined but already expired,they are no more useful than prices. They have to buy crude themselves, and it's very expensive and not even every fund can come to an agreement, and analyzing them requires a lot of work. For example you trade on grain futures, you get an image of agricultural fields every hour of 10-20% of the planet, from them you have to determine, first of all, the amount of various crops, degree of productivity, weather impact on the harvest and hundreds of other parameters and so on. This is the work of data analysts for man-year approximately, that is, about $ 150-200k per data stream, and the more data streams the more the costs and the data are very expensive, if you can even get them. Yes, it's expensive, but there are many cheaper data streams and the most interesting ones, that no one paid attention to)))
Yes ... I remember one thing about tick volatility) the thing is about time of modeling spikes with growing tick volumes, even on MT5 with "real" ticks the spikes happen in 2-3 seconds, on MT4 they happen in half a minute, and this, even 2-3 seconds is enough to earn about 1700000 points of profit on EURUSD for 10 years. In reality such emissions firstly occur in fractions of a second, secondly, the spread grows wildly, and thirdly, it is constant requotes. Using TickDataSuite I get a loss of 600000 points instead of profit of 1700000 points. Therefore I had to use a strategy, minimally depending on market noise, spread widening and working with opening prices, to maximally combine the results of tester and realset.
This kind of "problems" is a classic, when working with kitchens and low-quality data. But what conclusions would a reasonable person draw?
First of all not to work with kitchens, it is obvious, such horrors as "requotes", "re-quotes", all sorts of spikes and the like are just natural scams, and a reasonable person does not take part in such things. You can come up with hundreds of these kinds of scams to win a client, it makes no sense to gain such an experience. If you do not know the market from the scratch, you just register in half an hour, deposit 100 quid and start trading Eurobucks)). But it used to be like that, before the questions about your financial status and other "real exchange" in the KIS kitchens.
Secondly, again - DATA. The price series cannot be trusted from the kitchen server, it is clear, but unfortunately data from trusted sources, from the stock exchanges and brokers with high reputation, should be trusted with certain reservations, especially if the data are processed, such as candles, which can be filtered in many ways, even the orderlog in a feed backtester should be corrected by the time lag between the data you receive and where the orderlog is written . And that leads to one conclusion - you have to write the data yourself. Only self-written data from the correct sources can give a realistic picture of how the system would trade in real life.
Thirdly, the backtesteras the entire trading and analytical complex. For dating course MT-tester is good enough, but sooner or later will have to write your own backtester, it is incidentally not difficult, but very useful for brainwashing and appreciated at the interview, if you are aiming somewhere steeper than the "DC in your area. During backtesting and trading there should be no surprises, you should understand clearly at the level of elementary mathematics what' s going on If everything is accounted for, failures may occur only because of a severe force majeure, or "market surges" when the market crashes and so on. But not because of data and God forbid a scam like requotes/rejects.
These kinds of 'problems' are a classic, when dealing with kitchens and poor quality data. But what conclusions would a reasonable person draw?
First, not to work with kitchens, it is obvious, such horrors as "requotes", "redirects", all sorts of spikes and the like are just natural scams, and a reasonable person does not take part in such things. You can come up with hundreds of these kinds of scams to win a client, it makes no sense to gain such an experience. If you do not know the market from the scratch, you just register in half an hour, deposit 100 quid and start trading Eurobucks)). But it used to be like that, before the questions about your financial status and other "real exchange" in the KIS kitchens.
Secondly, again - DATA. The price series cannot be trusted from the kitchen server, it is clear, but unfortunately data from trusted sources, from the stock exchanges and brokers with high reputation, should be trusted with certain reservations, especially if the data are processed, such as candles, which can be filtered in many ways, even the orderlog in a feed backtester should be corrected by the time lag between the data you receive and where the orderlog is written . And that leads us to the conclusion - you have to write the data yourself. Only self-written data from the correct sources can give a realistic picture of how the system would trade in reality.
Thirdly, the backtester,as the entire trading and analytical complex later on. Of course for the acquaintance of course MT-tester is good, but sooner or later will have to write your own backtester, by the way it is not difficult, but very useful for a brain workout and appreciated at the interview, if you are aiming somewhere steeper than the "DC in your area. During backtesting and trading there should be no surprises, you should understand clearly at the level of elementary mathematics what' s going on If everything is accounted for, failures may occur only because of a severe force majeure, or "market surges" when the market crashes and so on. But not because of data and God forbid a scam like requotes/rejects.
For example you trade in grain futures, get an image of the agricultural fields every hour 10-20% of the planet
Can you give some more examples of the types of data used. Interesting!
You may say that the real exchange is not protected from lack of liquidity, and it is even quite normal for a real exchange. And the fact that they are skewing clients with hairpins, yes ... If you have some straightforwardness, you can do the same in an ordinary mt4, for example, re-compile a raw tick history into a file suitable for loading into MT4 and that's it. You may analyze and test as long as you like) But everything depends on your desire. You may start to analyze your own currency pairs and then you'll get the fx rate. We do not know how to use them, we just do not know how to react. But there is a chance to speed up the deposit from $100 to $ 1000-2500 using certain statistical tricks and cunning. True, there is a 100% risk of losing your money, because that is how the Forex market is set up, the two opposite processes have the same places to enter the market for earning on them. So you have to choose the lesser of two evils always)
No, there is no need to choose between the evils, you have got acquainted with the market, learned how to " speed up the deposit " in the slot machine, move on to the next stage, go to the local stock market, write your first trading and analytical infrastructure, then go to the top European and American exchanges, there adapt, then you will understand what to do. A more or less "mature" Forex, teenage I would say, where ECN liquidity is traded without a slump, but there is no sense to get into it with deposit less than 100$K. And adult Forex is an institutional subject, not for the average man.
No, there is no need to choose between the evils, you have got acquainted with the market, learned how to "speed up the deposit" in the slot machine, move on to the next stage, go to the local stock market, write your first trading and analytical infrastructure, then go to the top European and American exchanges, there adapt, then you will understand what to do. A more or less "mature" Forex, teenage I would say, where ECN liquidity is traded without a slump, but there is no sense to get into it with deposit less than 100$K. And adult Forex is an institutional theme, not for the man in the street.
The only thing is that the availability of data on local institutions is more accessible and trustworthy than in other countries.
For forex you just need to live in 2 countries to have some idea of what's going to happen with the local currency.)
Can you give some more examples of the types of data used. Interesting!
Open careers pics. Morbidity / mortality by country by industry, patents scientific and not only, grading and accounting for physicists from social media and not only for services.
Data gradation for good is just starting / going, so no textbooks yet. But AI in intelligence is already being screwed on by almost all countries)