Discussion of article "The algorithm of ticks’ generation within the strategy tester of the MetaTrader 5 terminal" - page 3
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
Post in a separate thread all detailed and repeatable information together with the Expert Advisor code - we will all check and discuss it. If it is a piper, then also logs of his real trading.
This is a correct and honest approach. The same as in the above article.
:o) I may be wrong and not honest, but the code of this EA, I'm sorry, but I will keep it a secret,
and even more than that I will not even hint on what principle it works (I have every right).
And I am not going to discuss my Expert Advisors with anyone.
I gave you the idea of importing ticks and you think for yourselves.
At this point I think the topic is over because I have heard everything I wanted from you, and you from me, and I think we will not be able to say anything new to each other.
:o) I may be wrong and not honest, but I will keep the code of this EA secret,
That's exactly what I wanted to show publicly.
Instead of public practical research of tick generation quality, you both refused and limited yourselves to theoretical and unsupported fabrications.
"Durations are often modelled using Poisson processes."
Irene Aldridge "High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems"(http://books.google.ru/books?id=8iCiOip5scIC&pg=PA121&lpg=PA121&dq=quote+arrival+frequency+distribution&source=bl&ots=L8SQvN1XVP&sig=uuoAHB7tv3ExC0W_xY_dKqnaU9U&hl=ru&ei=d7H2S7rOK4GROMf4-M8I&sa=X&oi=book_result&ct=result&resnum=5&ved=0CDcQ6AEwBA#v=onepage&q=quote%20arrival%20frequency%20distribution&f=false)
Thanks, will take note and do some experiments using this model.
But it is much easier for us - due to detailed history we have modelling in micro-periods (minutes), which drastically reduces (almost to zero?) potential error in tick frequency and distribution.
This is exactly what I wanted to publicly demonstrate.
Instead of a public practical study of the quality of tic generation, you both refused and limited yourselves to theoretical and unsupported fabrications.
You don't have to speak for everyone. Already on the first page I asked you questions, absolutely clear and you answered them yourself, what research do you need? If you immediately admit that you ignore these questions when modelling tics.
1. Your modelling ticks come from a quartz oscillator ("intensity is uniform across the bar"). This does not happen in REAL life !!!!.
2. You have a rigidly set order of Low and High. And it is determined only by the colour of the bar. The question of how often this is true on real data - you do not know the answer.
3. You ignored one question at all, to draw a multi-currency arrangement of ticks on the bar .
4. Remembering your "skill" in modelling and your attitude to it in MT4. I can ask one more question. Do you have the same number of ticks in a bar? In MT4 you threw out 20% of ticks. How is it now? If a real bar has 100 ticks, how many ticks will you have when modelling ?
We (me in particular) are trying to convey to you one simple idea understandable to any schoolchild. A model is always worse than real data ! You are stubborn and say no, not so. And cite philosophical reasoning, we 10 years, modelling, create robust algorithms ... What is this here? We are talking about the quality of modelling, about adequacy....
You are the creators of the model, you have to prove its adequacy (and not make bare statements look at the picture). Be kind enough to answer the posed questions concerning adequacy. Talks in the style of "we know everything, we know how to do everything, and you are all fools here, will not pass". I have modelled such processes, which you would not dream of, and proved adequacy of the model, not to amateurs in this field.
Renat :
Instead of public practical research of tick generation quality, you both refused and limited yourselves to theoretical and unsupported fabrications.
I am not interested in ticks and their history, because I design trading strategies in quantum space - where there is neither time nor price. But I saw something else in the article - it is an algorithm for modelling possible price movement. Moreover, the authors proved its adequacy. Having it in hand, you can set input parameters (OHLCV) to get possible movement trajectories. To study your algorithm, it would be good to get the public code of tick modelling in the form of a library.
What public code you need in the article is described in detail (you can even say algorithmically),
and having built an Expert Advisor using this algorithm, you can get profit immediately in the tester.
(it's a pity only in the tester because in real time the strategy built according to this algorithm will immediately lose).
And it will fail because it is NOT ADEQUATE.
I have already written (on the mql4 forum) about an Expert Advisor that uses the property of a series to move first against the main movement (and I did not set this property of the Expert Advisor, the machine itself selected it through optimisation among the set of properties),
so we get an improved entry and immediately determine the direction,
but the only problem is that this property exists only in the tester for a number of quotes, in real life it does not exist.
Again, another strategy (or rather indicator) that processes 5000 ticks and extrapolates to 1200, and this is about an hour and no one can call it a pipsar contemptuously, everything is robusto, but the adjustment of filters of such a strategy is possible only for months debugging in real time, because in the tester is a complete darkness.
And to prove it Renat offers me to sit for a month and then publicly wash my strategy's bones,
I will be S&M proof and Renat will be like a fly.
:o) I may be wrong and not honest, but I will keep the code of this Expert Advisor secret,
That's exactly what I wanted to show publicly.
Instead of public practical research of tick generation quality, you both refused and limited yourselves to theoretical and unsupported fabrications.You base your defence on public ridicule of your opponent, a very wise position.
Your tics were checked for adequacy and more than once, that's why the topic arose. And when the author saw the discussion, he came to the rescue and wrote an article, for which by the way I thank him very much, because the article is really informative, and everything in it is correct, I do not agree only with the conclusion about adequacy, but if it was said that the chart clearly shows that the quality of tick modelling in the MetaTrader 5 client terminal tester allows to test experts on historical data in a good approximation , I would be in full agreement with the author.
But it still does not change the fact that importing user's tick history is necessary.
ps at the second reading I found out that the author is MQ, so it is Renat'a who should be thanked for the article.
Don't speak for everyone. Already on the first page I asked you questions that are absolutely clear and you answered them yourself, what more research do you need? If you immediately admit that you ignore these questions when modelling tics.
1. your modelling tics come from a quartz oscillator ("intensity is uniform across the bar"). This does not happen in REAL life !!!!
Time between ticks doesn't matter unless you are a pips guy. Look at the comparison chart - everything clearly converges. This is a practical example.
2. You have a rigidly set order of Low and High. And it is determined only by the colour of the bar. The question is how often it is true on real data - you do not know the answer.
I suggested not in vain - "check it yourself in practice" by recording 1-2-3 days and comparing. We gave our evidence with pictures.
3. you ignored one question at all, to draw a multi-currency arrangement of ticks on the bar .
Each instrument is modelled independently.
4. Remembering your "skill" in modelling and your attitude to it in MT4. I can ask one more question. Do you have the same number of ticks in a bar? In MT4 you threw out 20% of ticks. How is it now? If a real bar has 100 ticks, how many ticks will you have in modelling?
We (me in particular) are trying to convey to you one simple idea understandable to any schoolchild. A model is always worse than real data ! You are stubborn and say no, not so. And cite philosophical reasoning, we 10 years, modelling, create robust algorithms... What is this here? We are talking about the quality of modelling, about adequacy....
And I'm saying that I've been seeing the standard misconceptions of pipsers for 5 years already. They are ready to self-deceive themselves time after time, trying to replace a trading strategy with tactical pipsing (which leads to failures in real trading).
You are the creators of the model, you have to prove its adequacy (and not make naked statements look at the picture). Be kind enough to answer the posed questions concerning adequacy. Talks in the style of "we know everything, we know how to do everything, and you are all fools here, will not pass". I have modelled such processes that you would never dream of, and proved the adequacy of the model, not to dilettantes in this field.
We have conducted practical research for many years, made the third version of the tester, conducted tests, showed the results, presented the method of verification, offered traders to check everything by themselves, and the answer is a naked theory.
I did not point out the problem of idealisation of the financial market by mathematicians with all the resulting failures for a reason.
Mathematicians do not want to think about the fact that there is a manual or automatically adaptive tick filter on the server, which can change a hundred times a day and from version to version. They usually have enough mat models, regardless of the words of people who know (and even developers of all this stuff) "don't get involved in ticks - it's self-deception".
You base your defence on public ridicule of your opponent, a very wise position.
Your tics were checked for adequacy and more than once, that's why the topic arose. And when the author saw the discussion, he came to the rescue and wrote an article, for which by the way I thank him very much, because the article is really informative, and everything in it is correct, I do not agree only with the conclusion about adequacy, but if it was said that the chart clearly shows that the quality of tick modelling in the MetaTrader 5 client terminal tester allows to test experts on historical data in a good approximation , I would be in full agreement with the author.
But it still does not change the fact that importing user's tick history is necessary.
ps at the second reading I found out that the author is MQ, so it is Renat'a who should be thanked for the article.
I base my defence on public proof of conclusions.
You haven't provided anything, and you also failed with your knowledge of MetaTrader 5 (you didn't even know that it has a debugger). Don't forget that we are talking about MetaTrader 5 and MQL5
Unfortunately, I did not write the article, although I would have been happy to do it.
I base my defence on public proof of conclusions.
You did not provide anything, and you also failed with your knowledge of MetaTrader 5 (you did not even know that it has a debugger). Do not forget what resource you are on.
Unfortunately, I didn't write the article, though I would have been happy to do it.
I have every right to poke fun at the knowledge of MetaTrader 5 because I never said that I am an expert in this platform,
and I'm certainly not going to compete with the developer.
Regarding public evidence,
here is a typical example if a neural network relies on the property of tester ticks to make the first tick in the wrong direction, and this property is right on the surface because the tester generates ticks knowing in advance where the bar will be drawn,
so here is the grail, we open on the third tick against the movement and five ticks later we take profit,
here it's enough to put a filter that the first tick should be not less than the spread,
and that's it, we're done with pipsing, we only catch movements greater than 30 pips.
But only to check the inoperability of this TS will require at least a month of intense sitting behind the monitor.
ps Actually it is asked from you to make the last instance before the realtime, let this check will not be long historically,
but it will give an opportunity not to sit at the monitor for months catching what can be done on the history of ticks.
I have every right to poke fun at the knowledge of MetaTrader 5 because I never said that I am an expert in this platform,
and I certainly don't intend to compete with the developer in knowledge.
Regarding public evidence,
Here is a typical example if a neural network relies on the property of tester ticks to make the first tick in the wrong direction, and this property is right on the surface because the tester generates ticks knowing in advance where the bar will be drawn,
so here is the grail, we open on the third tick against the movement and five ticks later we take profit,
here it's enough to put a filter that the first tick should be not less than the spread,
and that's it, we're done with pipsing, we only catch movements greater than 30 pips.
Since you do not know the MetaTrader 5 terminal, you are not aware of the trading modes in the tester and did not even pay attention to my explanation of these modes with the picture on the first page of this topic.
Testing modes are specially designed to sober up traders and write robust Expert Advisors. This will dramatically and qualitatively improve the Expert Advisors.
I have repeatedly written about aggressive modes in the tester in forums (MQL4.com and MQL5.com).
But to check the inoperability of such a TS will require at least a month of intense sitting at the monitor.