Distribution of price increments - page 9
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
Good evening, Maxim!
I'm kind of new to forex, but I have some background in physics, and some practice in technological processes. And there are other processes in technology, too...
I don't know anything about trading, so please don't judge severely.
For me it is important to understand the current process - then I will be programming. If traders-programmers, for example, are already using the exponential frequency instead of the real tick rate and obtain any unusual results - I'll be glad to hear their opinion.
In order to get some understanding of the "current process" of quotes forming in retail Forex, the best thing you can do is to take a course in your city. There are brokerage houses which offer free introductory courses. Forex companies don't reveal much about how the courses are created. It is their know-how. If you want to know the insider information, a lot of approaches are described here https://habrahabr.ru/post/202402/, but to understand it you have to attend at least an introductory course. I would also recommend searching on the internet for the words A-Book and B-Book, these issues are already largely open, published.
Since you like, to put it mildly, to "rework" the source data, here http://tradetrade.ru/MythBusters/2014/08/04/fundamentalnaya-oshibka-algotreyderov.html"Legend Breakers Fundamental Error of Algotraders" describes the method of correct data reduction, correct in terms of its subsequent analysis for profitable trading using the method of spatial arbitrage.
Finally, just on the number of ticks themselves (for banks and DCs) see here http://ru.forexmagnates.com/chto-takoe-toksichnyiy-potok-na-ryinke-foreks/ What is "toxic flow" in forex? 01.11.2017.
In order to get some understanding of the "current process" of price formation in retail Forex trading, the best thing you could do is enroll in a course with a brokerage company in your city. There are brokerage houses which offer free introductory courses. Forex companies don't reveal much about how the courses are created. It is their know-how. If you want to know the insider information, a lot of approaches are described here https://habrahabr.ru/post/202402/, but to understand it you have to attend at least an introductory course. I would also recommend searching on the internet for the words A-Book and B-Book, these issues are already largely open, published.
Since you like, to put it mildly, to "rework" the original data, here http://tradetrade.ru/MythBusters/2014/08/04/fundamentalnaya-oshibka-algotreyderov.html"Legend Breakers Fundamental Error of Algotraders" describes the method of correct data reduction, correct in terms of its subsequent analysis for profitable trading using the method of spatial arbitrage.
Finally, just on the number of ticks themselves (banks and DCs) see here http://ru.forexmagnates.com/chto-takoe-toksichnyiy-potok-na-ryinke-foreks/ What is "toxic flow" in forex? 01.11.2017
Thanks for the links!
About the courses - probably the appearance of a man under 50 would have caused laughter and amusement :))
I'll try to figure it out on my own. If before the New Year I see that in theory it's just a 50/50 game, I'll close the subject for good.
Right now I'm getting data from a real NDD account (not trading yet, just opened an account for pure experimentation). As I understand it - data from such an account can be trusted unconditionally.
Of course, you can believe it. At the moment you come into your terminal and provided you have not been individually quoted in the form of, for example, spread extensions (just for you or for the group of clients you have been put into). This is indeed a quote from the server of this company.
Look at the client agreement of this company, the section on disputes or claims - you will definitely find a clause like "When considering a dispute, the only source of quotes is the company's server quotation database, other sources cannot be an argument in the dispute". What do you think it is for?
Next, look for the words "glaring error", "failure", "non-market quote", "spike" in the contract documents - somewhere next to them will be written about the right of the company to make changes to the server quotation database. It's an obvious enough right, they might not even mention it.
To find or calculate the quotes that may be considered as quotes of Forex market (even if not real, only retail) is a separate task. I'm afraid if I go on about it, you will lose your hands, because you are interested in the smallest increments, where the most noticeable differences in rates in different companies (see, for example, http://www.gurutrade.ru/forex-quotes/).
Thanks for the links!
About taking the course - probably someone in their 50s would be hilarious and laughable:))
I'll try to figure it out on my own. If, before the New Year, I see that in theory it's just a 50/50 game, I'll close the subject for good.
You are wrong about age. I attended an orientation course in early 2007 and was surprised at the strong clustering of trainees - either students and young people looking for work, or recently unemployed and retired people. Only a few were of middle age. The oldest were in their seventh decade. How it is now, I can't say.
!!!!!!!!!!!!!!!
... In order to deduce something one must make sure that external conditions are the same or at least similar, otherwise it is "the average temperature in a hospital. Not even an annual average." Well one of the measurement scales has changed.
i.e. p1. identify which conditions are relevant
External conditions for currencies change significantly at least 2 times a day. The market before London opens and after NY opens are two different markets.
Absolutely agree. I've written about this (highlighted) here. The type of distribution can change significantly from the conditions, and its parameters are out of the question.
Right now I'm getting data from a real NDD account (I'm not trading yet, I just opened an account for pure experimentation). As I understand it - data from such an account can be trusted unconditionally.
Once again, a tick is a technological concept. If a broker has a sufficient number of clients and liquidity providers, the ticks are "even". That is, without gaps, at a reasonable pace, corresponding to the agreements you signed at the opening.
Otherwise, there will be gaps, gaps, spikes and so on. There is no market-wide concept of"tick volume". It's a private shop. The volume will be different on different servers. They are not sworn to be identical - it's a distributed system.
If your own conditions change over a long period of time (liquidity strains or you connect a new one, or switch to a different version of server software) the tick volume structure of a particular server changes. And this can be seen.
The data from __any_ account can be trusted unconditionally. Why else would you open it if you have doubts?
Counterexamples for those who like to "think":
If returns follow the AR(1) model - both trend and counter-trend trading can work; the resulting process is Markovian. If we add fractional order integration - the process becomes non-Markovian, but both trend-following and counter-trending will work.
The AR(1) model does not have so many parameters to understand which of them depends on the performance of trend trading, if the matter is not whether the process is Markovian or not.
I was talking about going from AR(1) to larger orders (memory is periodically included). The idea of trading along the trend or against it depending on marginality is not mine, I am the author of the thread. I'm sure we can't define it without lag, which would result in no more efficiency than, for example, a naked swing. So complication is required, including knowledge of external factors(news schedules etc).
In addition to the general idea - processing at tick level by this method is IMHO unpromising. It needs higher timeframes.
Alexander_K, what is the problem solved for in general? What is the tick analysis for? Is it for earning 15 ticks from an entry tick?
What data do you get... the speed of light - exchanges, servers, client computers are not located in the same data centre, and the speed of communication and equipment does not affect significantly the latency now, 15 years ago you could open a DT terminal and open some other terminal with quotes with lower latency and on the difference of 5-10 seconds between the terminal values try to earn (or other ways).
The initial minimum step increment in the terminal is equal to spread (or recalculate it from the commission), for example, within this step you can observe for 10 minutes development of the trend and its correction in 62%. During these 10 minutes 300-700 ticks will come (it is a significant value for statistics), but at this step it will not be possible to use statistics (at the level of your client terminal of your broker/dealer). Such conditionally "useless" for you sections, roughly, for 3-5 hours per day. You, this is a team from one person physicist, and how to draw the quotes in your terminal has thought and invented 100000 people physicists with more reliable and complete data - the algorithm of drawing/delivering quotes is constantly evolving. If before your terminal "bad" quotes are filtered out, then "combed" three times, then in your terminal the quotes are already all "bad". What is the point of analysing this series, to determine the algorithm of quote modification/supply to your terminal? - If you go to a big brokerage company, you may find out everything and get paid.
Take a more reliable third-party quotation source for the DJ FXCM USDOLLAR index (it is based on more reliable data) and calculate the same index in your terminal using quotes from it. With 99% probability you will find the difference. Whether you will be able to use this difference for earning - no.
The peculiarity of human psychology is in picking out of a line of sight what you like or what is more suitable, and in fact it turns out that in this situation you need to screw up or score a lot of additional conditions.
I was talking about moving from AR(1) to higher orders (memory serves periodically). The idea of trading along or against the trend depending on marcicity is not mine, I am the author of the thread. I'm sure we can't define it without lag, which would result in no more efficiency than, for example, a naked swing. So complication is required, including knowledge of external factors(news schedules etc).
In addition to the general idea - processing at tick level by this method is IMHO unpromising. It requires higher timeframes.
I tend to think that trading by ticks is the only right way.
Another question is which sampling volume you use in calculations.
At the moment I am investigating two ways to collect ticks - in real time (we have a non-Markovian process) and in intervals predefined by the exponential law (we obtain a Markovian process).
1. For a non-Markovian process we study integrodifferential equations of motion (in practice, the set of current and averaged process parameters)
2. For Markovian - only current parameters.
And draw conclusions.
Something tells me (including specialists on this thread of forum), that Markovian process may not be considered at all... Well, except for the sake of curiosity.:))))
Good luck to all!