From theory to practice - page 255
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
While I'm busy sorting out WebMoney and opening an unrestrained signal and PAMM account, I would like to dwell again and again on the key point - the time intervals between tick quotes.
I checked it once again. This is what I have got for the AUDCAD pair:
This is the distribution of time intervals between real ticks
I keep repeating that this is the DISCURRENT LOGARIFY DISTRIBUTION
Column C represents real values of the probability density function
Column D - calculated by formula fromhttps://ru.wikipedia.org/wiki/Логарифмическое_распределение with p=0.7.
Gentlemen!!!!!!!!!
Well, show me at least one theory that would work at such time intervals between events.
There isn't one and there isn't one to be expected.
That's why I partition this time series with an exponent, introducing pseudo-states into it and working with diffusion equations.
No, not a lognormal distribution, the graph looks like an exponent.
https://ru.wikipedia.org/wiki/%D0%9B%D0%BE%D0%B3%D0%BD%D0%BE%D1%80%D0%BC%D0%B0%D0%BB%D1%8C%D0%BD%D0%BE%D0%B5_%D1%80%D0%B0%D1%81%D0%BF%D1%80%D0%B5%D0%B4%D0%B5%D0%BB%D0%B5%D0%BD%D0%B8%D0%B5
but here's the exponent:
https://ru.wikipedia.org/wiki/%D0%AD%D0%BA%D1%81%D0%BF%D0%BE%D0%BD%D0%B5%D0%BD%D1%86%D0%B8%D0%B0%D0%BB%D1%8C%D0%BD%D0%BE%D0%B5_%D1%80%D0%B0%D1%81%D0%BF%D1%80%D0%B5%D0%B4%D0%B5%D0%BB%D0%B5%D0%BD%D0%B8%D0%B5
Here is the lognormal, the distribution of the number of ticks per bar.
https://forum.fxclub.org/threads/32942-prostye-nenuzhnye-veshhi?p=594214&viewfull=1#post594214
No, not a lognormal distribution, the graph looks like an exponent.
https://ru.wikipedia.org/wiki/%D0%9B%D0%BE%D0%B3%D0%BD%D0%BE%D1%80%D0%BC%D0%B0%D0%BB%D1%8C%D0%BD%D0%BE%D0%B5_%D1%80%D0%B0%D1%81%D0%BF%D1%80%D0%B5%D0%B4%D0%B5%D0%BB%D0%B5%D0%BD%D0%B8%D0%B5
but here's the exponent:
https://ru.wikipedia.org/wiki/%D0%AD%D0%BA%D1%81%D0%BF%D0%BE%D0%BD%D0%B5%D0%BD%D1%86%D0%B8%D0%B0%D0%BB%D1%8C%D0%BD%D0%BE%D0%B5_%D1%80%D0%B0%D1%81%D0%BF%D1%80%D0%B5%D0%B4%D0%B5%D0%BB%D0%B5%D0%BD%D0%B8%D0%B5
No. The time intervals between real ticks are logarithmic and full stop.
https://en.wikipedia.org/wiki/Logarithmic_distribution
If this is a property of a non-Markovian process, then I'm even glad. Let it be logarithmic. The "time spiral" comes immediately to mind... Beauty, in a word.
But, beauty is not enough for us, is it?
And yes! Gentlemen of physics and mathematics!
With my head bowed and my cap removed, I humbly ask you to post on this thread a REALLY work-proven formula for calculating the Hearst coefficient.
The formulas are long forgotten, as there are plenty of ready-made packages that, among other things, calculate the Hurst coefficient.
This is the so-called idea of fractional differentiation. The required values are searched automatically, e.g. auto.arima {forecast} . The result of the function will be two vectors containing three values:
You can name other packages as well.
PS.
In the mentioned package this Hurst of yours is a minor detail, and extremely rare. But that's for trifles, as this package is dumb, as the markets are not just NOT stationary, there are also more than a hundred models for which they have been invented.
While I'm busy sorting out WebMoney and opening an unrestrained signal and PAMM account, I would like to dwell again and again on the key point - the time intervals between tick quotes.
I checked it once again. This is what I have got for the AUDCAD pair:
This is the distribution of time intervals between real ticks
I keep repeating that this is the DISCURRENT LOGARIFY DISTRIBUTION
Column C represents real values of the probability density function
Column D - calculated by formula fromhttps://ru.wikipedia.org/wiki/Логарифмическое_распределение with p=0.7.
Gentlemen!!!!!!!!!
Well, show me at least one theory that would work at such time intervals between events.
There isn't one and there isn't one to be expected.
That's why I break this time series with exponent, introduce pseudo-states into it and work with diffusion equations.
Tell me, Alexander, what reasons do you have for repeating relentlessly that this is "DISCRETE LOGARIFMIC DISTRIBUTION"? Attached to your message is a spreadsheet, from the data of which, as I guess, a histogram of sample frequencies is constructed using Vissim tools, then in column D the calculated probabilities of this distribution are added for comparison at p=0.7. Why don't you logarithm the ordinate axis of the graphs? See, I just logarithmed it in the same table by ticking it, and added a comparison of the two frequency series (columns C, D). After the row highlighted on the left, the discrepancy becomes noticeable, and at the end of the value interval you selected for analysis (30), the calculated probability differs from the measured frequency by a factor of 10. And yet you catch the behaviour of ticks on the tails of their distributions. I don't get it.
That's exactly what I told you) at the output of Vissim's buy/sell commands. On these commands the robot sends OrderSend at the current price. It doesn't have SL/TP, it also closes on command.
You have overcomplicated things).
So, roughly speaking, the algorithm is as follows:
1 Take the last n ticks.
2 Rebuild their time(only time)
3 believe that now you can apply the diffusion equation, and apply it
4 we determine if the current market moment is far away from the mathematical expectation (we took n before the last tick)
5 if the market is far enough away (the definition is very vague, but you don't mention the other one), we throw OrderSend(on the market!). To buy or to sell is determined by the direction in which the deviation is found.
6 While the deal is open the second one in the same side (!) on the same asset do not open
7 ticks continue to read and do the same from point 1
Note: a trade will be closed when our esteemed VisSim decides in point 5 that we have come to the mathematical expectation. There are no other conditions for exiting a trade? No additional signal filters?
Right???
Or correct me.
Re:
By how much is the maturity expectation it calculated there different from the one you can calculate directly from the market data? Their values should be dumped into the log for each trade. Are the RMS's very different? The levelsbehind which it goes back to the average are not calculable without converting to its space? How doesthe number of trades and percentage of profitable ones change if these levels are placed further/closer to M?
Then you and the author simply do not have the answers.
There is no time, "get your pockets ready". Right?
This time the error is irrelevant. All that matters is that these intervals are not initially uniform or exponential.
The flow of events is clearly not random! This once again shows the non-marking nature of processes in the market.
I repeat - we do not have a developed mathematical apparatus for describing non-markovian processes, that's why most traders get confused, invent more and more new ways to fight the market, etc.
I, on the other hand, do it simply - I forcefully read the quotes through the exponent.
That gives me the right to use the equations of diffusion, where, by the way, the time root that you love appears, that's what I'm talking about.
Sincerely,
Alexander_K2
Checked again. This is what I got for the AUDCAD pair:
This is the distribution of time intervals between real ticks
Well, show me at least one theory that would work with such time intervals between events.
I'm probably being childishly obtuse right now, but still.
Why aresuch time intervals disconcerting? Because we will increase the weight of those sections where there were powerful trades? What if we simply draw candlesticks by these ticks, at least ten seconds long, if ticks allow? Then take the median of a candlestick? And we obtain a series of pseudoticks equal in time. Why not?
No, well, you write all the right things, but for some reason you pull the RMS in your posts.
I don't count CCO. It's not there. There is a variance calculated by the formula through the diffusion coefficient and time.
And I have an additional parameter for input - it is the asymmetry coefficient nonparametric skew https://en.wikipedia.org/wiki/Nonparametric_skew, but it does not work well. I'm looking for a replacement. It's either Hurst or nonentropy.
I'm probably being childishly obtuse right now, but still.
Why aresuch time intervals confusing? Because the weight of those parts will increase where there were powerful pro-trades? And what if we simply draw candlesticks by these ticks, even ten seconds long, if ticks allow? Then take the median of a candlestick? And we obtain a series of pseudoticks equal in time. Why not?
Look on the Internet - is there a mathematical apparatus to describe processes with a flow of events in logarithmic intervals? And through exponential ones, there is! Through uniform time intervals my strategy works the same way, but, slightly worse.
This time the error is irrelevant. All that matters is that these intervals are not initially uniform or exponential.
The flow of events is clearly not random! This once again shows the non-marking nature of processes in the market.
I repeat - we do not have a developed mathematical apparatus for describing non-markovian processes, that's why most traders get confused, invent more and more new ways to fight the market, etc.
I, on the other hand, do it simply - I forcefully read the quotes through the exponent.
That gives me the right to use the equations of diffusion, where, by the way, the time root that you love appears, that's what I'm talking about.
Sincerely,
Alexander_K2
Alexander, how many values did you take for the sample to understand that the distribution is logarithmic?