The future of automated trading: round two - page 20
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
For all my loyalty to your ideas, I do not share your stubbornness.
I will try to justify it:
Ticks come in increments of 3 seconds, whereby the increments are variable and there is essentially no time reference.
A tick will come in 3 seconds or 20 seconds. The tick is the price change. This means that in essence, there is no time in Forex, only price change.
In the same way as in physics, there is no time, there is only the change of objects position and we judge that there is time by the amount of changes.
If no atom in the universe changes its position then we say that time has stopped.
In physics we use reference changes in atoms to measure time.
In forex, this elementary reference change is a tick. So tying a tick to the discreteness of time is not reasonable.
Secondly, if you are using formulas that require time as a parameter, you should use something more stable in discreteness than ticks.
And the starting point can be the M1 Open, it is stable in time and has the fixed nature (it is set once and for all without changes, unlike the Klose that floats all the time till the last moment).
Now about the noise, how it is possible to calculate position of a rocket having all knowledge of position of all atoms of a rocket? Will not you analyze all this information or you will generalize it to formality object-missile and already will process information on object as a whole.
This is the generalisation that is the bar. For the intercept, it does not matter that the position of the right aileron is now 2 cm higher than the position of the left aileron, what matters is the position at the last count and the position at this count. Using three counts it is possible to calculate any trajectory by one count.
It is possible to calculate the position of a future open minutiae from the 3-minute optics (if your formulas are at all applicable for such a prediction).
It is only possible to get away from the noisiness by taking small compensating displacements as noise and generalizing them to the notion of object trajectory, the main thing in this process is that the speed of the trajectory is less than the discreteness, otherwise the object will do several cycles of the trajectory in one counting. Such discreteness is not acceptable since one can easily miss the bifurcation point and the reaction to changes becomes strongly delayed.
Look closely, how many minute bars exceeding 3 spreads can you find?
If we look at the table for 10000 bars, we have 3 spreads exceeded 34 times on Close-Open differences and 87 times on High-Low differences.
Can we use these exceptions to build a strategy?
It is not unimportant that the overshoots are in steps that SO>3*spred ()=34 HL>4*spred()=24 it means that both shadows are within 1 spread.
Based on this analysis, I propose that the stops should only be triggered once a minute. This will be quite acceptable?
don't expose the stops but control the virtual stops once a minute. What is the problem.
ZZY Generally speaking talk about prognostic system with a horizon forecast of 1 hour 4 hours and based on the calculations on the minute for example transform Fourier transform 1024 minute bars is done forecast for the next 15 minutes or an hour. so triggering stops within the spread can not be a serious disturbance.
don't expose the stops but control the virtual stops once a minute. What's the problem.
Don't set stops and control virtual stops once a minute. What's the problem.
Apparently it's an understanding. I don't need a personal one. It was a question and a hint of a flawed approach to analysis.
so that the triggering of stops within the spread cannot be a serious nuisance.
Where do they give forex that the price doesn't go further than one spread in a minute? I want one too.
Don't put stops and control virtual stops once a minute. What's the problem.
Why is that? Because you and Lenya do not see the point in moving within a minute, and if tomorrow you say together that less than 5 minutes is all noise and there is no point in looking there, then what?
It is scary to think about. It is 2010, the speed of the Internet is adequate and I personally do not see downloading problems. Moreover, the speed increases in two - three times a year. It is more logical to give everyone the opportunity
It makes more sense to give everyone a choice when it comes to downloading ticks or not. The developer may have many reasons not to give ticks and it is not debatable, but it's cool to have a vote of traders on who and where to find what.
I understand Prival. You can argue about why you need tics, of course. Well, he needs them, that's all. I remember back in the day I asked the creators of mt deep hole-free history because my advisor was looking for the nearest neighbour, and I thought the longer the story, the more neighbours, the more protesters. I do not remember who (Renat or Roche) told me to create a random generator and test your EA using generated quotes. At first I thought it was a joke. But then it gradually came to me. There is a lot of noise in quotes and it is almost impossible to catch a signal. If we measure statistics (moments of different orders) of quotes, we can successfully generate them and they will look very close to the real ones. And what if we do the following for tick data: measure its moments (the 1st, 2nd, 3rd and so on), create a random number generator with such moments and generate a tick history for any time? Of course, we need to think about what to do with the times of news release. Maybe these times should be ascribed more volatility?
Where do they give you forex that the price doesn't go further than one spread in a minute? I want it too.
This is the statistics for 10000 bars CO>3*spred ()=34 HL>4*spred ()=24 those in 34 bars the body of which exceeds the spread by 3 times Shadow exceeds the spread by 4 times only 24, and 10 exceeds 3 spreads but no more than 4 and so on 5 spreads exceeds only 13 HL minute bars. 10 spreads no spread at all.
and what is 10 spreads 30 pips, those spreads exceeding 30 pips no spread at all, an error of 15 pips appears 13 times per 10,000 bars.
369 bars has a margin of error of 6 pips if you measure on an open M1.
Why is that? Because you and Lenya do not see the point in moving within a minute, and if tomorrow you say together that less than 5 minutes is all noise and there is no point in looking, then what?
It is scary to think about. It is 2010, the speed of the Internet is adequate and I personally do not see downloading problems. Moreover, the speed increases in two - three times a year. It is more logical to give everyone the opportunity
It makes more sense to give everyone a choice when it comes to downloading ticks or not. The developer may have many reasons not to give ticks and it is not debatable, but it's cool to have a vote of traders on who and where to find what.
This is not a voting, I have given my opinion on the feasibility of working with ticks, you can collect ticks by yourself and process them as you want.
I am more interested in news history directly in MT and mql-interface to these news, in my opinion (imho) these are more important things.