From theory to practice - page 138
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 there is no deep tick history, and there is a glimpse into the future in the calculation, I am very much in doubt.
I will check these calculations at the weekend with different reading methods.
I have tick data when reading:
1. quotes without pseudo-states in 1 sec.
2. quotes without pseudo-states through exponent.
3. quotes with pseudo-states through the exponent (this is the case we are studying now)
My broker does not have tick archives, so I have to gather the history by myself anyway
I will check these calculations at the weekend with different reading methods.
I have tick data when reading:
1. quotes without pseudo-states in 1 sec.
2. quotes without pseudo-states via exponent.
3. quotes with pseudo-states through the exponent (this is the case we are studying now)
My broker does not have tick archives, so I have to gather the history by myself anyway
On the chart, the marks are in increments of 1,520. If an exponent, it's a mess with the marks, because our scale starts on the right and not on the left.
No. Just a random exponentially distributed number generator generates something like 11223391128145 and I read quotes at these intervals. I collect these quotes into a buffer of 15625 and make calculations of current parameters and average parameters at each new quote. The longer the process goes on, the more noticeable it is that yes - the current variance value (according to the formula from the attached file earlier) changes insignificantly, and the average variance does not change at all with time. Note that this all goes with respect to 0. If we plot the variance histograms (Column A in tab Sheet 2), we see a near-normal distribution on average:
This is with an archive of historical values = 211690 quotes
I think, when we will have the history of 1.000.000 values, we will see a perfect bell
Do you realise that this is a peek into the future?
In history these sorts of twists go through, in real time they don't.
Why would there even be a pullback to the mean?
Anyone who trades knows that a trend correction through sideways is quite possible, not by Elliot waves. I purposely selected a sample of three years where there were several sections with sideways trend correction and the total movement with sideways trend correction was over a thousand pips.
All these gifts are clearly visible in the signals of the averaging lovers.
Why should there even be a pullback to the mean?
Anyone who trades knows that a sideways trend correction is quite possible, not by Elliot waves. I purposely selected a sample of three years where there were several sections with sideways trend correction and the total movement with sideways trend correction was over a thousand pips.
All these gifts are clearly visible in the signals of the averaging enthusiasts.
Why should there even be a pullback to the mean?
Anyone who trades knows that a sideways trend correction is quite possible, not by Elliot waves. I purposely selected a sample of three years where there were several sections with sideways trend correction and the total movement with sideways trend correction was over a thousand pips.
All these presents can be clearly seen in the signals of averaging lovers.
According to the algorithm suggested by your old friend Vizard_, the pullback to the average will definitely be for the increments.
Take another close look at the lower graph of the mean value of increments in the sample of 15625 relative to the price itself (upper graph)
and the distribution that this average value of increments forms over time (see my previous post).
PS I publish again the linkhttps://yadi.sk/d/Q26c4qoS3RbJRn not to be lost...
Why should there even be a pullback to the mean?
Anyone who trades knows that a sideways trend correction is quite possible, not by Elliot waves. I purposely selected a sample of three years where there were several sections with sideways trend correction and the total movement with sideways trend correction was over a thousand pips.
All these presents can be clearly seen in the signals of averaging lovers.
Because the market is constantly looking for the price optimum. And when there is a change, the average is pulled up and the market comes back to it.
But graylists forget that the average should be set on the midpoint of the window, i.e. shifted by half a period from the right point backwards.
The market always comes to this middle point. But in this case it is nothing at the right border.
If we do not show humor, the market does not always return to the middle point, especially when it follows the trend. It will touch the wrist after a while, but you will be sitting in a specific minus.
No. Just a random exponentially distributed number generator generates something like 11223391128145 and I read quotes at these intervals. I accumulate these quotes into a buffer of 15625 and make calculations of current parameters and average ones at every new quote. The longer the process goes on, the more noticeable it is that yes - the current variance value (according to the formula from the attached file earlier) changes insignificantly, and the average variance does not change at all with time. Note that this all goes with respect to 0. If we plot the variance histograms (Column A in tab Sheet 2), we see a near-normal distribution on average:
This is with sample size = 211690
I think that when we get a history of 1,000,000 values, we'll see a perfect bell
This is something I don't understand at all.
If the time stamps are random numbers (which have an exponential distribution), then regardless of the distribution it will result in a shuffling of the original quotient, and you clearly see a regular quotient, i.e. everything is ordered.
But that's just a matter of speaking.
The point is the principle.
Your system DOES NOT work.
Systems like this are very well researched, even nobility people have got (co-integration is called). You can build decently working arbitrage systems. But it's always multiple instruments. No one has ever managed to do this on one currency pair - the reason is named above: you can wait for a pullback to zero through a drawdown of several thousand pips and you can NEVER wait for it.
This I don't understand at all.
If the timestamps are random numbers (which have an exponential distribution), then regardless of the distribution this will cause the original cotier to shuffle, and you clearly have a regular cotier, i.e. everything is ordered.
But that's just a matter of speaking.
The point is the principle.
Your system DOES NOT work.
Systems like this are very well researched, even nobility people have got (co-integration is called). You can build decently working arbitrage systems. But it's always multiple instruments. No one was able to do it on one currency pair - the reason is named above: you can wait for a pullback to zero through a drawdown of several thousand pips and you can NEVER wait.
SanSanych, on this one I will not waste my time arguing.
Once again - this algorithm was actually given to us by your old friend Vizard_. I have no reason not to trust him - his calculations were fully consistent with mine.
If you still don't believe me or don't understand it, you can do the calculations yourself.
SanSanych, this time I won't waste my time arguing.
Once again, your old friend Vizard_ actually gave us this algorithm. I have no reason not to trust him - his calculations fully coincide with mine.
If you still do not believe or do not understand it, you can do the calculations yourself.
At weekend I will do it in mql, but on minute bars. I did not know how to deal with ticks, there is more hassle than money.