Market phenomena - page 45
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
Is the blue one the result of a tweak or just drawn for clarity?
It's a Gaussian fitted by least squares to the raw data.
Show the same for the Eurobucks, for M15 or smaller. And correct the sign.
Please.
Here is the distribution of the EURUSD 15m series minutes for Open[i]-Open[i+1] - red and Open[i]-Close[i] - blue:
As you can see, there is no significant difference.
The sign of what to tweak?
Here is the distribution for the sentinels. The blue one shows the normal distribution. A good convergence to the Gaussian distribution can be ascertained.
I would like to add one more nuance to independence: the presence of a deterministic component (or is it a dependence?). There is no point in talking about the statistical characteristics of BP if there is autocorrelation in it. The deterministic component will score everything and nothing can be trusted.
Autocorrelations of price increments are hardly significant.
Autocorrelations of price increments are hardly significant.
Yes. :)
The market phenomenon I have been able to comprehend so far is the following:
1. You can analyse the history of the market and identify the various "patterns" by which the market has supposedly moved;
2. You can predict the market with varying degrees of probability;
3. Most importantly: The market will still act in a more original way than items 1 and 2;
4. For the trader, the most valuable pattern of the market is volatility, i.e. the principle of "everything returns to normal", presence and consideration of mutual correlation of trading instruments.
5. It is impossible to squeeze anything else out of the market - it will cost you money.
It is only a formal coincidence of one line with another. From this you cannot conclude that the mechanism of formation of both lines is the same. If you have dependence between samples on ticks, it exists also on the watch. It does not disappear with zooming in. Therefore, you cannot call this distribution normal. It's similar in form but not in content.
It does not make much sense to state anything on the subject without the results of a particular instrument. I did not plan in this thread to delve deeply into the question of the mechanism of baroque formation and expressed only my general opinion on the subject. But I can also justify it.
If you build an AR model for a tick process you will see that the AR coefficients are non-zero only for the first and less often for the second term. This tells us that in a linear model each tick remembers history no deeper than the previous tick. Next, in depth, by induction. If, for example, the relationship to the previous tick in our BP is a=1/2, then the relationship to the pre-previous tick will be 1/4, and so on. Now, what is an hourly bar? - It is a thinning of a tick series with a non-equidistant step. For estimation, this step can be taken to be a constant, and on an order of magnitude estimate of 1000 ticks. So, if we have a link between ticks, as you correctly noted above, then there is one on the clock as well. Let's estimate this relationship by order of magnitude: aH=(1/2)^1000=10^-300->0. With great accuracy it is zero!
So your ". It (dependence) does not disappear anywhere with increasing scale..." needs justification and the hourly bar distribution I cited most likely tends to be normal both in form and in content precisely because of the lack of a meaningful relationship between the counts. Of course, I am going too far and understand that ticks inside of an hour bar and on its border are not the same due to the crowd psychology, which takes part in the price formation mechanism. Also I cannot exclude the presence of non-linear links between ticks that are not considered by the AP-model. I think that these effects do not determine the shape of the distribution of bar amplitude on large TFs.
Autocorrelations of price increments are hardly significant.
It does not make much sense to say anything on the subject without the results of the research on the particular instrument. I did not plan in this thread to delve deeply into the question of the mechanism of baroque formation and expressed only my general opinion on the subject. But I can also justify it.
If you build an AR model for a tick process you will see that the AR coefficients are non-zero only for the first and less often for the second term. This tells us that in a linear model each tick remembers history no deeper than the previous tick. Next, in depth, by induction. If, for example, the relationship to the previous tick in our BP is a=1/2, then the relationship to the pre-previous tick will be 1/4, and so on. Now, what is an hourly bar? - It is a thinning of a tick series with a non-equidistant step. For estimation, this step can be taken to be a constant, and on an order of magnitude estimate of 1000 ticks. So, if we have a link between ticks, as you correctly noted above, then we also have it on the clock. Let's estimate this relationship by order of magnitude: aH=(1/2)^1000=10^-300->0. With great accuracy it is zero!
So your ". It (dependence) does not disappear anywhere with increasing scale..." needs justification and the hourly bar distribution I cited most likely tends to be normal both in form and in content precisely because of the lack of a meaningful relationship between the counts. Of course, I am going too far and understand that ticks inside of an hour bar and on its border are not the same due to the crowd psychology, which takes part in the price formation mechanism. Also I cannot exclude the presence of non-linear connections between ticks that are not considered by the AP-model. I think that these effects do not determine the shape of the bar amplitude distribution on large TFs.
The market phenomenon I have been able to comprehend so far is the following:
1. You can analyse the history of the market and identify the various "patterns" by which the market has supposedly moved;
2. You can predict the market with varying degrees of probability;
3. Most importantly: The market will still act in a more original way than items 1 and 2;
4. For the trader, the most valuable pattern of the market is volatility, i.e. the principle of "everything returns to normal", presence and consideration of mutual correlation of trading instruments.
5. It is impossible to squeeze anything else out of the market - it will cost you money.