Market phenomena - page 61
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
Swinosaurus was the parent of that branch, but the interesting thing is, as Mathemat said , in the middle to the end of the branch. And that, interesting by and large, is not his merit, but
Yurixx, Mathemat, Candid, MetaDriver , etc. Read it, you'll find it interesting.
Friday's just started and they've been at it since Thursday...
Only Neutron has illustrated something sensible.
2 Neutron. The next step is to see the interval distribution of the "second" process. Show me, please!
rsi:
The next step is to look at the time interval distribution of the "second" process.
I'm digging.
The interesting thing is that the equity of the TS based on this idea is really creeping upwards. I am looking at the randomness of it. It's clear that it's probably not, but it's fascinating!
And with non-linear regression analysis, what can you determine in this case?
To begin with, the task must be broken down into parts:
1. Based on the fact that if we have optimized the TS, its successful forward tests are crowding in certain places, i.e. if the profit factor is 1.6 or higher, the forward test is most likely to be a drain, if it is below some value, then it is also a drain. If drawdown at the optimization point exceeds a certain limit, the forward test is most likely to be unsuccessful. The same is with the expected payoff. Unimportant, as compared to spread, expected payoff in optimization results leads to unsuccessful forward tests. I.e. we have some dependencies between optimization results and forward test success and therefore, we need to be more accurate. Let's look through the reference books to find an appropriate (contextually relevant) method for studies. We found that in our context, logistic regression followed by ROC analysis seems to be appropriate, i.e. it allows to calculate the probability of success of an event (forward test) depending on its features (fitting parameters). Theoretically it fits, although I am not sure, as the most common logistic regression is linear, and it would be better to bring it to a non-linear form. But this is speculation so far, it is quite possible that linearity may be more than enough.
2. we have the initial segment of the forward test and we need to build its mathematical model using a non-linear regression for extrapolation, for example, using OLS or power polynomials approximation.
3. We have the data in item 1 and the model in item 2. Examine the model in item 2 for a deviation (residuals in econometric terminology) of the known forward segment from the model. Investigate characteristics of the known interval of the forward test and its deviations and take data from step 1 and perform analysis, for example using the logistic regression described above, calculate a probability that the forward test is not exhausted and has a sufficient potential for profitable trading in the future (otherwise, we should again re-optimize everything and look for another successful forward test).
This is roughly the plan for forward test research.
To begin with, the task must be broken down into parts:
1. Based on the fact that if we have optimized the TS, its successful forward tests are crowding in certain places, i.e. if the profit factor is 1.6 or higher, the forward test is most likely to be a drain, if it is below some value, then it is also a drain. If drawdown at the optimization point exceeds a certain limit, the forward test is most likely to be unsuccessful. The same is with the expected payoff. Unimportant, as compared to spread, expected payoff in optimization results leads to unsuccessful forward tests. I.e. we have some dependencies between optimization results and forward test success and therefore, we need to be more accurate. Let's look through the reference books to find an appropriate (contextually relevant) method for studies. We found for example that in our context, logistic regression followed by ROC analysis seems to be appropriate, i.e. it allows to calculate the probability of success of an event (forward test) depending on its features (fitting parameters). Theoretically it fits, although I am not sure, as the most common logistic regression is linear, and it would be better to bring it to a non-linear form. But this is speculation so far, it is quite possible that linearity may be more than enough.
2. we have an initial segment of the forward test and need to construct a mathematical model using a non-linear regression for extrapolation, for example, using OLS or power polynomials approximation.
3. We have the data in item 1 and the model in item 2. Examine the model in item 2 for deviations (residuals in econometric terminology) of the known forward segment from the model. Investigate characteristics of the known interval of the forward test and its deviations and take data from step 1 and perform analysis, for example using the logistic regression described above, calculate a probability that the forward test is not exhausted and has a sufficient potential for profitable trading in the future (otherwise, we should again re-optimize everything and look for another successful forward test).
This is roughly the plan for forward test research.
Everything is fine if your TS converts a non-stationary quote into a stationary profit.
Is it a bad thing if the profits are non-stationary?
To begin with, the task must be broken down into parts:
1. Based on the fact that if we have optimized the TS, its successful forward tests are crowding in certain places, i.e. if the profit factor is 1.6 or higher, the forward test is most likely to be a drain, if it is below some value, then it is also a drain. If drawdown at the optimization point exceeds a certain limit, the forward test is most likely to be unsuccessful. The same is with the expected payoff. Unimportant, as compared to spread, expected payoff in optimization results leads to unsuccessful forward tests. I.e. we have some dependencies between optimization results and forward test success and therefore, we need to be more accurate. Let's look through the reference books to find an appropriate (contextually relevant) method for studies. We found for example that in our context, logistic regression followed by ROC analysis seems to be appropriate, i.e. it allows to calculate the probability of success of an event (forward test) depending on its features (fitting parameters). Theoretically it fits, although I am not sure, as the most common logistic regression is linear, and it would be better to bring it to a non-linear form. But this is speculation so far, it is quite possible that linearity may be more than enough.
2. we have an initial segment of the forward test and we need to build its mathematical model using a non-linear regression for extrapolation, for example, using OLS or power polynomials approximation.
3. we have the data from step 1 and the model from step 2. Examine the model from step 2 for deviation (residuals according to econometric terminology) of the known forward segment from the model. Investigate characteristics of the known interval of the forward test and its deviations and take data from step 1 and perform analysis, for example using the logistic regression described above, calculate a probability that the forward test is not exhausted and has a sufficient potential for profitable trading in the future (otherwise, we should again re-optimize everything and look for another successful forward test).
This is roughly the plan for forward test research.
Even though I've been familiar with MQL and programming in general for only two months and I don't know a lot, but I still think it's technically very difficult to test this idea using MQL.
Here we need the tester to call itself when testing. Although, alternatively, you can use testers from two terminals, if it is possible (and if it is at all possible to start tester from Expert Advisor).
And if the profits are unsteady, is that a bad thing?
ISC's assessment is untenable.
Don't be ridiculous! What has ISC got to do with it? What does stationarity and non-stationarity have to do with it?
You're picking up words and dropping them everywhere you go... You've heard the sound, but you don't know where it is.