Machine learning in trading: theory, models, practice and algo-trading - page 1770
![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
Delta is the difference between buyers and sellers during a bar (as an example) if the difference is positive, then there were more buyers, if negative, then sellers. Please check out the service clusterdeltalkak, there are even paternas, by the way the only market paternas that pay attention, because they are stock, not this all of yours.... hammer or takeover..... as an example...
If you look for patterns, all the data that you have are important for the purity of the experiment and research, so not taking into account the volume and delta is wrong in general. The tick volume or any other volume, this is only data. It is not correct to discuss their correctness and accuracy without analysis. Why are you so emotional? I have been working on this idea since year 15, 16 or something like that, it is a very short period for results and emotions in such problems. What feeds is already good))))
1) you choose your own ZZ parameters, you can for example by max profit, it all concerns the global target, but in the local targets you can use hundreds of parameters of different ZZ, it would be good
2) Predictors are different, if you have the sense to take memory into account, you should take it into account.
3) First: We can make predictors with local targets that refine the entry point (the target is not the direction of the PZ, but a reverse/non-reversal, etc.)
Second: The better we get to the global target, the more accurate the entries will be
I've made a screenshot of ZZ forecast on the chart somewhere, if I find it, I'll paste it here
I found#17408 I think if I increase the quality to 95% it will be great.
But doing it alone is boring, and ideas are running out, so I came up with the idea - a public dataset, but realized that one (.
4) It is possible, but remember you are working with a local attribute, it's just a part of the puzzle which helps to put it all together
5) I have a simple script that always unloads quotes from mt4 into tcht at the end of a candle, then I read it from R and do what I want )
I get an accuracy of 97%. But what is the use of it? One of my predictors is vector ZZ on the current bar - I think it will give 70% of correct classification....
The training for 2017-2019 is minutes, and the independent sampling is 2020, the tool is Si glue.
As a rule, the market-neutral start-ups are used by banks which earn 0.1% but in money it is enough to cover the operating expenses and not only. And not infrequently they do so at the expense of depositors, because the strategy is a no-lose.... don't you think?
There is no evidence. There is a lot of frequency trading in the public that transactions are milliseconds and like the reports of profile agencies that such trading takes almost half of the market, but there are no algorithms, experts, courses.
There is no evidence. There is a lot of frequency trading in the public that deals milliseconds and like the reports of profile agencies that such trading takes almost half of the market, but algorithms, experts, courses are not visible.
I got an accuracy of 97%. But what is the use of it? One of the predictors I have is the vector ZZ on the current bar - I think it alone will give 70% correct classification....
Training on 2017-2019 - minutes, and independent sampling 2020, tool is Si glue.
didn't get the shit, akurasi on new data ? 2020 ?
I do not know, I work within M15 positionally and do not worry. I'm quite well fed.... Moex, just a fairy tale after 15 years of forex. I think, I should go here and voila, and it goes on as usual. But maybe it's just my luck... although... I'm going to drink some more beer... :-)
Throw me a link to your monitoring. Show how you have voila and how this experience is not lost. It is interesting to see if you have written.
I do not know, I work within M15 positively and do not worry. I'm eating quite well.... Moex, just a fairy tale after 15 years of forex. I think, I should go here and voila, and it goes on as usual. But maybe it's just my luck... although... I'm going to drink some more beer... :-)
Alcoholic, you...cunt ))))
hawthorn maybe better? just for doshik ))))
Throw me a link to your monitoring. Show how you have voila and how this experience is not lost. It is interesting to see if you have written.
Alcoholic, you...cunt ))))
hawthorn maybe better? just for doshik ))))
I have something similar in mind:
...and makes a prediction about the destruction of the pattern. Probably at least two NSs should work. SOM classifies similar vectors (vectors from a floating "visible" window of some width (in time)). We work, for example, with the most dominant class. Well, and MLP, can already predict fidelity/falsehood of the new pattern, based on statistics (for example) of such repetitions, taking into account all sorts of factors.
About cooperation, then, I don't mind. Rather, get tired of my inefficiency. My code formulation and debugging process stretches sometimes for a long time...
What Alexey Martyanov said a long time ago, I think will be relevant for a long time to come... https://www.youtube.com/watch?v=0BwEeOyUa9w&t=734s
You're thinking right, the way you think and the way you think is right, credit for understanding Maitreyda too !
It's necessary to forecast the market maker, if it works, it will be something like this
I think that the tools you have are not very good.