Machine learning in trading: theory, models, practice and algo-trading - page 272
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
Because even on such primitive signs, on minutes, you can draw 3-5%, that is to predict the direction with 53-55% probability. It won't work that way with white noise.
I do not know what 3-5% is, but predicting "white noise" with an accuracy of about 50% is easy. You turn on the PRNG and go ahead.
And in pieces - even under 90%!
I do not know what 3-5% is, but predicting "white noise" with an accuracy of about 50% is easy. You turn on the PRNG and go ahead.
And in pieces - even under 90%!
3-5% is an advantage over randomness (53-55%) - 50%, with samples of hundreds of thousands of samples (not chunks).
PS: Dmitry, I see you want to fool around, unfortunately I'm not a good candidate for that. Good luck with the white noise and infinite variance.
...............
For example "levels", could be one of the most important signs, I mean those levels that we see on the chart with our eyes, where people put stops. Try to formalize and program such a sign and check how statistically significant it is.
Everyone sees "levels" in his own way, some people builds quarts, some people build bollinger levels, some people build minimum/maximum and so on. It's difficult to say how to formalize this "most important sign", maybe because it is the most important, like Elliott Waves, it has too many faces and is subjective, that it can be fitted with all formations that predetermine predictable market behavior, but nobody wants to do it with non-predictable ones...
I personally do not draw any "trend lines". I personally do not draw any "trend lines", I use zigzag lines that have not been covered by the price. How to make it a sign I don't know.
Where do you get the date with the stakes from?
Everyone sees "levels" in his own way, some of them builds quadrants, some of them build bollinger, some of them build minimum/maximum, etc., I use ZZ . It's difficult to say how to formalize this "most important sign", maybe because it is the most important, like Elliott Waves, it is too variable and subjective, it can be fitted with all formations that predetermine predictable market behavior, but nobody wants to do it with non-predictable ones...
I personally do not draw any "trend lines". I personally do not draw any "trend lines", I focus on zigzag reversals that have not been covered by the price. I don't know how to do it.
But it's not the only one, I can only say that I found quite a clever way to make these "peculiar points", "attractors", very important features, but I won't deny the respected gentlemen the joy of invention)))
Everyone sees "levels" in his own way, some of them builds quadrants, some of them build bollinger, some of them build minimum/maximum, etc., I use ZZ . It's difficult to say how to formalize this "most important sign", maybe because it is the most important, like Elliott Waves, it is too variable and subjective, it can be fitted with all formations that predetermine predictable market behavior, but nobody wants to do it with non-predictable ones...
I personally do not draw any "trend lines". I personally do not draw any "trend lines", I use zigzag lines that have not been covered by the price. I don't know how to do it.
================================================
Of course levels are important. But it is more important to see forces acting on this timeframe. And we should not forget that each TF has its own resistance and support levels. To process the variety of conditions on many TFs we need models of MO.
How does it look? In attachment.
Good luck
Are there any brokers that don't provide stakes? You get the stakes and you write, that's the only way, it's the most reliable way to get a synchronized data stream from many sources, as far as I know everyone writes the data themselves, those who moderate HFT trades.
I asked the wrong question, I meant how / by what do you write the glass? with your software or . ?
Well, yes, the nearest non-crossed zigzag knees like hai\loi traders put stops and limiters, but not only there, I can only say that in a rather artless way, I found how these "special points", "attractors", make very significant features, but I will not deprive the honored gentlemen the joy of invention)))
Why not? I would love to hear about it. Other people's experience saves you time.
And another question, how did you measure this significance that would then know that they have become more meaningful?
it looks strong! but it's also not clear (
It looks strong! but it's also not that clear (
I agree. But to explain how it works you need to write a considerable article. And the time is disastrously short.
The indicator is available on the Market. But the example I gave was not for advertising purposes, but to show the complexity of analysis.
Good luck
1) I asked the wrong question, I meant how / by what do you write the glass? with your software or . ?
2)Why not? I would love to hear about it... I would love to hear about it... someone else's experience saves my time...
3)And another question, what do you measure this significance that would then know that they have become more meaningful?
1) My software, of course))
2) About the features of the levels, I'm not going to spoon-feed them, I've already said a lot. If you are a genuine trader, you should have an irresistible motivation to do it yourself and a whole bunch of ideas how it can be implemented in your system of abstractions, starting with the lowers, pivot points, etc. You know where you put the stops, according to what algorithm, these points are "attractors", or rather anti-attractors, from them the price bounces, from the top and bottom.
3) Checking the significance is as simple as a bagel. You have a model, (re-)teach it with and without this attribute, calculate the difference in accuracy, it will be the contribution of the attribute to the model.