Machine learning in trading: theory, models, practice and algo-trading - page 2710
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
I don't have anything else to use, so I'll use them. Time series classification hasn't gone anywhere. It uses its own terminology, such as attributes rather than events.
Events are described by traits.
A trait/predictor/feature can be either a set of data or the result of processing that data.
Events are described by attributes.
A trait/predictor/feature can be either a set of data or the result of processing that data.
STOP IT.
You don't want to learn and expand your consciousness - fine - everyone's choice.
I think I understand what people don't understand, here's the diagram, it should be clearer.
(about the picture) to make it work, prices should be read from right to left :-) and this is not a joke
I think I understand what people don't understand, here's the diagram, it should be clearer.
This picture is your idea?
(about the picture) to make it work, prices have to be read from right to left :-) and that's not a joke
I'm not sure I understand you, but if the point is that recent events have a greater impact more often than past events, then I agree.
And so, the system is closed loop - the input there is also data from the chart.
this picture is your idea?
The idea is more about implementation, the picture shows what data we are talking about.
The idea is more about implementation, the picture shows what data we are talking about.
I'm not sure I understand your point, but if the point is that recent events have a greater impact more often than past events, then I agree.
And so, the system is a closed system - the data from the graph is also fed to the input.
When the "event" comes (becomes obvious or reflects in the past calendar) it is too late to rush and trade something somewhere. In your picture "price has reached ATR D1" is already the consequence, the final, serious people have made all purchases/sales, then you can trade only noise.
That is the scheme as if it corresponds to reading in reverse time, from right to left: "price is going somewhere" -> "bought/sold" -> "damn, zigzag turn" -> "news, emotions" :-) .... but hypothetically it can be used by feeding quotes into it also from right to left, then the revealed consequences are close to the sought predictors and some of them can even be used.
And what's the idea behind the implementation?
Typing Events. For each type of Event we look for a different type of Catalyst - probably an indicator, i.e. something basic that will be at the root of the rule/list. Then we look at the behaviour of this indicator in three spaces and count the metrics for each space, if the metrics are within the required limits, we save the indicator settings and space settings to a file.
After the data set of such settings and spaces, which are essentially a tree stump or a simple set of rules (I don't know yet whether to hit the rules here at once - to carry out the procedure of binarisation), we connect other predictors (description elements) of the Event peculiar to this type of event. At the output, we recalculate the metrics and do the final screening - the predictor is ready.
This is in brief.
Further grouping of the resulting predictors with the purpose of screening or obtaining the average value, maybe using MGUA, but I don't know how to implement it in MT5.
What are the questions - what is the best way to deal with the target. So far I think that we should take a neighbourhood near ZZ extrema, and if an event has already been identified in the neighbourhood, then we should not count the metric for its rules/leaves again.