Machine learning in trading: theory, models, practice and algo-trading - page 3738

[[Deleted]]

Oh, that's funny. I divide it up too. What criteria do you use to divide them apart from volatility? Of course, there are differences in volatility, but dividing by them does not give a meaningful result, for some reason

[Ivan Butko]

Aleksei Stepanenko #:
We may be right in our entry, or we may be wrong, and in this case we should look for solutions based on our specific circumstances, not only on general market patterns, until we close this trade. That is, we should teach the neural network not regularities, but behavioural strategy. Without pretending to be super knowledge, it seems to make sense

This is reinforcement learning.

It is an extremely complex topic in terms of constructing an agent's policy.

But, it is important to take into account that stop-losses should be structural, not statistical, because the latter have no intellectual meaning.


Here is an example of short structural stops

Smartmakers



Snipers




That is, a stop-loss is meaningful and is a part of the structure, like a cancellation of a development scenario or some kind of force majeure.

That is why you should first form an informative field from the price chart, and then models should work with it.

Any attempt to make stop-loss undesirable leads to self-deception, to crutches and half-measures. Yes, a model without stops or with huge stops can work on the return to the average, but roughly, dirty and not always.



Besides, the same wavemakers have such a concept as resonance. Not for ellioticians, but for more competent ones, who are not tied to shitty waveforms. Price builds itself, but sometimes the "snake" crashes into its tail. Also in price: there comes a moment when the price has built such structures that at all wave levels they should statistically go in one direction.

A resonance effect occurs and the price shoots out an impulse, allowing to set a short stop and huge targets.


And the task in this case comes down to how to prepare the price chart for structural formatting, in order to then feed this information to the neural network.

[[Deleted]]

Aleksei Stepanenko #:
Oh, that's funny. I divide it up too. What criteria do you use to divide them apart from volatility? Of course, there are differences in volatility, but dividing by them does not give a significant result, for some reason

Depending on the TS and not every trading symbol will work. You need to go through the options.

[Inquiring]

Ivan Butko #:
Besides, the same wave-makers have such a concept as resonance. Not for ellioticians, but for more literate ones, who are not tied to shitty waveguides.

Can I clarify, who do you mean?

[[Deleted]]

Inquiring #:
Can I clarify, who do you mean?

Advertising his 3.14 indicators in the topic, spamming these pictures everywhere. It is not clear what it has to do with this topic.

[Ivan Butko]

Inquiring #:
Can I clarify, who do you mean?

Who considers a three-wave as the only structural unit of a wave of a larger movement.



TS GOR
TS Pathfinder
TS Formula Forex

I remembered three trading and analytical strategies in which Elliot was sent to hell and replaced subjective 5 with objective 3.

[Inquiring]

Ivan Butko #:
Who considers the three-wave as the only structural unit of the wave of a larger movement.



TS GOR
TS Pathfinder
TS Formula Forex

This is a memory of three trading and analytical strategies, in which Elliott was sent n*x and replaced subjective 5 with objective 3.

In general, Charles Dow first proposed a 3-wave market movement. 5 waves also look quite logical. Another thing is the selection of coefficients for each wave. That's where the dog lies.

[Inquiring]

If there is a place to look for input parameters, it is in the ratio of waves. Only in the code it is a lot and complicated, - you can't do it alone, and you can hardly gather a team here.

[Ivan Butko]

Inquiring #:
In general, Charles Dow first proposed a 3-wave market movement. 5 waves also look quite logical .

7 waves is also quite logical. If you want to.

Inquiring #:
The other thing is the selection of coefficients for each wave. That's where the dog lies.

An extremum is either there or it is not.

The question is deeper: how to determine the wave levels / orders / seniority / dimensionality / scale / class of these extremums.

At the moment, this question is answered by one well-known crutch - timeframes.

Well, there they are and that's it. Chain Beal's fractals and highlight swings on each TF, like.

But it is more competent to format the price chart by self-similarity and self-organisation (fractal, in short), starting from the atomic level (M1 is enough, an intraday trader does not need to go down to ticks).


But even nowadays, using TFs and ratios of neighbouring mashka periods, I used to identify waves and make jewel-like forecasts of movement (resonance) on the basis of "unfinished" ABC.

Forum on trading, automated trading systems and testing trading strategies.

Forex - Trends, Forecasts and Consequences 2025

Ivan Butko, 2025.02.28 04:52 AM

A good place to make a decision and find entry points on the euro

Taking into account the cascading flat-like price movement, as well as the "sluggish" support of the senior trend, which is active, the level of 1.03700 is a borderline level, where the local blue ABC trend upwards ends.

In addition, the price likes to "prick", popularly called - false breakdowns, round the level, gathering liquidity (stops) and go to new tops.

And the result

And I am not a venerable trader, I just tried to analyse structurally.

In MO you should prepare input data in a structural way and give them to DBSCAN and others without teachers.

[[Deleted]]

Especially for careless three-wave-watchers: the best articles on the subject, so far.

1, 2