Machine learning in trading: theory, models, practice and algo-trading - page 3017

[mytarmailS]

Ivan Butko #:

From extremum to extremum?...

Completely indecent. At least it's down a few points.

Lucky you.)

[mytarmailS]

What is a trend? I asked myself.

First of all, from what point of view?

From the point of view of TA, it is a certain movement that is the strongest in comparison with other movements.

Can it be formalised and programmed? Not really, everything is relative, the wording is not very good, it doesn't carry much information....

If there is a complex phenomenon and it is not so easy to understand it, then

1) it is necessary to break it down into simpler parts.

2) to study not the phenomenon itself, but to create a simplified model of the phenomenon and study the model.

I will do both.

So the market model will be three sinusoids - general trend(slow movement), average movement and fast movement.

like that.

Now we can identify the dominant movements as in the case of technical analysis.

So what are trends from the point of view of this model?

A trend is when all the waves are lined up in the same direction.

For some people it's obvious, but I think it's useful to think about it.

R code if anyone wants to play with the parameters of the movements.

my.sin <- function(ve,a,f,p)    a*sin(f*ve+p)

ve <- 1:1000
a <- 1
f <- 0.02

trend <- my.sin(ve = ve,a*2,f/5,0)
s1 <- my.sin(ve = ve,a,f,0)
s2 <- my.sin(ve = ve, a/2, f*4, 2)

al <- s1+s2+trend


par(mar=c(2,2,2,2), mfrow=c(2,1))
plot(al,t="l",lwd=2) 
matplot(cb,t="l",lty=1) 

[Igor Makanu]

Aleksey Nikolayev #:

The problem is that the tree is not built according to a profit maximisation condition, but according to a loss function convenient for package programming.

So you have an unpleasant choice - either try to reconfigure a complex, tricky package, or build a cramped bicycle. It is also possible to "successfully" combine both of these options)

IMHO, if you choose to fiddle with an existing package on trees, you should try to use pruning (pruning) - with the condition of profit maximisation on the forward, for example. It might be possible to avoid manual fiddling with the rules.

Yandex wrote something similar https://academy.yandex.ru/handbook/ml/article/optimizaciya-v-ml

[Ivan Butko]

mytarmailS #:

So what are trends in terms of this model?

A trend is when all the waves line up in the same direction

Reminds me of a well-known academy about 15 years ago. One of the faculties developed a similar idea, well, and accordingly, made money on it.

They called this phenomenon "resonance" of the price. The correction of a higher wave level ended, and the supposed impulse of the same level began, then the correction ended at a smaller VU and the supposed impulse began, and at an even smaller VU the same thing happened. In the end, the price resonates and goes into a big impulse, which forms the directional movement (trend) familiar to the eye. And the point was to patiently wait for such conditions, ignoring ranges, corrections, flat spots.

So it is potentially a working scheme.

[mytarmailS]

Ivan Butko #:

They called this phenomenon "resonance" pricing.

That's quite a phenomenon.

[СанСаныч Фоменко]

Ivan Butko #:

It reminds me of a well-known academy about 15 years ago. One of the faculties developed a similar idea, and, accordingly, made money on it.

They called this phenomenon "resonance" of the price. The correction of a higher wave level ended, and the supposed impulse of the same level began, then the correction ended at a smaller VU and the supposed impulse began, and at an even smaller VU the same thing happened. In the end, the price resonates and goes into a big impulse, which forms the directional movement (trend) familiar to the eye. And the point was to patiently wait for such conditions, ignoring ranges, corrections, flat spots.

So it is potentially a working scheme.

On the history ....

And behind the right side of the monitor is darkness and devilishness.

[Aleksey Vyazmikin]

mytarmailS #:

So about the paid help, waiting for the question to be answered.

[mytarmailS]

Aleksey Vyazmikin #:

So about paid help, waiting for the question to be answered.

The answer is no.

[Aleksey Vyazmikin]

mytarmailS #:
The answer is no

To the question "Why?"

[Aleksey Vyazmikin]

Vladimir Perervenko #:

folk wisdom says you can't see the forest for the trees. I wonder if you can see a tree by picking leaves. I'm not asking about the forest.

Is this the only algorithm you know? Or is it the most efficient? Why are you fixated on it?

It's a passing thought.

Good luck

What's the point of looking at a tree? :) I don't think one should aim to get 100% Recall.

Yes, since about alternative thinking models, can you suggest a model that combines well (works) discharged (rare units - about 5% in the sample) binary features? There are a lot of them.