Machine learning in trading: theory, models, practice and algo-trading - page 814
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
It has long been noted that when the futures are new, the TCs are small and short-lived. The older the futures, the more predictable they become, and when they expire, it's a piece of cake.
And who speculates on new futures? Only the last three months. If the previous one is over (or 2-3 days before) - go to the next one.
And then it's about the same all 3 months except for the last days of existence. No - the older...))
and with your attitude, you're gonna have a long time to make contact with the local population.
He doesn't need to, he's just here to troll. There are literally piles of grail algorithms in the thread, if he was not tongue-twisting, and tried them - would have long come out of the permanent moose. He even posted here almost 90% of ready-made grails, but in order to finalize them he needs knowledge that he lacks. All the missing steps are described here in the thread, but he told everybody, who tried to help him and direct him in the right direction - to fuck off )))))
Ironic.
He doesn't need to, he's just here to troll. There are literally piles of grail algorithms in the thread, if he did not move his tongue, and tried them - would have long been out of the permanent moose. He even posted here almost 90% of ready-made grails, but in order to finalize them he needs knowledge that he lacks. The missing steps are described here in the thread, but he sent everyone who tried to help him and direct him in the right direction - fuck off ))))).
Ironic.
oh teacher, give me the remaining 10% and I will serve you faithfully
Forgive the stupid student who didn't see the spark of truth in your posts.
Patents denied.
The pictures are beautiful, of course.
But in simple terms, you can do this and get that. You can do it without pictures. I trust people that way.)
Saw Shura, they're golden. https://www.mql5.com/ru/articles/2930
If you have at least one predictor with the distribution shown, then you do not need anything: urgently go to a warm island and live there.
Usually the picture is like this:
And here's an absolutely gorgeous one.
Here is the reality of hard life with real predictors.
If you have at least one predictor with the distribution shown, then you do not need anything: urgently go to a warm island and live there.
Usually the picture is like this:
And here's an absolutely gorgeous one.
Here is the reality of real life with real predictors.
By Probability Distributions we mean Bayesians. I'll write later if the topic turns out to be interesting, for now I don't know...
and did you mean probability distributions relative to the target on the OOS?
Probability Distributions refers to Bayas. I'll write later if the topic turns out to be interesting, for now I don't know.
and did you mean probability distributions relative to the target on the OOS?
Writing for the hundredth time.
I take the predictor and divide it into two parts for a target of two classes: one part belongs to one class and the other to another. Then we build two curvilines and superimpose them. Underneath them, we make a caption that says, "f*ck you, not the money."
That's the whole job.
PS.
These curvulines are constantly moving relative to each other, for one predictor less and for the other more than the width of the curvuline. This determines the non-stationarity of the input data for the classification models, any.
I'm writing for the hundredth time.
I take the predictor and divide it into two parts for the target of the two classes: one part belongs to one class, and the other to the other. Then we build two curvulines and superimpose them. Underneath them, we make a caption that says, "f*ck you, not the money."
That's the whole job.
PS.
These curvulines are constantly moving relative to each other, for one predictor less and for the other more than the width of the curvuline. This is what determines the non-stationarity of the input data for the classification models, any.
Now take for each predictor a historical by\sell\hold estimate, translate it into a probabilistic estimate
take several predictors, do the same for each of them
find conditional probabilities of profit
and then you feed it into NS or fuzzy sets, like in this example
the average estimation will fluctuate around 0.5 for each predictor, but the wonders of the Bayesian approach will bring totals to an acceptable level
this in theory :)