Machine learning in trading: theory, models, practice and algo-trading - page 2766

New comment
 
СанСаныч Фоменко #:

What's all the fuss about?

Do you not have a "prediction error - window size" graph? Or any other similar information?

We're basing this on the properties of fractals, which have been discussed before. Hands used to trade pretty well back in the day, until I got bored of it.

like suggesting the topic and doing it myself
 
Maxim Kuznetsov #:

new section of the special Olympiad: one NN trades on ticks, the second predicts the results of the first one, and the third one warns of the second one's fakups. (it can be combined indefinitely). The battle is fought on bets and forward, because even on demo it's a bit scary to run this game.

I had a similar idea...)))) I wanted to teach the neural network to change the periods of mashka, say, improved AMA.

 
I don't think anything is worth trading right now. What's happening now has never happened before. There are no examples for the D.O.D. And what's next is unknown.
Plus everything is manipulated. Why didn't the Nord Stream explosions bring down the Euro? On the contrary it is rising today?
 
elibrarius #:
I don't think anything is worth trading right now. What's happening now has never happened before. There are no examples for the D.O.D. And what's next is unknown.
Plus everything is manipulated. Why didn't the Nord Stream explosions bring down the Euro? On the contrary it is rising today?
The Baltic pipe has started up, they don't need that pipe now.
And guess who gave Norway that hydrocarbon rich shelf 😀
And just by that time they had pumped the storage facilities to the max, so everything is logical. Business and nothing personal.
 
Ivan Butko #:

I had a similar thought...))))) I wanted to teach the neural network to change the periods of the mashka, like an improved AMA.

The application of neural nets for their classical purpose is hindered by a categorical lack of images and even understanding of what to count with them

There's nothing to train them on.

Those who think that you can take a bunch of ticks and feed an open-close sample into a neural network have never encountered pattern recognition.

There is training on tens of thousands of positive examples and hundreds of negative ones, and the "image" itself is very strongly transformed (on the basis of science: physics, mathematics and psychology) and processed before being fed to the neural network.

And we have no sufficient volume of objects for training, no expert assessment of their quality, no theory on which they are selected and what they are in general. There is nothing to train on and nothing to objectively check the result on (forward-back, it's a feedback loop, you can't do that).

By the way, the same applies to MO: DL and NN are only methodology of methods, application tools. They are powerless to find the unknowable in who-knows-what.

 
Maxim Kuznetsov #:

By the way, the same applies to MO: DL and NN are just methodologies, application tools. They are powerless to find the unknown in the fucking thing.

It's all about information processing algorithms/trait formation.

If you feed the last 10 candles, it is clear what will happen, even if you train GPT-5.
 
mytarmailS #:
It's all about information processing algorithms/trait formation...

If you feed the last 10 candles, it is clear what will happen, even if you train GPT-5.

Did it work? I mean, what if you feed 10 candlesticks for training and get not 50/50 as a result?? on demo at least

unlikely...and there is no grounded theory to measure something, and why 10, why exactly these candlesticks, why exactly now...what was expected and how exactly GPT-5 helped.

even down to the basic "evaluating the result of an experiment". Without vs with GPT - how can it be evaluated? Whether the experiment was successful or not?

 

for the last six hours, everyone's been up against the dollar.

Is it an evening pullback? Is it a reaction to critical news? Is it just a fluctuation, or did the Fed decide not to hike rates the day before?

No one who looks at just a quote on a single instrument will be able to answer that. From the point of view of an individual quoter/sticker all options are equal.

PS/ about windows and reference points: from the point of view of 10-year bond holders, the 5000p rise in XAU today is a load of rubbish that is not worth mentioning. And someone must have lost their deposit.

 
I don't know what to say.
 
Maxim Kuznetsov #:

for the last six hours, everyone is against the dollar.

Is this an evening pullback? Is it a reaction to critical news? Is it just a fluctuation, or did the Fed decide not to hike rates the day before?

No one who looks at just a quote on a single instrument will be able to answer that. From the point of view of an individual quote/sticker all options are equal.

PS/ about windows and reference points: from the point of view of 10-year bond holders, the 5000p rise in XAU today is a load of rubbish that is not worth mentioning. And someone must have lost their deposit.

It's a bit of a rambling exclamation point. Either the fundamentals are analysed or the market reaction. Often both when trading manually. Patterns sometimes anticipate some fundamental events not in terms of information, but in terms of their significance and impact on the market.

But most traders, like humans, are irrational, subject to emotion. They cannot concentrate. And therefore vulnerable to risky events 😀

That's why they try to shift the responsibility to bots, because they have proved to themselves over time that there is no way to rely on themselves. In periods of particularly emotional states they like to argue without backing it up with deeds. Psychologists would say it's from their own insecurity. But in fact, the person was fooled, lured into doing something he does not understand, made a slave, zombified... by external forces.
1...275927602761276227632764276527662767276827692770277127722773...3612
New comment