Machine learning in trading: theory, models, practice and algo-trading - page 2359
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
Can MO calculate the quality of signals by probability? To then filter out those with more than 90% probability?
There are already probabilities in the model output, but they are pseudo, i.e. have no relation to the general population.
you can filter through thresholdAutoMO implies going through the models and choosing the best one
What's the point if catbust beats everyone on all datasets?
https://mljar.com/machine-learning/compare-ml-algorithms/
Well, here's a confirmation that models don't solve anything, the difference between models is ~5%...
Only signs and ways of presenting information are decisive...
But fools still believe in models, love models, pray for overseas articles with new models, is there anything easier than to train a model on "raw" data? :)) You don't have to think or know anything, just copy the code and you're a cool MO trader in the MO branch)))
WAKE UP!!!!
There are so many signs and methods of processing that you don't have enough computing power to check everything, even a group of people!!!
Are your brains so screwed up that you see nothing but returns and multiple models...
Study DSP, system modeling and other sciences, one MO without knowledge, it's just the coolest fitting and nothing more...
Well, here's a confirmation that models don't solve anything, the difference between models is ~5%...
Only signs and ways of presenting information are decisive...
But fools still believe in models, love models, pray for overseas articles with new models, is there anything easier than to train a model on "raw" data? :)) You don't have to think or know anything, just copy the code and you're a cool MO trader in the MO branch)))
WAKE UP!!!!
There are so many signs and methods of processing that you don't have enough computing power to check everything, even a group of people!!!
Are your brains so screwed up that you see nothing but returns and multiple models...
Study DSP, system modeling and etc science, one MO without knowledge, this is just the coolest fitting and nothing more...
And can you without fooling around?)))) it has many nicer synonyms)))
Anyway their science has more money and therefore not us ahead yet(
Processing raw data is also a model. And of course testing a model is not understanding it)
Can you do without fooling around?))) It has many nicer synonyms)))
No offense, it's not to you at all...
It's for those who read western articles and think that for example "GPT-3" will tear up the market...
At the entrance, of course, 10 returns in a sliding window, you have no brains for more, but what else? The network will think of everything, yeah...
In any case, their science has more money and that's why we're not ahead of them yet(
That's not what I mean...
There are two problems
1) "Information starvation" in models, those are few and bad signs
If you want to predict the process, but you have signs which describe only 5% of the process, you should train at least a 100-layer, 8 times convolutional mega-duper-super-ultra-extra GPT-5.
The output will be the same prediction with an error of 95%.
And people don't understand that, but fall for the architecture, and what do you call them?
conclusion the solution to the problem is not in the MoD
2) Features do not live long, they lose their useful properties very quickly, no MO can see the dynamics of usefulness of their features, you should work with your brain, not with models.
Conclusion the solution to the problem is not in the MO.
Processing of raw data is also a model. And of course, testing of a model is not its understanding.)
Well yes, the question is about adequacy...
When a plane flies over my head and I buy a euro it's also a model ...
Yeah, it seems that doing ME sometimes leads to a nervous breakdown.
maybe someone else will try to explain the sacred meaning of AutoML?
Yeah, it seems that doing ME sometimes leads to a nervous breakdown.
maybe someone else will try to explain the sacred meaning of AutoML?
can someone else try to explain the sacred meaning of AutoML?
Apparently, it's something like those "two from the box, identical in face" from the cartoon, which will do everything for us.)
maybe someone else will try to explain the sacred meaning of AutoML?
Marketing.
I.e. the task is to win crumbs of quality gain through a significant increase in time, model enumeration
but there's also auto preprocessing and auto exploratory analysis