Machine learning in trading: theory, models, practice and algo-trading - page 122
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
Trolling is cool)))) hilarious...
Not for everyone. What it's for, said. It can also indicate a failure of the algorithm or etc. Well, let's not
about sad))) Nickelodeon likes it (just look at the crap it puts out first))), you like it, that's fine))) Automatic
scaling happens?, fuck it!, you can shove anything you want))) hilarious... But it's not pies... fast food)))
I looked it up for interest... On the same data, I get trend=93.2%, test (applying the model to new data) =92.7%.
"Fastfood_#9" (Tested: Load-> create model-> screenshot-> reload-> create model...) gave the following results-
WHAT IS THAT?
My point is that you need to understand what this neural network is and what to feed into the inputs.
So far you haven't gotten any further than 25,000 patterns to inputs.
And you should think about it, maybe the neural net just stupidly learned them.
Or found a bigger chunk and learned from it?
My point is that you need to understand what this neural network is and what to feed into the inputs.
So far you haven't gotten any further than 25,000 patterns to inputs.
And you should think about it, maybe the neural net just stupidly learned them.
Or found a bigger chunk and learned from it?
I don't know what you're trying to say. Can you rephrase that?
I know my model perfectly understands my data, but the new price reacts differently to the old patterns I've learned - often the opposite
I know that personally my model understands my data perfectly, but the reaction of the new price to the learned old patterns is different - often the opposite
And how do you form patterns?
Well, it's the same as everyone else... The neural network forms during training
Well, it's the same as everyone else... A neural network forms a pattern when you learn it.
A neural net forms a pattern? - I thought that the neuronet only learns how to react to patterns correctly. A pattern is a current market situation described by indicators, geometric drawings or other price manipulations.
So, how do you describe a pattern? - The market reacts differently to the patterns in the future, isn't that because there were no patterns at all, or rather the way you described it as a pattern is not what the market actually reacts to.
There is another very important point. If the network is trained for too "hard" answers, then in the future the slightest deviation from the correct answer will be an error. Try to formulate more "soft" answers, then things will be more fun.
1) A neural network forms a pattern? - I thought that a neural network only learns how to react to patterns correctly. A pattern is a certain current market situation described by indicators, geometric constructions or some other price manipulation.
2) So how do you describe a pattern? -
3) The market reacts differently to the patterns in the future, isn't that because there were no patterns at all, or rather the way you described it in the form of patterns is not what the market actually reacts to.
4) There is another very important point. If the network is trained for too "hard" answers, then in the future the slightest deviation from the correct answer will be an error.
5) Try to formulate more "soft" answers, then it will be more fun.
1) We speak different terminology.... predictors (information) are fed to the neural network, usually in the form of BP, as a rule indicators, they are not patterns, they are predictors....
Then the network when training on data divides this data into groups of similarity or clusters or patterns is what I meant when I said pattern...
2) I understand that you call predictors as patterns, if so I will answer using your terminology
3) If there were no patterns it would be chaos and random and the output of my network would not be a function that is almost 100% inversely correlated to the price(the reverse of the learned patterns), but just some random function, unrelated to anything, and there is no such thing...
4)This is retraining #### misunderstood the line..... Yes, you are absolutely right, I am thinking about it, but so far it is not working, which is why I test my model exclusively on trading on new data, not on recognizing new data but on trading on new data
5) My target (answers) are reversals, how can they be mitigated ? I tried to make a target that imitates a take profit and stop loss but i got the same softening, but i didn't get any interesting results, maybe i didn't look hard enough, i don't know.
1) If there were no patterns there would be chaos and randomness and the output of my network would not be a function that is almost 100% inversely correlated with the price(the inverse reaction to the learned patterns) but just some unrelated, random function, but there is no such thing...
I have a target (answers) which is reversals, how can i soften it ? I tried to make a target that imitates take profit and stop loss, but i got the same softening, but i don't get any interesting results, maybe i didn't look carefully, i don't know.
i did not get an answer to this question, how do i build/detect a pattern? - I realize that the question is probably too intimate, you don't have to answer.
2. Turns - this is not even too "hard" an answer, but in general from the category of "I don't know where from and I don't know what". Here's a reversal on the next candle, no? - One more? - No, wrong! - Maybe on the fourth candle there will be a reversal? - Yes, a reversal, 150 points passed, and it turned back, but no, it wasn't a reversal but a correction, though it was a reversal nevertheless... There's no way to define "pivot"! - It means there is no possibility to teach how to detect it, not only beforehand but even in the current moment.