Machine learning in trading: theory, models, practice and algo-trading - page 3575
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
Some time ago brought to the demo account TS with the teacher of price increment with classification error below 20%, or even 10% .
And here it became clear that my simple teacher taught the model to predict a profit of 10-15 pips, but was wrong on price movements over 50 pips.
Finally...
Estimating models by probability (osh. kl.) makes sense only if TP=SL. This is too small a set of strategies.
In other cases, you should evaluate by loss to profit ratio.
Earlier you emphasised exactly this 20-10%. Billing it as a meaningful achievement. Glad it has shifted to profit.
Sanych, a teacher is a person, not a target.
You are stubborn (like everyone else here), I thought we discussed the topic six months ago. There is no point in repeating it.
You are stubborn (like everyone else here), I thought we discussed the topic 6 months ago. There is no point in repeating it.
I am not stubborn, it just creates a mess in the head of the writer and the reader.
Usually you need to repeat 8 times to make a person understand. Statistics.Sanych, a teacher is a person, not a target.
and you also deliberately emphasise your illiteracy by specifying"construct a teacher (target variable)".
There is learning with a teacher and there is learning without a teacher. There is no other.
There is learning with a teacher and there is learning without a teacher. There is no other.
Also wrong, there's reinforcement learning.
Finally...
Estimating models by probability (osh. cl.) only makes sense if TP=SL. This is too small a set of strategies.
In other cases, we should estimate by the ratio of losses to profits.
Earlier you emphasised exactly this 20-10%. Billing it as a meaningful achievement. Glad it has shifted to profit.
You are stubborn (like everyone else here), I thought we discussed the topic 6 months ago. There's no point in repeating it.
It's not accurate.
If we're talking about classification valuations, there are no earnings valuations out there. There are their valuations and their packages (ROC).
If we talk about CU, the valuation is only earnings and a number of other metrics. By the way, all except P-squared.
Also wrong, there's also reinforcement learning.
With "reinforcement" - a variation of "with a teacher", when a previous result is taken as the teacher and targeted learning takes place. "Without a teacher" is when it is simply broken down into subsets.
With "reinforcement" - a variation of "with teacher", when a previous result is taken as a teacher and targeted learning takes place. "Without teacher" - when it is simply broken down into subsets.
No
One more time (hopefully the eighth). Learning with a Teacher is learning on human-prepared data (traits + labels). That is, it is known in advance what is being taught.
For example, learning to solve the multiplication table.
Good replacement for vpn, especially for those who use different LLMs.
I don't get it - the link is a collection of software for local PC.
I don't get it - the link is to some collection of software for local PC.
https://www.comss.ru/page.php?id=7315