Machine learning in trading: theory, models, practice and algo-trading - page 3444
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
then I don't understand why there's such a difference in results
I have two assumptions - either it is the effect of forced randomisation (the numbers change), or there is an effect from different normalisations. I also did not expect such a result.
But how to check which is the most important of these variants is a mystery.
I have two assumptions - either it is the effect of forced randomisation (the numbers change), or the effect is due to different normalisations. I also did not expect such a result.
But how to check which is the most important of these options is a mystery.
I tried it on my own data - any normalisation worsens the result for me )
I'll double-check later, if I remember.I tried it on my own data - any normalisation worsens the result for me )
I'll double-check later, if I don't forget.I should note that I used sequential clustering - a tree 3 levels deep with 3 clusters per leaf - maybe this has some effect, but I'm not sure.
Below in the graph on the x-axis is the probability shift for the cluster as a whole for the train sample, and on the y-axis is the financial result in the cluster for the three samples. For the train sample, the financial result is reduced by a factor of 3 because it contains three times as many samples, and we wanted comparability in the graph,
Yes, not everything is so perfect, and with a small shift to the negative probability zone it is possible to make a profit (here the markup is for catching trends), but in general the slope of the regression lines for each subsample captures the dependence of profit on the probability shift in the cluster.
A year ago I was implementing MO in a custom EA, now I saw it and decided to see how the model worked.
At the beginning, the results without the model - the period from 2023 to the current date.
And now with the model
I was implementing MO in a custom EA a year ago, now I've had a go - saw it - and decided to look at how the model worked.
At the beginning, the results without the model - period from 2023 to the current date
And now with the model
What does it mean : I implemented the model, here is the result without the model
There was an Expert Advisor trading according to its own logic - the model became an additional filter - the last instance giving the right to open a position. The approach that I use in general - a signal according to a clear logic and its confirmation by the model.
There was an Expert Advisor that traded according to its own logic - the model became an additional filter - the last instance giving the right to open a position. The approach that I use in general - a signal according to a clear logic and its confirmation by the model.
There was an Expert Advisor that traded according to its own logic - the model became an additional filter - the last instance giving the right to open a position. The approach that I use in general - a signal according to a clear logic and its confirmation by the model.