Machine learning in trading: theory, models, practice and algo-trading - page 1212
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
Of course, if there is a clear hopelessness, then you can remove
Thank you for your opinion. It is not clear whether it is "promising" or not, but in my right mind I would not choose such a model, which let's say drawdown by 50%, though it is easy to come out in profit afterwards.
I'll add page analysis via google yandex search engines to it
Why? What is the point of the analysis?
Why? What is the point of the analysis?
That's the troll from the next thread.
That's the troll from the next thread.
aah)))
Does anybody know if I develop some cool neural network algorithm on Java and attach it to google yandex search engines, will nobody "lock" me up?)
Ahh)))
One guy's asking about how to filter his orders, and the other one is answering with an article on predicting trends by accidental forests... :)) After that the first one immediately intends to write his own neuronet in Java :))) So, here they are
and the crazy green-eyed kitty just quacks... it's impossible to read without smilingThe figures show the financial result (y-axis) of the model when choosing different probabilities for the binary classification (x-axis). On the test sample it turned out that one should always enter the market when an activation signal appears (training decides whether to enter the market or not). The resulting paradox is that training only worsens the basic activation signal and I would not have seen it, if I hadn't decided to see how the financial result changes depending on the shifting of the classification point on the segment of probability.
We have a very different approach to the problem. I am alien to a purely mathematical description of price with no tangible (visually observable patterns) justifications. On the contrary, I apply ZZ and see efficiencies from it (pedicators on ZZ are always at the top of the list in all MO packages). I think combining the two approaches could improve the results.
Selecting models through significance is nonsense - I've shown before that removing different significant predictors on the same model can improve learning results and form new, more productive and stable relationships in tree leaves. All this "importance" is the principle of greed in tree construction, which is not a priori correct, so we need separate meaningful predictor evaluation methods - I don't have them yet.
A year and a half ago Alyosha gave me advice not to use ZZ for inputs or outputs, as they peep into both past and future. Alyosha didn't explain the details - here's my interpretation of how things work:
When applying ZZ as a target, it is the past on which this ZZ began to be built. It is built on previous bars (from the zero point), which are fed to the input. In other words, knowing those bars and ZZ (partially based on them) you will get a peek into the future at the input.
When I started using price increments as an output, everything became less rosy than with ZZ.
If this is the case, I don't know what to do with it. It's unreal for a beginner in MO to read 1200 pages. And if you read it, you will miss it, if there are no explanations.
ah... zigzag... That makes sense then... why there are the same errors... I missed it.
Who even came up with this zigzag crap, I wonder... where the roots come from
Although it's probably the first thing that can come to mind - what to teach the model ... and here you can see the vertices and troughs, so the zigzag - the perfect solution :)KsanKsanych and his grandson Kesha came up with the idea. Who else?!
:))) someone even wrote about it before, but I don't remember those times.