Machine learning in trading: theory, models, practice and algo-trading - page 458
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
Why has he written in java and not in mql? There he uses one third-party library, the sources of which are only available in java at a third-party site, maybe because of this?
What's a lib? What's the name of it? And what does it matter????
What's the name of the liba?
I'll find it later from another computer, I can't remember now.
there, if you look at the list of libs with the reflector and google some, there will be links to third-party sites
I'll find it later from another computer, I can't remember now
there if you look through the list of libs by reflector and google some, there will be links to third party site
So what's the big deal if you use a third-party library???? Or do you think it's some kind of trick?
Sensei, I had a lot of fun yesterday. One does not understand what we're talking about and cut in))))
The other fucking saw it six times, but asked for the seventh.) What are you going to do with this information?
You know it won't do you any good, but people with a normal level will compare it in 5 minutes.
a rattlesnake with a little bit adequate models and throw it away. Thank you for telling me)))
How to get it under the plinth was announced yesterday, but of course you do not understand...
You haven't said anything of substance, you've just trashed the thread without bothering to back up your words with any arguments at all. It's all "I laugh" and "I don't give a shit". That's your entire argument. Troll!
So what's the big deal if you use a third-party library???? Or do you think it's some kind of trick?
No, I just tried to understand it, but some things are still unclear.
So it's not possible to rewrite it on mql in the original and use the same cloud for quick calculations so far
That's why I'm writing my original NS)
No, I just tried to understand, but some things remain unclear, what from where and where.
To be honest, I myself opened the project in Eclipse yesterday, browsed through the classes, looked at the surface and realized that for me Java at this stage is a complete forest.... But let's try it, let's see what's up.... Maybe I'll get the hang of it....
No, I was just trying to figure it out, but some things remain unclear, what from where and where.
So it's impossible to rewrite it on mql in its original form and use the same cloud for quick calculations
That's why it should be rewritten in SI, there already can be run on multicore procic.....
Anyway, I don't know why... Maybe my data is good or what, but on average the polynomial works for about 70-80%, that is, out of 10 signals 2 errors maximum. Taking into account that it is trained correctly.....
The question arises, and where is the fit? How did you even determine that the Predictor is overtrained, etc.? Questions????
You haven't said anything of substance, you've only polluted the thread without bothering to back up your words with any arguments at all. It's all "I laugh" and "I don't give a shit" about anyone. That's your argument. Troll!
Here I totally agree with you.... Wizard is throwing poo around the corner. Maybe he won't hit us, but he will definitely have his own hands in shit...
That's why it should be rewritten in SI, there already can be run on multicore procic.....
Anyway, I don't know why... Maybe my data is good or what, but on average the polynomial works for about 70-80%, that is, out of 10 signals 2 errors maximum. Taking into account that it is trained correctly.....
The question arises, and where is the fit? How did you even determine that the Predictor is overtrained, etc.? Questions????
Have you tried it on real or forward, how stable is the system and when do you need to retrain? For example, informative predictors have been selected for this certain timeframe, and tomorrow the market has changed, patterns have changed and the polynomial has gone to hell.
By the way, RNN also works perfectly without radical market changes. But its disadvantage is that there is only one neuron, which is not capacious and cannot be trained properly on a long history, the classification becomes too rough and the profitability goes to zero, on short periods everything is fine
Have you tried it on real or forward, how stable the system is and when it needs to be retrained? Fitting can be exactly the construction of a polynomial that may show quite different results, even the opposite, in case of market changes... for example, informative predictors for this particular timeframe were selected, and tomorrow the market has changed, patterns have changed and the entire polynomial has gone to hell.
Well, in general I calculate the period of the feedback roughly as half of the training set. I.e. I trained it on 50 signals, it should work from 25 to 100 signals approximately. If the polynomial works for 100% of learning period, then I consider it a good result. The market is always changing..... it needs to be constantly re-trained....