Machine learning in trading: theory, models, practice and algo-trading - page 2280
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
If on eurusd the price is 1.2 and ma 1.19
then on usdeur should be 0.83333 and 0.84
this way you can check the correctness of the forecast
1) Why is it worse to check the correctness of the forecast in the conventional way?
2) Make two/hundred benefit neuronics on eurusd and usdeur and make p. 1)
interesting approach to solving the problem of market forecasting through a knowledge base
When the rules will be under a million, it should work.
When the rules are under a million, it should work.
Why a million and not 100 or 100 million?
why should it make money and not unprofitable?
why so much bullshit in just 7 words? )
interesting approach to solve the problem of market forecasting through the knowledge base
The third test (false patterns) is especially useful because of the sufficient proximity of prices to the SB.
The third test (false patterns) is especially useful because of the sufficient proximity of prices to SB.
Uh-huh
how do you like this neural network prediction quality? of the last 20 predictions guessed 18
(no need to banish, there are no paid services, ads and all sorts of links to third-party resources)
how do you like this neural network prediction quality? of the last 20 predictions guessed 18
(no need to banish, there are no paid services, ads and all sorts of links to third-party resources)
It's hard to think of a more uncomfortable way to evaluate the bot, I don't know what to do, what to look at.
Make some kind of weekly reports, because in this chaos of mixed messages nothing can be understood.
Was the retraining going on all the time?
How do I solve the problem? I need to predict the waveform for half a period ahead.
There are 3 sources of information:
1) analyzing the mach itself and extrapolating it
2. If we shift the timeframe by half a period, then its prediction will be reduced to price approximation.
3. There are several currency pairs connected by a correlation. The simplest case is eurusd and usdeur, the prediction on one must match the prediction on the other. The more pairs are involved, the more accurate predictions should be obtained.
We should somehow combine these three objectives into one equation.
It seems that any approach of predicting price ahead solely on recent price data (and MA pairs among them) would result in "buy when all goes up". )
You may make your forecast using neurons or a fortune-teller.
I mean if you don't select/find predictors for a model of real importance - success is either by chance or fitting (while adaptive selection of a MA point period is a completely inborn idea)
And predictors may be in absolutely different planes and not directly related to the price data. What do you think?
It's hard to think of a more inconvenient way to evaluate the bot, I still do not understand what to do, what to look at...
Make some kind of weekly reports or something, because in this chaos of heterogeneous messages do not understand anything