Machine learning in trading: theory, models, practice and algo-trading - page 697
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
Everything I try to predict usually has an error of at least 30%, and here is such an extraordinary thing, but useless
Well, okay, maybe someone will have some thoughts - after all, future values are very important in machine learning.
f_ma=(sum(Cn(1:11))+ftr_cn)/12 => ftr_cn=12*f_ma-sum(Cn(1:11))
f_ma is the future SMA(12) value you predicted, ftr_cn is the future bar price value, sum(Cn(1:11)) - is the sum of values of prices of the previous 11 bars.
f_ma=(sum(Cn(1:11))+ftr_cn)/12 => ftr_cn=12*f_ma-sum(Cn(1:11))
f_ma is the future SMA(12) value you predicted, ftr_cn is the future bar price value, sum(Cn(1:11)) - is the sum of values of prices of the previous 11 bars.
This variant does not pass - look at the size of the error, moreover there are sections where the fact is completely the same as the prediction by 1-step-ahead. Calculate the error in your variant.
This variant does not pass - look at the size of the error, moreover, there are areas where the fact completely coincides with the prediction by 1-step-ahead. Count the error in your variant.
Doesn't your graph have a minus-16 degree, or am I confusing something? And here the error only increases by an order of magnitude.
You have a minus-16 on your chart , or am I confusing something? And here the error only increases by an order of magnitude.
That's just it. This is EURUSD.
I just happened to learn how to predict the SMA(12). Here is a graph of the divergence between the moving average and its one-step-ahead prediction.
If you use such a predictor that looks one step ahead, then... It's not the price, though...
Or is there any other way to use it?
Yes, if such indicators were on the price, that would be awesome.
But wait, the MA is a reversible transformation, well, at least for a period of 100 out of 1000 bars, 900 for sure can be reconstructed. Reconstruct the predicted price from the MA and check the error already at the price.
ZZZY MA(11)=(MA(12)*12-Close[i-11])/11; by this example go to MA(1), this is the original series.
If there is a prediction for one step of the SMA, then there is a prediction for one step of the price. The SMA formula does not change, it only changes a number of price values 11 of which are the same. )
Yes, yes already corrected the post.
When I read it I wrote it down and there are a lot of people like me here.)
Yes, if such indicators were at the price, it would be great.
But wait, MA is a reversible transformation, well, at least with a period of 100 out of 1000 bars 900 exactly can be recovered. If you want to check the error in the price, you have to do it again.
The price has nothing to do with it. There is a mistake between the fact of the MA and the prediction of the same MA...
Fa you're a pain in the ass. Pull the forecast price out of the forecast))))))))
Exactly, and count the error in the price. The grail is sitting idle.)
That's just it. This is EURUSD.
Something I don't understand - you have SMA(12) prediction with accuracy of 10 minus the 16th, it means that you can predict the future bar with accuracy of 10 minus the 15th. The price in EURUSD is at most 5 digits after the decimal point, which means you can absolutely predict the value of the next bar, unless of course you are mistaken anywhere.
I don't get it - you have an SMA(12) of 10 minus 16, that means you can predict the next bar with an accuracy of 10 minus 15. The price in EURUSD is at most 5 digits after the decimal point, that is, you can absolutely predict the value of the next bar, as long as you are not mistaken anywhere.
What "of" the value of the next bar? What are you all laughing about?