Machine learning in trading: theory, models, practice and algo-trading - page 2459
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
It worked for me for 2.5 days.
So in the end - taught you how to park?
Greetings Brothers!!!
I remember I've already said it more than once, but I'll say it again. Yes the method of training and the architecture of the NS is important, but much more important is the data you use. In many respects with well prepared data will work qualitatively a wide range of network architectures. Naturally each type of NS requires its own specific preprocessing, but if the input data, the information you take to enter the network makes sense for the target then the result will be visible immediately. The point of digging different methods of constructing a system, if you exit only on the unique configuration will not work anyway.
Well it's me so, suddenly young people read :-)
to distinguish between a Trend and a Flat to include the appropriate TS
the option of searching for fat tails in the Asymmetry distribution in the time-series analysis, is still the way to a prediction model based on only one factor - the time factor... And multifactor models (even those that are incorporated in the pricing of futures options) still can only be mathematically expressed in general terms and, of course, the forecast(which is determined by statistical regression model data) and its confidence level(as estimated by the adequacy of the model - for example, Fisher's F-criterion, etc.), as well as the limits of the model accuracy.), and the limits in which this prediction is most probable, - lead to the necessity to take into account model error and prediction error as well...
- Then we'll put all this stuff into neurons and work out algorithmic dependencies of further events on factor values and their errors... -- that's, I suppose, if you do it right... But to build such a BC (computing system), to get only a fraction of probability in the output anyway, I'm still too busy...
thanks for feedback from those who tried these NS, but, really, without correct input data you can get only "what God will give" (which you can't get without supply and demand anyway) at the output
... because it seems to me, anyway, that the more factors, the greater the cumulative error... Although, of course, you should only allocate the driving factors in your analysis... but that's at least Interest Rate & Money Supply (data on which we, as retail taders, appear at the last, if at all)... we should also add Net Exports and Net Foreign Investments in the analysis of D & S, of which no one lets us know either
P.S.
so we just have to intuitively rely on the forces of interest rate auto-regulation, and monetary and fiscal impulses to regulate the exchange rate.... and wait for real Driving Events, not gossip in the news... a robot will not distinguish the latter two at all (especially if it does not get a large enough sample of historical data for statistically reliable analysis)... As for news and gossip and reactions to it, it's probably worth running through the whole history of mankind to estimate where and when an economic crisis or economic recovery will occur -- and there will still be few cases for statistical validity
That is why long-term interest rate(s) are the analyst's best friend,
short-term interest rates are a trader's best friend... imho (if you study its behavior)... because if money is taken from the open market (D & S), it's at interest
Here's about machine learning
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
because it will boomerang back on your children and grandchildren - think before you fool the people
because it's like a boomerang
It's more like manure - not all manure is good for fertilizing the soil you're offered... - you have to know its sources!... imho
(in moderate doses and good quality can be beneficial)... Of course, otherwise it can boomerang...
Here's about machine learning
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
because it will boomerang back on your children and grandchildren - think before you fool the people
Here's about machine learning
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
It will boomerang back on your children and grandchildren - think before you make a fool of people.
What does this have to do with the Ministry of Defense?
With such logic, you can go as far as the midwife who delivered the baby )))))
What does the MO have to do with it?
With this kind of logic, you can get as far as the midwife who delivered the baby )))))
It's just a pretty wrapper for beginners, and inside, 0
inside will be what you feed the NS's input -
(just like with human brain neurons)
inside will be what you feed the NS's input -
(just like with human brain neurons)