Machine learning in trading: theory, models, practice and algo-trading - page 1125
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
Is there anything wrong with your brain? What the fuck is this "forward" ??????. You've adjusted to the noise (10 days of minutes) and run this random on the demo. I don't need to see "what will happen", I know it beforehand.
For two-scholars: forward is at least 200 deals (just as a hint for elaboration), while the norm is 1000-3000 deals if SR>3, if SR<3 then the need for even more deals, on data which in no way could be seen when training, have trained on the minutes on a year, tested for 2-3 months, or better trained for 3-5 years, tested on the year, and of course the test take it from the future)))) By God, kindergarten, gypsy bears and near-marketers, fuck them...
Dude, come on, don't give a shit about the dude. He trained for ten days and by the rules of the genre, the model can't run for long. The question is another. Will he be able to repeat this result with the next optimization:::
You're out of your fucking mind.) You're our fucking idol, teacher, mentor... ...you're our fucking idol, teacher, mentor...!)
OOO I see the heavy artillery is already here, and I'm drunk.... ♪ Hold on to me, seven of you ♪)
Is there anything wrong with your brain? What the fuck is this "forward" ??????. You've adjusted to the noise (10 days of minutiae) and run this random on the demo. I don't need to see "what will happen", I know it beforehand.
For two-scholars: forward is at least 200 deals (just as a hint for the study), and the norm is 1000-3000 deals if SR>3, if SR>2 then the need for even more deals, the data which in no way could be seen when training, trained on the minutes on a year, tested for 2-3 months, or better trained for 3-5 years, tested on the year, and of course the test take it from the future)))) By God, kindergarten, gypsy bears and near-marketers, fuck them...
for brains need information, show at least something to support what is described, and so, judging from the previous false messages, it is an idle chatter...
On one random. The second, please do not ask to add, you will drool.
Some time you will work on any sheep, the good))).
Listen, do you really think the OOS is cool in this example???? I don't think so..... Tell me, at these points, what are you going to do?
Misha, I'm not going to do anything there))) If you do not know what to do with it, do not try to freeze it out (for example, if you do not know what to do with it).
So that's the whole point that 90% of users of this particular optimization are likely to turn off the robot. The drawdowns are too high. That's when you wonder. What is it? Just a slump in balance or rollback. That's just... ...information for beginners. The problems in real trading are of a different kind....
Listen brothers I'm in a little shock from checking my metric and to solve it you need only help with the answer
In java, I have this expression
double[] pattern = min.get(i);, where min is an array-list and in this example the entire string is passed to the pattern.
But I need to pass this string to the pattern, but only from 1 to 5 elements, and from the fifth to the tenth to another variable. Any ideas????
Misha I will not do anything there)))) This is an example of how to make a rattle on 1 random, no more))))
To avoid asking these questions, you need to make a normal model with a normal
The model should be made with a normal teacher and watch the AOS on the marked data, and not guess by eye on the equi, just repeated for the hundredth time yesterday.
just repeated it for the hundredth time. 0.7 is good on the candle color, yours is rubbish...
You described the classic use of the mo, I gave you the mo, you had everything, but...
you miscalculated again and went to work on Monday)))
Yeah, well... You know, I'm glad I work there. It's not gazprom,you know? It's google, just in its own field.... :-)
Spit it out! Where and what you work. Let's have some fun...
And help with java???? That's just it...... Just think of it as one line. I don't know how to refer to a line item in a list. I've already tried everything by gut feeling. And you say put it out there... and I'm not talking about the work, I'm talking about the whole thing....
Is there anything wrong with your brain? What the fuck is this "forward" ??????. You've adjusted to the noise (10 days of minutiae) and run this random on the demo. I don't need to see "what will happen", I know it beforehand.
For two-scholars: forward is at least 200 deals (just as a hint for elaboration), while the normal 1000-3000 deals if SR>3, if SR<3 then the need for even more deals, on data which in no way could be seen when training, have trained on the minutes on a year, tested for 2-3 months, or better trained for 3-5 years, tested on the year, and of course the test take it from the future)))) By God, kindergarten, gypsy bears and near-marketers, fuck them...
Yes, you are a maximalist and sado-masochist. Well, 200 deals - that's still there and then, but 1000-3000 - it's already a mockery). And training for 3 years for a number of instruments is nonsense. 2 years - this is the lifetime of a system due to changes in the market. Even my systems - from 4 to 10 deals per day, for 2 years will make about 2000 deals before their death.
Learning from one year old data is a waste of time. The previous 6 months is all we have for training and tests. That's ~600 deals total. That said, there is no argument that 3 years and 3,000 trades would be good. Uh-huh, okay, but where do we get them?
Another question here is what to teach on, and what to test on. It seems more logical to teach on the last 3 months. Then we have to test on the previous ones. And if we teach the previous months, we will have to teach something which we haven't had for a long time.
So the theory is good, if we take the market as a rigid and preserved system. Imho, the market changes significantly, even within a few months.