Machine learning in trading: theory, models, practice and algo-trading - page 799

 
Mihail Marchukajtes:
.....

You have a system that's proven itself in forensics, and then you get an offer to use it in medicine. What? You're gonna say no?

That's a bit of an overreaction.

You have to earn money, take it out, earn it again, take it out again.

then...

 
Mihail Marchukajtes:

What kind of proof do you need? I don't understand... Isn't real time trading the most reliable proof. Or do you still want to see everything on history in the tester....

You're constantly losing money in real time, and the last piece doesn't break out of the overall statistics.

this is the 1000000th time! it's a very simple thing, you've had better periods in the past

statistically nothing has changed. And your conclusions about robustness are premature and false

Usually, a person needs to be told eight times for him to understand

 
Maxim Dmitrievsky:

you have constant losses on the real, and the last piece of the dynamics does not stand out from the overall statistics

It's a very simple thing, you've had better periods in the past.

statistically nothing has changed. And your conclusions about robustness are premature and false

Normally, a person needs to be told eight times before they understand.

That's it...bowing my head and serving my sentence in the corner. Not another word. Really what is it me.... cardboard idiot....

 
Maxim Dmitrievsky:

the only thing that someone suggested was Alexander, but I think it is necessary to look not there but somewhere in the vicinity

The probabilistic approach was also mentioned here by Yuri Asaulenko, but neither he nor anyone else could show or reveal the potential of the topic

I am now writing about non-parametric methods, where only price and its derivatives act as a predictor

Well, why not? I told you everything about the methodology, I showed you how to walk and how to pass. I even demonstrated some real-time trades. I showed that the approach really works. And how to build a system, and was not planned to disclose - that's for yourself, if the approach is of interest. The forum is for the exchange of opinions, not systems.

Plus, showed a couple of test grails.)

One to show that with 30-40% of successful trades you can be in good profits. And just then, for some reason everyone wants more than 50% successful trades, it is unclear why. Although, of course - the more - the better, but it is not necessary for profit.

The second test grail - is a system a la A_K2, but made much easier. It does not have to be as complicated as A_K2 - the approach is not new. Conceptually, it's just a modification of the game on Bollinger.

In terms of implementation, these Grails are not worth it, there is no need. Although if implemented, the systems will work, but it's just experiments and nothing more.

 
Yuriy Asaulenko:

Well, why not? I told you everything about the methodology, I showed you how to walk and how to sell. I even demonstrated some real-time deals. I showed that the approach really works. And how to build a system, and was not planned to disclose - that's for yourself, if the approach is of interest. The forum is for the exchange of opinions, not systems.

Plus, showed a couple of test grails.)

One to show that with 30-40% of successful trades you can be in good profits. And just then, for some reason everyone wants more than 50% successful trades, it is unclear why. Although, of course - the more - the better, but it is not necessary for profit.

The second test grail - is a system a la A_K2, but made much easier. It does not have to be as complicated as A_K2 - the approach is not new. Conceptually, it's just a modification of the game on Bollinger.

In terms of implementation, these Grails are not worth it, there is no need. Although, if implemented, the systems will work, but it's only experiments, and no more.

If these approaches are not new, they should have had some names for a long time.

i don't mean the probabilistic approach when the probability of a profit trade is less than 0.5, but the average profit for them is higher than the average loss for the others.

I mean the distribution of quotes, for example how to predict the expected density of the distribution or something like that. The bollinger works in flat.


 
Maxim Dmitrievsky:

If the approaches are not new, they should have had names for a long time, not on the fingers but by names.

i don't mean the probabilistic approach when the probability of profit trade is less than 0.5, but the average profit of these trades is higher than the average loss of other trades

I mean the distribution of quotes themselves, e.g. how to predict the expected density of distribution or something like that.


I mean the A_K2 approach. It's not new conceptually. The implementation, yes, is different, but the concept hasn't changed much - a return to the mean.

Sorry, I'm not in the business of predicting the density of distributions.

 
Maxim Dmitrievsky:

If the approaches are not new, they should have had some names for a long time, not on the fingers but by names.

I don't mean the probabilistic approach when the probability of profit trade is less than 0.5, but the average profit for them is higher than the average loss for the others.

I mean the distribution of quotes, for example how to predict the expected density of the distribution or something like that. Bollinger in flat only works


 
SanSanych Fomenko:

Thanks, I'll read it.

Here, by the way, is the appendix to the vid from Kuznetsov above. 2018 something interesting, but I don't understand it yet. There are examples of predicting bitcoins, forex, etc. And a comparison of his method with arima.

https://arxiv.org/pdf/1803.05814.pdf

 
Maxim Dmitrievsky:

Bollinger in the flat only works

This is an erroneous opinion. The notion of a trend-flat is very relative. What is a flat for one person may well be a trend for another.) And vice versa.)

Of course, Bolinger-like strategies have their limitations. This is just like any other strategies.

 
Yuriy Asaulenko:

This is an erroneous opinion. The notion of trend-flat is very relative. What is a flat for one person may well be a trend for another.) And vice versa.)

Of course, Bolinger-like strategies have their limitations. Just like any other.

I think that Bolinger strategies do not work at all, because they are a consequence of price. That is, they change after the price has changed not in time of course, but causally. And the percentage of 30-40 deals is a bullshit. I'm counting on the ratio of more than 70% of profitable ones. It was so until I screwed up this ratio and it is about the same average loss and profit. This is what I understand a strategy that you can trust. And you can see it all on the balance curve when you have enough experience. Just by its appearance alone you can tell a lot about the TS, because there are certain conditions to which it must adhere. And if you were more experienced you would have seen it in those screenshots, which I threw off two deals on purpose, coded so to speak ... and thus confused you, allegedly TC shit, but in fact it's not that... And in general, who cares, if some and 30-40% of transactions enough. I do not understand how it is possible to work when the TS is in a terrible drawdown. Look at my screen separate sections first and second. Only up, only forward.....


I will add that this is the period of change of futures and time zone....