Machine learning in trading: theory, models, practice and algo-trading - page 996

 
Alexander_K2:

And yet.

Now it is fashionable, in view of the small profits from the trade itself, to trade by signals.

Why none of the old-timers (Koldun, Toxic, etc., including, of course, you) do this? I've never even seen a report from MT on a real trade.

Is there no desire to just show off their skills and get a well-deserved round of applause?

Alexander, you are beginning to see the light. They do not have any reports, I will never tire of repeating this to you. All the more they do not have them from the buffoon of peas Asaulenko. In other words, they're not even one iota closer to your understanding, and you're looking for answers from them??? Wtf?!? Aleshenka has never traded at all, judging by his inarticulate nonsense. It's not even funny, his uncles will order him to babble... what a clown. These kids just don't... send them away and that's it.
 
Alyosha:

I'm not going to say anything about the first couple of lines, I can speak for myself in the third point, I don't trade on DC Forex and MT respectively, I don't trade on any public platforms at all, everything is through api, the tester has been my own for a long time too, I cannot publish my own equities as I trade other people's money (DU) with tough guys who don't like my public comments, even if they are false, while my desire to propagate myself among newbies disappeared about three years ago when I got a hint of something profitable.

Asking such a question, says about you or a novice, who does not distinguish between DC-forex and circum-market subculture, motivating "to get rich by investing $ 1000 in a grail" with "signals" and "affiliate" from real professionals like Rentech and Citateli, or what is more likely you want "to trick" me for information or even to get exposed to penalties or other damages. I refuse, with all due respect.

I was asking with absolutely no subterfuge or any intent.

I'm really a beginner, so to speak, although I've been dealing with Forex for about a year now. As it turned out, it's quite a difficult task and (I brag a little) even my, as I thought before, pretty good education and life experience is not enough to solve it.

I consider it an honor to talk to old-timers with real results.

 
Maxim Dmitrievsky:
Alexander, you are beginning to see the light. They don't have any reports, I won't get tired of telling you this. All the more there are no reports from Asaulenko, the buffoon of peas. In other words, they are not even one iota closer to your understanding, and you are looking for answers from them??? Wtf?!? Aleshenka has never traded at all, judging by his inarticulate nonsense. It's not even funny, his uncles will order him to babble... what a clown. These kids just don't... send them away and that's it.

Maybe so. :)))) I do not know - I just want to believe that there are real craftsmen.

 
Alexander_K2:

Maybe so. :)))) I do not know - I just want to believe that there are real craftsmen.

Normal guys do not even so communicate, not to mention the men. It was evident from the very beginning, hu from hu. If you have something behind you, the behavior is different. Or maybe people are just so sh... like Asaulenko that they allow themselves such behavior, but it's unlikely, more likely just weakness. Psychology.
 

Anyway, here's the deal. I started to use boosting, because in addition to accuracy and high generalizability in this algorithm model building is more unambiguous. The method is also easier to set up because of a small number of specific external parameters. The only disadvantage is the loading of RAM when calculating, and accordingly the model size increases in tens and hundreds of megabytes depending on the number of iterations. As the result of comparison with random forest and shallow neural network methods I came to conclusion that boosting is more preferable for classification tasks.

I tested a lot of predictors. Mainly they are sequential time series formed from the most various indicators and their combinations. Testing was performed in multicurrency mode (27 currencies) by the program method taking into account the real spread (2 points). Timeframe - an hour. In the output - a binary class, calculated using a zigzag signal with a step depth of 100 points. Almost all results are negative. If you exclude the spread, the plus can be substantial. As an option, you can try to take a higher timeframe.

I have in mind how to further develop the study:

1. to try another type of output zigzag or other parameters.

2. to use cyclic component signals selected using the Fourier method or wavelet filters.

3. use real values of indicators on the output (regression). For example, difference of close and open prices of candlesticks, price change for several bars ahead.

4. use inconsistent data as predictors, for example, key points or levels

5. filter the initial sampling on different indicators (Volumе or ATR indicators), i.e. to teach working only on certain parts of the market.

I will be glad to hear your opinions and advice.

 
Maxim Dmitrievsky:
Alexander, you are beginning to see the light. They do not have any reports, I will never tire of repeating this to you. Moreover, Asaulenko, the buffoon of peas, does not have them. In other words, they are not even one iota closer to your understanding, and you are looking for answers from them??? Wtf?!? Aleshenka has never traded at all, judging by his inarticulate nonsense. It's not even funny, his uncles will order him to babble... what a clown. These kids just don't... ...send them away, that's all.
Yeah. That's a daily occurrence.
Psychiatry knows this condition. Have you ever tried to contact them?
 
Alexander_K2:

In order not to be hopelessly sad, I once again attach Kolmogorov's work on BP forecasting.

From it, it is simply obvious that we should work with the first and second price differences. And if the expectation of returnees is always strictly = 0, then with the second differences we need to work - to do such a tricky (according to Aleshenka - exponential) sampling, so that the ACF took some tricky form (I did not know about it).

And voila - the Grail was taken from crying in tears Aleshenka and distributed to the suffering people.

I guess you're wrong. Non-stationarity of prices (as well as their increments and increments of increments, etc.) is not particularly disputed by anyone. So the article (about stationary random sequences) doesn't help us much.

 
Alyosha:

It's when he starts praising you that's cause for worry:)

God forbid. For all intents and purposes).
 
Alyosha:

There is no need to try to change Maxim, he condemns many, from his vague emotional motives, he has such a feminine petty-evil nature, just invert his throw-ins, what he condemns means he approves, what he approves means he condemns, but when he starts praising you, it is a reason to worry:)

I'm not judging anyone, I'm talking to the shitheads as it should be, it would be strange if there was a different reaction to you jokers. Get into a pile and start defending yourselves all together now, it's the last thing you can do. Don't you wonder why others respond normally?
 
Ilya Antipin:


1. try a different type of zigzag or other parameters on the output.

I would be glad to hear your opinions and advice.

You should start with the target, the zigzag. A very sneaky thing with a deceptive visualization. For example: a trend at the beginning of the zigzag, and at the end, the same trend? Or does the trend start before the zigzag turns? Or maybe the trend is a part before the reversal and a part after the reversal such that to the reversal will give some target profit?


There are a lot of questions with this zigzag, and then there is a question: are there any predictors for the desired target formed from the zigzag, because the questions put into question the possibility of using it as such?