Machine learning in trading: theory, models, practice and algo-trading - page 917

 
Maxim Dmitrievsky:
This is pennies, it's not even enough for a normal car. Norm investments from 1 mio are usually considered. You just have to move in the right circles.

This is an argument.

Where to look for these circles now....?

 
Maxim Dmitrievsky:

I know, everyone is looking for something different :) I already have some connections

If you're cool, the environment will change

Cool (have achieved results in anything) or a skilled showman (know how to attract attention using different technologies)? This topic is really important for those looking for investors.

 
Maxim Dmitrievsky:

I dunno, everyone has their own understanding, many people immediately have twisted thoughts about being fucked everywhere... :) which is partly true... if you don't have a brain of your own

there is no point in discussing it further

Well, what does their understanding have to do with it, we're talking about a scientific approach with the purpose of extracting funds!

 
Maxim Dmitrievsky:


If other people think you're cool about something, then you're cool. Prove it to investors and they'll follow you like the Jewish people.

How you do it and what the consequences will be - your problem :)

Yeah, so it's important to understand what others (potential investors) think is cool, and our job is to give them what they want!?

It's all interesting, without these skills even with a good TS it won't work...

 
Maxim Dmitrievsky:
Well at least their expectations should be supported at least in the drawn future :) otherwise no one can be sure

We are up to this task :)

I had one investor who demanded to increase the risks, and when he lost everything in one day, I was the last one to blame. I have been telling him 100 times about the dependence of profits on the probability of sinking: I used TS with averaging.

 
Maxim Dmitrievsky:

Many investors are even less rational than traders.

I think there is a grain of truth in this, so the investor should be given as little opportunity to influence the TS as possible, I did not understand it or I was greedy myself. I will only say that at that time I also lost more than the investor, but I understood the risk, and he did not (I could have saved 60% of the money, but he said he wanted to fully copy my decisions on the current situation - I recommended to fix the loss). I am not sure how to make a profit if I do not know what to do with it.

 
Maxim Dmitrievsky:

The main thing is to stipulate your share in advance and take it away in time. From practice - people with bloodshot eyes and fantasies are not able to act rationally, it is useless to explain:)

There are meticulous people who think they are smarter than you and begin to give advice - better not to have anything to do with them. They better not have anything to do with you.

That's right.

 
forexman77:

It's time to cool down the dreamers) This is just the beginning. And it's the best test yet.

The forward itself.


A perfect example of artificial rules - i.e. describing the past due to statistical coincidences but not based on a pattern.

Alas, statistics are against us, as you can describe the market using different methods, each method will find its statistical patterns, but they do not coincide with real patterns, alas.

 
Maxim Dmitrievsky:

Loophole - add losing strategies to the system (so that they learn to lose)

Then on the OOS will work steadily, such as here (for the best option, just as an example)

This is not magic, game theory :)


I.e. target at a loss to put? And then combine different strategies in one file for training?

 

I experimented tonight. And I succeeded, the entire curve without losing trades at all. I hardly fell asleep, excited).

In the afternoon I checked, and it turned out that the program only saw training data)