From theory to practice - page 786

 
Alexander_K2:

It's not a reaction, it's a simple question:

How is it that you are able to determine that this trending movement is over?

After all, at that point all statistics, including Chebyshev's inequality, are torn into fur. Only Markov's inequality works there, up to 100*sigma.

That's the only thing I'm guided by.

 

I am absolutely convinced that the process in the market is very similar to the random Ornstein-Uhlenbeck process with a tendency to "return to the mean".

With one exception.

Here if you plot the histograms of everything in the market - increments, correlations, deviations from expectation, here literally everything has almost a Laplace distribution - double exponential + outliers in the super-heavy tails.

This suggests that we are facing some kind of symbiosis of an Ornstein-Uhlenbeck type SB process and a process with memory.

If there were no these outliers, my TS would have a 100% profit, and anyone who asserts that you can't make money on SB is weak-minded. On a coin, I don't know, but on an OM, it's easy.

And it's impossible to eliminate these outliers in the tails - this has also been proven in this thread. You have to live with them somehow... How? No answer...

 
Maxim Kuznetsov:

You may have said the thing that A_K screwed up - the apparent reaction is delayed and prolonged. That is, the tick that is coming now is not a direct and significant consequence of the previous one. The whole system has not yet had time to fully react to those events, only the nearest node (and/or its neighbours).

And also possibly with the larger bars.

A_K seems to have calculated for a fully quantized case - the event came, everyone reacted and here is the next result.

Unfortunately, the event came, but some participants reacted to one event, others to another, etc. Everyone has their own plans: some for minutes, some for months. And everyone has different money and goals. And events can be unrelated and differently directed, and everyone's agenda is different. At the end of the day, there is no such thing as a line up. For example, the price went up - it was an event for me, while for him - Lukoil signed a contract - also an event. And the third one has decided to return to the average - another event).

It turns out that ticks are nothing. We need the dynamics of the hospital average, which also does not guarantee anything.)

 

I respect your confidence.

 
Алексей Тарабанов:

I respect your confidence.

Well, I have both demo and real results.

If you exclude the most shameful trades against hellish trend movements, my purse would already stupidly not close from the strength of the elasticity of the cash.

Alas, how to bypass trends - haven't figured it out yet.

Out - let Asaulenko think and report on the proper form.

 
Alexander_K2:

Alas, I have not yet figured out how to get around the trends.

Out - let Asaulenko think and report on the proper form.

Already said: Trajectory measurements

Forum on trading, automated trading systems and testing trading strategies

From theory to practice

Yuriy Asaulenko, 2018.11.27 19:00

Shit. What kind of memory? Don't enter a trade until you are sure the price is really going in the right direction. Not relative to the average, but the actual price.

And exit the trade if the price suddenly turns around and goes in the wrong direction.

To make it clearer to physicists - it is necessary to look at the trajectory and its parameters in the real space.

That is all. There are no secrets here. Don't bother with the gral.

By the way, a few months ago I recommended you a book, where the entire methodology is described in detail. Look it up in the thread.
 
Alexander_K2:

Well, I have both demo and real results.

If you exclude the most shameful trades against hellish trend movements, my purse would already be stupidly closed by the strength of the elasticity of the cash.

Alas, how to bypass trends - haven't figured out yet.

Out - let Asaulenko think and report on the proper form.

Trends should not be bypassed, but go with them in the same direction. For the hundredth time... and you still don't get it.

 
Олег avtomat:

Trends don't go around them, they go in the same direction. For the hundredth time... and you still don't get it.

So here's the trading plan: don't open this thread before 1000 pages, trend)).

 
Unicornis:

In general, the trading plan is as follows: do not open this thread until page 1000, trend.)

It's not a trend here, it's a mess.

 
Yes, it's past page 786...