Machine learning in trading: theory, models, practice and algo-trading - page 1682

 
Aleksey Nikolayev:

The problem I see is that this approach is applied to a two-player game. It is not very clear whether it can be generalized to games with an arbitrary number of players. In the market, there is no clear and immediate sense of confrontation between traders (unlike chess and other one-on-one games) because there are so many of them. And there is a significant difference even between games with two and three players - enough to play "backgammon" with different number of players), and theory gets complicated there...

Everything is simple here - the market and I

Aleksey Nikolayev:

With the market everything is somewhat simplified because everyone has only two strategies (buy and sell) and all traders are the same - this is called a symmetric binary game. In a symmetric game there is always a symmetric (possibly mixed) equilibrium

I disagree, the main problem here is that there are no formalized rules of the game,

buy and sell is we'll discuss the even and odd of roulette.

The market rules are more complicated, you need to know the type of action (buy - sell) and WHEN to commit profit/loss fixing, for example if the market is sideways you can buy and sell and get profit from different actions at the same time ... but only on history

i haven't watched the video, although i did it last year.


But how to evaluate the strategy qualitatively? - the main problem here is that the proposed statistical indicators (MO, continuous loss/loss,z-score) are statistical indicators that depend on the moment of observation and the total number of observations

how can we tell if the strategy indicators have changed? - i can't even think of something to analyze in the Strategy Tester (((

 
Igor Makanu:

everything is simple here - the market and i

how do i know that the indicators of the strategy have changed? - Even in the tester I can't think of anything to analyze (((

It's not a game after all, but a probabilistic guess, after all. Symmetrical binary behavior is not a game, traders are simply called players and we also play roulette. There's not much you can take from game theory in tick data analysis. From weather, Brownian motion, calculations of turbulence onset and laminar stability limits, it is probably better to look at algorithms, here it is more like the market
Or at least something similar.
If the TS works, then stops, then starts, and has no idea where the transition from work to pest may make sense to determine the quality of work not in two criteria, but to gradate and optimize the gradation when to go out.
 
Valeriy Yastremskiy:

If the TC works, then stops, then starts, and there is no understanding of where the transition from work to pest, it may make sense to define the quality of work not in two criteria, and make a gradation and opt for when to enter the exit.
I did not quite understand what I wrote. The idea is to divide into smaller sections, where the results of TC are stable, and to gradate them. And the stability can be determined by the similarity of the results on even smaller segments, the number of orders, profitability and lifetime.
 
Igor Makanu:

Everything is simple here - the market and me.

Since the influence of one ordinary trader on the market is negligible, we get what in game theory is called "playing with nature". This is all the same matstat, machine learning and non-stationarity problems.

Igor Makanu:

I disagree, the main problem here is that there are no formalized rules of the game,

buy and sell is we'll be discussing the even or odds of roulette.

An approximate formal model is not difficult to build. It is possible to approximate a discrete time - no one will change positions too often. We get a repeating game (that's a game-theory term) of the same rounds, with a minority winning each round-it looks like even-odd, but it's not. I then make the assumption (which I am not prepared to prove mathematically) that the resulting repeating game also has a symmetric equilibrium, which is constructed as a sequence of equilibria for the round games. That is, all players in each round flip coins and win only if they are outnumbered. In this equilibrium, the price behaves like a SB.

This equilibrium will be unstable, because staying in it is not profitable, but it is also dangerous to move far away from it. The situation is remotely reminiscent of the Lorentz attractor. As I have already written, there is no practical use in it. Except that non-stationarity, which makes the existence of the grail impossible, becomes more understandable.

 
Aleksey Nikolayev:

An approximate formal model is easy to construct. It is possible to approximate a discrete time-no one will change positions too often. We get a repeating game (that's a game-theory term) of the same rounds, with a minority winning each round-it seems like even or odd, but it's not. I then make the assumption (which I am not prepared to prove mathematically) that the resulting repeating game also has a symmetric equilibrium, which is constructed as a sequence of equilibria for the round games. That is, all players in each round flip coins and win only if they are outnumbered.

OK, consider this to be the formalized rules of our game

To the same question: how to consider this game from theposition of the Sprague-Grandy function , everywhere follows the statement that one can consider any equal-opportunity game as an elementary game of "Nim" - is this possible?


ZS:

Aleksey Nikolayev:

We get a situation remotely resembling the Lorentz attractor. As I have already written, this is of no practical use. Except that the non-stationarity, which makes the existence of the grail impossible, becomes more understandable.

I like to consider the market as a Cantor set, most of the market processes correspond to the division into "triples", as per the participants - there was a volatility spike and the 1st ones left with a profit, the 2nd ones left the market with losses, the 3rd ones just entered the market hoping the movement will continue or the Fibo levels are a multiple of 3, .... This is a question of who sees what in the starry sky - it's senseless to discuss or try to apply the existing models or abstractions.

 
Igor Makanu:

OK, consider that these will be the formalized rules of our game

to the same question: how to consider this game from theposition of Sprague-Grandy function, everywhere follows the statement that it is possible to consider any equal-right game as an elementary game "Nim" - is it possible?

From wikipedia: "The Sprague-Grandy function is defined for games with two players. You need some generalization of this function to games with more than two players. I couldn't find one. Perhaps we should look in the direction of potential games.

"Me and the market" is clearly an unequal game. A possible two-player option is "all traders vs. state," but that too is clearly an unequal game. Probably you could also consider games for several players, each of which is some group of traders.

Igor Makanu:

It is a question of who will see what in the starry sky, it is senseless to discuss here or try to apply existing models or abstractions.

I agree)

 
Maxim Dmitrievsky:

NEW GETTING THERE

Ahem, ahem. Funny. Well done. You did it...
 
Mihail Marchukajtes:
Ahem, ahem. Funny. Well done. You got out of it...

Repetition is the mother of torment

You better write a video. Why did you buy a camera?
 
Valeriy Yastremskiy:
INTELLECT is the thinking (mental) capacity of a human being; it may be identified with the intellect,reason and intuition.Concise Dictionary of Philosophical Terms

Speech in general, if the definitions, terms, concepts have not yet settled and are understood different ly, it is better to explain in more detail what is meant ... There will be less swearing.

In the '90s, I wrote a little program in Basic, a hint, what to do, a yellow icon appeared, press F5, appeared green, find and open the black closet, like that.... The manual called this program a program with elements of AI))))



I gave you a clear definition, and you ...

what is mental capacity?

what is a mental capacity?

what is intuition?

so it is understood differently.... Again, I gave a clear definition, but you do not need it, you just want to "talk trash".

And in general if you do not agree with something, then you should say - I do not agree with this and that, because......... )


Otherwise this conversation is no different from farting.

 
mytarmailS:

I gave you a clear definition, and you...

what is mental capacity?

what is mental capacity?

what is intuition?

that's why it is understood differently.... Again, I gave a clear definition, but you do not need it, you just want to "talk trash".

And in general, if you do not agree with something, then you should say - I do not agree with this and that, because......... )


Otherwise this conversation is no different than a fart.

In general, I agree with the definition), logical and capacious definition, but few people know it and in school and university do not teach it in that context, so it is difficult to call it a generally accepted definition. And for many disciplines, it is even controversial. For MoD and AI it is the best.