What's so special about the markets that you can't compare it with any other game?

To add comments, please log in or register
Mrluck07
425
Mrluck07  

Nowadays we have big companies with good AI systems, today there is an algorithm called Alpha Go zero, this AI can win basically any game, you just have to tell it the rules, and then he get better at playing, and in hours, can outmatch any other player. My question is: if forex/ stock market are games, then why Alpha Go, and other AI systems can't return good profit and beating the market?

Eleni Anna Branou
Moderator
6646
Eleni Anna Branou  
Mrluck07:

Nowadays we have big companies with good AI systems, today there is an algorithm called Alpha Go zero, this AI can win basically any game, you just have to tell it the rules, and then he get better at playing, and in hours, can outmatch any other player. My question is: if forex/ stock market are games, then why Alpha Go, and other AI systems can't return good profit and beating the market?

First its not a game, second all big institutional participants have already implemented computer aided trading for many years now and finally forex as all other markets its driven by news and unpredictable factors.

"The best way to predict the future is to create it" 

Abraham Lincoln

Amir Yacoby
1222
Amir Yacoby  
Who said AlphZero can't learn to trade? I'm sure it can, and very good. 

And in order to do so it doesn't have to predict the future, it can just find the holes in probabilities, where it stands the best chances. 

And also, if you love so much to talk in theory, then let's assume a future situation where all market participants will have a private alpha-zero to trade. Then, in that situation, alpha-zero will lose it's edge, and they will all lose money (in average) - at least the spreads+commision (in the best case they will lose only spread and commission - if all somehow manage to break even).

In current situation, given it's abilities, I assume it can have an edge over most humans and software currently in the market. But that's changeable. The traders ability may evolve, but the game will always remain a zero sum game - meaning, the losers pay for the winners. 
Cristian Eriksson
1058
Cristian Eriksson  
Amir Yacoby:
Who said AlphZero can't learn to trade? I'm sure it can, and very good. 

And in order to do so it doesn't have to predict the future, it can just find the holes in probabilities, where it stands the best chances. 

And also, if you love so much to talk in theory, then let's assume a future situation where all market participants will have a private alpha-zero to trade. Then, in that situation, alpha-zero will lose it's edge, and they will all lose money (in average) - at least the spreads+commision (in the best case they will lose only spread and commission - if all somehow manage to break even).

In current situation, given it's abilities, I assume it can have an edge over most humans and software currently in the market. But that's changeable. The traders ability may evolve, but the game will always remain a zero sum game - meaning, the losers pay for the winners. 
Agree what you say. But I'm thinking, that there could be more important factors why money flow in one direction , than oversee-able time trading.  Maybe politics and safety and other things are more important than profit when exchanging money to some. I guess the market will always have participants that doesn't care about if they loose or win when they exchange money, because they have other interests or business-ideas. So I would not say that forex is an isolated zero sum game where one has to loose and one has to win. It's only those strategies that starts to extract too much profit that will stop to work, and those that tries to trade, but cant read the market that will loose. The winners will be those that adapt to what the sum of the rest is.
Mrluck07
425
Mrluck07  
Amir Yacoby:
Who said AlphZero can't learn to trade? I'm sure it can, and very good. 

And in order to do so it doesn't have to predict the future, it can just find the holes in probabilities, where it stands the best chances. 

And also, if you love so much to talk in theory, then let's assume a future situation where all market participants will have a private alpha-zero to trade. Then, in that situation, alpha-zero will lose it's edge, and they will all lose money (in average) - at least the spreads+commision (in the best case they will lose only spread and commission - if all somehow manage to break even).

In current situation, given it's abilities, I assume it can have an edge over most humans and software currently in the market. But that's changeable. The traders ability may evolve, but the game will always remain a zero sum game - meaning, the losers pay for the winners. 

If Alphago can beat the market they are keeping it very private, i've found no news about AI's with a good edge on forex-stock market

Amir Yacoby
1222
Amir Yacoby  
Mrluck07:

If Alphago can beat the market they are keeping it very private, i've found no news about AI's with a good edge on forex-stock market

Because they probably didn't even try it yet, it's not their first interest to trade forex.
FxTrader562
157
FxTrader562  
Amir Yacoby:
Because they probably didn't even try it yet, it's not their first interest to trade forex.

You may be correct..because they are tech guys and they usually try to implement AI in real life situations like self driving cars, scientific research etc. Probably, they never tried or no one thought in that direction yet since as per their explanations the alpha zero algorithm succeeded beyond their expectations so quickly and it is a recent event of a couple of months or years.

I am very curious in implementing the alpha zero algo in forex and studying their white paper for long time, but when it comes to implementing it to MQL5 it is not a one man job. If many programmers will join, then we can create a dedicated thread and work together towards it. 

To add comments, please log in or register