Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Pipimint Fresh Chart Sharing-Where Feedback Flows & Intelligence Grows 34 replies

Artifical Intelligence EA 28 replies

Artificial Intelligence EA 16 replies

Attachments: Do we need Artifical Intelligence in trading?
Exit Attachments

Do we need Artifical Intelligence in trading?

Quoting jdieieh
{quote} Have you seen work of Boston Dynamics and alike? Have it ever crossed your mind that actually it might work like the Terminator and it wont be just a sci-fi movie. As Jules Verne's submarine fantasy is actually reality today. Most of those concepts are already alive and working on battle fields. How much battle field has changed from 100 year ago, when a peasant took his rifle and went shooting. Today is not even heavily armored soldier with self aiming rifle that matters. But a swarm of armed drones able to cooperate and ambush. War was...
Hypothetical case in point.

Inserted Video
It is not hypothetical. It is already being implemented, we just dont know all and what we know is already advanced.

https://www.newscientist.com/article...rom-a-missile/

http://www.popularmechanics.com/mili...arming-drones/

https://www.rt.com/news/350736-russi...s-drone-swarm/

Inserted Video
Quoting jdieieh
It is not hypothetical. It is already being implemented, we just dont know all and what we know is already advanced. https://www.newscientist.com/article...rom-a-missile/ http://www.popularmechanics.com/mili...arming-drones/ https://www.rt.com/news/350736-russi...s-drone-swarm/ https://www.youtube.com/watch?v=uwQ3gdmA7sQ
Not so hard to achieve. Just buy a few drones from the local retailer. Install sensors and develop stigmergy algorithms to communicate between drones with a coordinated objective....voila. It makes me wonder how this is all going to be effectively and ethically regulated. As advanced as we may be with AI....it's the lack of advancement in our thought processes that has me wondering.
Quoting Copernicus
As advanced as we may be with AI....it's the lack of advancement in our thought processes that has me wondering.
Maybe we are just a next manifestation or Fermi paradox
Quoting jdieieh
{quote} Maybe we are just a next manifestation or Fermi paradox
Perhaps just another step on the ladder to the Borg :-)

We're home Mum. What's for Tea? Jay got into a fight and lost his eye....again!!!
Attached Image
Quoting jdieieh
{quote} Have you seen work of Boston Dynamics and alike? Have it ever crossed your mind that actually it might work like the Terminator and it wont be just a sci-fi movie. As Jules Verne's submarine fantasy is actually reality today. Most of those concepts are already alive and working on battle fields. How much battle field has changed from 100 year ago, when a peasant took his rifle and went shooting. Today is not even heavily armored soldier with self aiming rifle that matters. But a swarm of armed drones able to cooperate and ambush. War was...
I agree with you but how can we know it is still embryo not a grown man? All latest technology informations are being shared with public? Correct me if i am wrong but internet was being used by US military more than 2 decates before revealed to public, named arpanet.
Maybe AI has already involved but we are aware? I have looked last bis reports, havent seen clear info about this.

Read an article here about AI trading maybe a year before, one european (i think german) fund invested huge money to AI trading and researches took some time without succcess and they decided to end the program. dont remember exactly, something like this. So it is not certain they will succeed.

One more thing, there is also quantum processored computers thing. Maybe they will be game changer or maybe already involved. Do you have some infos about that?
And how exactly AI could possibly beat the OCHL trading styles?:-)))))

On the other hand, if the trader is looking for the Moon to rise on the fifth day after the Fed's chairman took a shit, at open skies, on the lawn of the Fed's headquarters, but still waiting for a brown pigeon to CONFIRM the delivery (if it is not a brown pigeon, IT IS NOT VALID SETUP!)... Then, any AI will just blow that trader away...
Maadik Hugiis. IQ 69.
  • Post #48
  • Quote
  • Edited Dec 4, 2017 12:02am Dec 3, 2017 10:34pm | Edited Dec 4, 2017 12:02am
Quoting cbo
{quote}One more thing, there is also quantum processored computers thing. Maybe they will be game changer or maybe already involved. Do you have some infos about that?
The race for the first quantum computer is still being pursued actively but there is still a long way to go. In Aussie we are getting close with the flip-flop qubit......however the key lies in the development of an environment where a quantum superposition of state (quantum entanglement) is maintained long enough to allow for qubit processing, without associated signal interference and the collapse of the superposition into a classical state.

Irrespective of whether they are able to achieve this feat of technology the question then boils down to the nature of the markets themselves. Yes, qubit processing will be able to vastly speed up computer processing allowing for superfast identification of anomalies....however provided order front-running is prevented then ultimately the question boils down to whether any advantage can be taken from a quantum as opposed to classical interpretation of market activity. Given that the fundamental unit of a transaction is classical (eg. a bit) and not quantum (an emergent feature of a qubit) then I personally don't believe that more powerful computers are the solution. If order front-running continues then the game will be lost to AI (technology and latency).

Provided order front-running is prevented.....unless you can predict with precision the exact timing and reason for all participant decisions (and this is beyond what any computer can access as it requires access to ALL possible data), then risk will always be a central feature of modern markets. Market anomalies that operate outside normal rules of efficient markets with non-linear power laws are emergent features of complexity and the more efficient they get, the more prone to catastrophic risk.....a lesson Nassim Taleb and Benoit Mandelbrot teaches. With uncertainty comes risk and it can never be unhinged from return provided perfect information is not attainable. Markets are not perfectly closed systems. Impacts external to the market can and do have material effects on the markets themselves through cascading dependencies. For example an individuals personal financial financial position..a cat running over a keyboard......etc etc etc.......namely events external to the markets themselves (tsunamis, wars etc etc etc) that cause selling/buying frenzies etc. ensure that fat tails will remain that simply cannot be predicted irrespective of processing power.

The game gets harder for humans with AI in that markets get more efficient however 'uncertainty' (being lack of total information) will always have the last laugh.
There are many easy to deal in with the Forex Market, you need to have a good amount of knowledge and experience and skills to make some good amount of profit in the Forex Market.
Lets assume for the moment that AI will be dominant in the forex market, what are we likely to see as a result?
Will the movements be merely speeded up?
Will spikes be even more dramatic, or in fact lessened?
Will AI use strategies that institutional traders already employ e.g. counter moves to grab better prices?
Will only daytraders get hammered?

Intelligent speculation would be appreciated.
Quoting profitkeeper
Lets assume for the moment that AI will be dominant in the forex market, what are we likely to see as a result? Will the movements be merely speeded up? Will spikes be even more dramatic, or in fact lessened? Will AI use strategies that institutional traders already employ e.g. counter moves to grab better prices? Will only daytraders get hammered? Intelligent speculation would be appreciated.
My hunch is that we will see two things happen with AI dependent on whether:

  1. AI uses 'dumb' programming (eg. non-adaptive algorithms); or
  2. AI can 'self' problem solve (eg. adaptive algorithms).

The former will result in impacts that simply relate to their ability to rapidly respond. The quest for latency . But their side effects could be more intense spikes, propensity to cause runaway markets and flash crashes etc. Impacts associated with this form of AI could be quite unpredictable akin to irrational human behaviour when the **** hits the fan.

The latter would perhaps create hitherto new market structures as they adopt a presence more akin to a human participant but with deliberate 'rational intent'. These IMO have the ability to create new emergent market structures if they evolve in time with the market. Tied to their adaptive nature would be the over-riding objective linked to their survivability....eg. ability to score high in risk weighted returns etc. and their ability to reproduce etc.

AI that has had a chance to evolve with the market will have different inherent capabilities than humans who have evolved in a different system and are trying to navigate an alien environment (aka the financial markets).

Just my hunch :-)

  • Post #52
  • Quote
  • Edited 7:51pm Dec 4, 2017 2:13am | Edited 7:51pm
Quoting profitkeeper
Lets assume for the moment that AI will be dominant in the forex market, what are we likely to see as a result? ...
The attempt to use AI for trading had actually begun with the development of HFT algorithmic programs, which are already running as of today ...

From what I understand, the current AI developments are geared more toward recognizing “good” traders and “bad” traders ... the program will then follow whatever trades the “good” traders are putting in the market, possibly in a bigger size as the AI developers usually have bigger capital ...
Given that’s the case, the market should move substantially faster and/or further in a direction when the AI programs are following the same trades as the “good” traders ... But again, there will be some discrepancies or different opinions even among the so-called “good” traders ...

So IMHO and as the forex market has been around for decades, this “irrational” market will behave the same as today but at a bigger volume ... $10T a day???
Market is not random but unpredictable
  • Post #53
  • Quote
  • Edited 3:08am Dec 4, 2017 2:41am | Edited 3:08am
Quoting Copernicus
Given that the fundamental unit of a transaction is classical (eg. a bit) and not quantum (an emergent feature of a qubit) then I personally don't believe that more powerful computers are the solution.
You are putting equal sign between AI development and increase in computational power. But that is not the case. Although more computing power is useful, the key is way of thinking.

Let me come back to an example I provided before and I guess nobody bothered to read. AlphaGoZero beat AlphaGo at Go game 100 times in 100 games. Total defeat. Yet AlphaGoZero uses less and less computing power and at the same time achieves higher and higher ELO ratings and better and better real life wins. Clearly not because of more computing power, but because newer, more intuitive way of thinking.

Attached Image (click to enlarge)
Click to Enlarge Name: Capture.PNG Size: 92 KB


Speed of thinking is not sign of intelligence. Einstein was know for being a slow thinker, yet nobody can deny his intelligence. Proverbial Rain-man was faster than a calculator, yet dump like a penguin.

Quoting Copernicus
...unless you can predict with precision the exact timing and reason for all participant decisions (and this is beyond what any computer can access as it requires access to ALL possible data), then risk will always be a central feature of modern markets.
And how do you know how advanced AI will think? Nobody knows at the moment... hence you make mistake of projecting AI's capability based on what human can do now or what primitive Narrow Artificial Intelligence can do today. But that might not be a case. AI I have in mind (or people like Kurzweil) will be so different from us in a way of thinking that will not be able to follow.

In a case of the markets, all you know you today about markets and how to speculate, will have to go to a bin and learning process will start from scratch. And how quickly wil you be able to re-learn and start wining against that AI if it will able to change its strategy fluidly on fly to adapt to behaviour of crows that try to catch up?

That is my point.
Quoting 4for4
{quote} The attempt to use AI for trading had actually begun with the development of HFT algorithmic programs, which are already running as of today ...
AI was operating in the markets in the 90s if not earlier.
  • Post #55
  • Quote
  • Edited 3:24am Dec 4, 2017 3:13am | Edited 3:24am
Quoting mbrown
{quote} AI was operating in the markets in the 90s if not earlier.
What exactly AI do yo mean. NAI, GAI or SAI?

HFT is nothing more than fast calculator.

90s systems you are referring to are known as expert and advisory systems.

Do you realize, we are talking here +GAI? We are talking here about actual intelligence.
Quoting jdieieh
{quote} What exactly AI do yo mean. NAI, GAI or SAI? HFT is nothing more than fast calculator. 90s systems you are referring to are known as expert and advisory systems. Do you realize, we are talking here +GAI? We are talking here about actual intelligence.
Buddy, I know what expert and advisory systems are and I am not referring to those. I am referring to relatively deep adaptive recurrent networks. I am also referring to reinforced AI algos. It may also help if you define your definition of actual intelligence (as you call it); is your definition related to the Turing test or do you have something else in mind?
Apart from the fantasies on FF.

Ai simply analyzes all the possibilities and finds the best solution. it cannot anticipate anything.

"Computers are incredibly fast, accurate, and stupid. Human beings are incredibly slow, inaccurate, and brilliant. Together they are powerful beyond imagination." —Albert Einstein

and I don't agree with him that human beings are necessarily "incredibly slow and brilliant" lol
Quoting HeyYou
Apart from the fantasies on FF. Ai simply analyzes all the possibilities and finds the best solution. it cannot anticipate anything.
In reality, that is all humans do except we can't analyze all possibilities and we deal with fuzzy outcomes more efficiently although AI algorithms have been developed which are quite good in this area such as the Support Vector Machines.

Anyway, where did jdieieh runaway to? Seriously, anyone who writes the only trading AI prevalent in the 90s was expert systems/advisory systems simply isn't worth having a conversation with.

Back to the real world of AI trading.
Green pips y'all!
Quoting mbrown
{quote} In reality, that is all humans do except we can't analyze all possibilities and we deal with fuzzy outcomes more efficiently although AI algorithms have been developed which are quite good in this area such as the Support Vector Machine.
we are slower and kinda inaccurate in analyzing the possibilities. Ai is faster... that's it!
Quoting HeyYou
{quote} we are slower and kinda inaccurate in analyzing the possibilities. Ai is faster... that's it!
You made a theory... now show me supporting evidence. Ideally refer to research within last 10 years. Otherwise we are talking about shape of clouds, everyone sees what they want to see...

So far I have shown practical, working, improving examples of (embryo) NAI surpassing or nearly surpassing human abilities. And as a opposing argument I am given opinions. In this case it is not even debate.... but then what one would expect from forex forum...
0 traders viewing now