When do you think AI will dominate trading field?

[IAS_LLC]

my opinion is that AI is very good at deterministic modeling - those things where inputs and outputs are governed by certain relationships that can be modeled (even mildly chaotic relationships). These could include things like weather forecasting, medical diagnosis (robotic doctors/surgeons), software that could replace accountants+lawyers, self driving cars, etc.

But the financial markets are something else altogether - they do not follow any physical or natural laws. Daily price movements are governed by factors that are at least an order of magnitude more complex and discontinuous than any other system we know of. I don't think there will ever be an AI program that will achieve 100% accuracy in predicting market price direction one day in advance. If such a program came about in the future, there eventually would be multiple players competing for the same profits, thereby reducing the forecast efficacy.

Take a look at Bayesian Machine Learning. In a nut shell, really good machine learning algorithms take into account the fact that nothing can be predicted with certainty, but we can provide a best guess, and an estimate what the uncertainy surrounding that estimate is. Sort of like how the Kalman Filter estimates the mean and covariance of a stochastic process.

I use machine learning for my real time trading system, and in no way do I expect it to 100% accurate. I just expect it do be able enter trades where the expectation is positive (i.e over a large number of trades, I expect to be profitable) and the variance on that expectation is withing my risk managment tolerances. Even if I expect it to be positive, I don't expect it to work on its own on any market.... there is a lot of human touch required to tune these algorithms. And the algorithm doesn't find the opportunity on its own, the human has to show it the way.

 

[Money Trust]

All of this electronic trading and markets still move the same as they did 100+ years ago.

 

[wrbtrader]

my point is, there will be time when big companies,hedge funds, will have so much computer power and so strong AI/algoritms that they will simply crash all the "small" (retail/individual) traders and the latter won't be able to survive in trading at all

but the question is when is this going to happen?

More and more "small" (retail/individual) traders are using algorithms. In fact, more forums have algorithm sections where these traders talk about their algorithms. This forum (ET) has algorithm traders and growing.

Simply, the retail trader has consistently adapted or adopted new technologies as the rest of the world since the birth of Wall Street. Those that don't...they get left behind.

Also, keep in mind that Microsoft, Google and a few others are designing computers (not yet available for the masses such as quantum computing) for the average consumer that will be using latest technologies used by the big boys as they have always done so in the past.

The issue really is this...when its available...will you use it or just sit around and bitch about how difficult trading has become (not able to survive while others can survive) ?

 

[O(1)]

my point is, there will be time when big companies,hedge funds, will have so much computer power and so strong AI/algoritms that they will simply crash all the "small" (retail/individual) traders and the latter won't be able to survive in trading at all

but the question is when is this going to happen?

it's not gonna happen. in my opinion, it seems more likely that "AI" would blow things up.. with unaccounted risk with bugs

 

[pinabetal]

all over the world companies invest in AI/machine learning and that field is developing exponentially, time will come that AI will perform nearly every possible task faster and better than any human can possible do (already happens in a lot of fields and robots starting to take jobs from people faster and faster)

when do you think that will happen to trading? when will AI become so powerful and intelligent that it will be able to beat any human in that area with ease?

Not in our lifetime and not with present technology.

 

[learner2007]

Ai, algos, HFT, machine learning etc. All great for certain tasks, I guess. But is a guy with a Ph.D really going to take off 2+ years to learn TA, and then another few years to learn how to trade just so that he can attempt to create a program that will defeat all the other programs and retail traders??
And in the end, as one of the world's greatest traders once said "There will always be trends in the markets as long as there are people (and algos!) who can't read them".

 

[Overnight]

...
The issue really is this...when its available...will you use it or just sit around and bitch about how difficult trading has become (not able to survive while others can survive) ?

If the world starts to get funky and all currencies weaken because some Mad Max thing is coming, gold won't go down, it will go up. This is not something that can be programmed into a box.

So you let your algobox short the thing, while I, with my human brain, long the thing. Human minds and discretion is going to win in the end. Not "algo". Remember: All things automatic are programmed by...You guessed it...Humans!

 

[IAS_LLC]

If the world starts to get funky and all currencies weaken because some Mad Max thing is coming, gold won't go down, it will go up. This is not something that can be programmed into a box.

So you let your algobox short the thing, while I, with my human brain, long the thing. Human minds and discretion is going to win in the end. Not "algo". Remember: All things automatic are programmed by...You guessed it...Humans!

What makes you think the algobox won't go "long the thing" before you do ?

 

[vanzandt]

Simply, the retail trader has consistently adapted or adopted new technologies as the rest of the world since the birth of Wall Street. Those that don't...they get left behind.

 

[vanzandt]

oops

More and more "small" (retail/individual) traders are using algorithms. In fact, more forums have algorithm sections where these traders talk about their algorithms. This forum (ET) has algorithm traders and growing.

Simply, the retail trader has consistently adapted or adopted new technologies as the rest of the world since the birth of Wall Street. Those that don't...they get left behind.

That is not exactly true. Retail traders, as we know them at least, didn't really start popping up until the 90's. The internet changed everything.

This entire thread is funny. And to be honest, its no wonder 99% of the people out there "blow up" and lose it all several times over.

It seems to be some kind of mantra here on ET that one has to spend years and lose piles of money to figure out some kind of "edge". There's one post on here (ET) this weekend where someone advised someone else to "keep their day job and keep losing money until you get it right" WTF? You guys are all nuts, and I'm not trying to be mean here, but you are all playing a losing game.

I'll be honest...I have never been around an HFT operation, or a prop shop, but these are concepts that are what--- for the most part less than 20 years old? And how many of them have went tits up? A lot. And how much money has been lost? A lot.

Then one counters..."But Zandy....for every loser there's a winner.... trades are two sided. Somebody's making that money". I'm sure that's true. But its not going to be anyone that ever reads these words that I am writing right now.

A whole industry has popped up that makes people think they can get rich and make a living trading. Its too funny. Look at the commercials. IB's girl on the date: "Hold on honey, a NATO plane just crashed...I have to make a trade". Or some of E-Trades with that goofy looking dude riding around some office on a Segway learning about E-Trades powerful trading systems. Housewives changing a diaper talking on a phone saying "I just made $300 trading with my Fidelity Active Trader account".... Its f'ing is comical.

There's only one way to make REAL money in the markets. That is to buy and hold good quality company's when their stock is cheap. Or short the pigs and hold on.

The rest is just pissing in the wind. You might find the "holy grail super fly 'edge'" one week or one month, but I suspect most lose it all the next month, or day ....or whatever.

If there's people out there that are consistently winning because they have a 1/10,000th of a millisecond advantage on a data feed...or whatever it is you guys like to chat up....more power to em.
At least daytrading with a system is not gambling...its using programing as a way to attempt to skim dollars. But its semantics...who cares what you call it,.... ("gambling" has a negative connotation that most hardcore day-traders despise and scream at ("I'm not a gambler!))....losing money is still losing money

And swing trading (which I enjoy), ......playing earnings or events......ahhhh...now that is gambling. The edge there is the ability to predict the herd. Kinda like counting cards works in blackjack. But its still gambling.

You can all sit here and say I'm wrong, but I suspect that's denial. The only way to really make money in the markets is to use the markets as they were designed 100 years ago. Buffett/Lynch style.
The rest of this is just a government/bank/broker/media sanctioned game of Three-card Monte. It really is.

Nothing has passed anyone by. In fact, it hasn't changed one bit.