machine learning

>>2340086
I've never seen a single shred of evidence that AI can do it consistently. If someone knows, they aren't telling anyone how to use it.

>>2340140
I'm assuming computer, AI, and machine learning technology isn't to the point where it is capable of doing anything noteworthy yet, but my question is will it ever be.

Bumping for interest

>>2340199
From what I've read, ai is getting returns of about 2% on the nyse. I think assisted investing will be as far as it goes

File: 1494189739958.gif (504KB, 1080x1080px) Image search: [Google]

504KB, 1080x1080px

End game, yes. There's no reason AI can't be impossibly better at everything than humans, including trading.

Just a matter of timing. If you believe people like Ray Kurzweil, maybe by 2045.

There are too many factors that affect stock prices. In shorter timescales, it's practically gambling, because the only factors in play are the chaotic traders themselves. You all know this.

But, taking data from a conglomerate of inputs and spitting out something halfway usable is an AI's forte.

Let's assume we build an AI to predict stock prices for a real company (not a shell company). If you build an AI to take in news sources, and use an algorithm to determine the context and tone of the article, you can suspect with good probability that the stock price will go up or down based on the article. Obviously, one article is not indicative of a particularly impactful change in price. However, if many articles are indicating a positive or negative tone, or if many articles are reprinting from already-seen sources, we can assume that it is impactful, and the probability of a price change will go up.

This technique has obvious demerits. It's bottlenecked by the output of articles. It requires a lot of processing time. As such, it's only suitable for swing and long haul trading and is not expected to bring good returns. Articles are biased and lie often, and as such are often not very good indicators. Still, we take it as obvious that businessmen tend to not understand the gritty details of the businesses they decide to invest in, and that biased or untruthful articles are not worth worrying about--the only thing that matters is the perceived sentiment towards a stock.
Some articles have nothing at all to do with a stock yet can impact stock prices immensely (i.e. the election of Trump). Essentially, the limitation of AI is the creativity of the human behind the design.

>>2340086
No because human emotions lead to irrational trades. Unless the computers can predict irrational actions then no.

>>2340308
An example of how this works in play
http://www.cnbc.com/2016/10/28/donald-trump-will-win-the-election-and-is-more-popular-than-obama-in-2008-ai-system-finds.html

>>2340328
A super AI definitely would be able to predict irrationality.. The AI would be the first to read whatever article or event that cause people to panic sell, and it would respond in the perfect way no doubt