Natural Language Processing

[2weels]

I don't recall specifically where, but I've seen discussion that some quants use natural language reporting (like articles in investing publications, Wall Street Journal, Barron's, industry-specific newsletters, information from area-specific analysts, etc etc) as input into their trading algorithms. Does anybody know more about this? Examples of this being used, what sort of algorithms are used for NLP (it seems the deep learning area has risen into prominence in this above traditional syntactic/semantic/symbolic parsers).

I wasn't able to find much on text to proportional logic translation since 2011 but in the deep learning field things are quite active. See the Stanford NLP group for example.

But how is this then used in trading algorithms? Kind of like sentiment analysis, which judges some verbiage (from an expert analyst or trade publication or .....?) on a positive - negative continuum and treating this as a factor to add into the mix in making predictions?

Any more substantive contributions welcome....

 

[globalarbtrader]

A lot of these things (NLP, AI, deep learning,... ) are like sex for teenage boys in high school. Everyone claims to be doing it loads, but the truth is that nobody has been able to do it properly, but they don't want to lose face admitting the truth. There is just one cool guy who has actually done it, and everyone knows who it is, but he isn't giving away any details- just sitting there looking damn smug the whole time (in this metaphor, Jim Simons).

GAT

 

[traider]

Jaffray Woodriff also claims to have achieved transcendence

 

[dumpdapump]

That is not accurate. Quite a number of shops nowadays not only employ NLP algorithms for sentiment analysis but also Deep Learning Networks. I am not sure how cool Jim Simons is but he is definitely not the only game in town.

A lot of these things (NLP, AI, deep learning,... ) are like sex for teenage boys in high school. Everyone claims to be doing it loads, but the truth is that nobody has been able to do it properly, but they don't want to lose face admitting the truth. There is just one cool guy who has actually done it, and everyone knows who it is, but he isn't giving away any details- just sitting there looking damn smug the whole time (in this metaphor, Jim Simons).

GAT

 

[fordewind]

That probably refers to the Speach Recognition language

 

[2weels]

That is not accurate. Quite a number of shops nowadays not only employ NLP algorithms for sentiment analysis but also Deep Learning Networks. I am not sure how cool Jim Simons is but he is definitely not the only game in town.

dumpdapump, do you know of any articles, news, or anything that talks more specifically about how NLP and Deep Learning are actually used? Most particularly NLP.

Thanks.

 

[dumpdapump]

Specifically targeting financial trading? There is not much in public domain but I know for certain that multiple hedge funds employ learning algorithms and staff their quant research seats increasingly with those who possess solid AI skill sets.

There is more out there in NLP space. You can search for "sentiment analysis artificial intelligence" or similar key words. I know a guy who applies natural language processing to tweets and uses such as input to some of his quant models. But he is quite secretive about it hence I don't know much details of how he structures his framework.

dumpdapump, do you know of any articles, news, or anything that talks more specifically about how NLP and Deep Learning are actually used? Most particularly NLP.

Thanks.

 

[2weels]

Specifically targeting financial trading? There is not much in public domain but I know for certain that multiple hedge funds employ learning algorithms and staff their quant research seats increasingly with those who possess solid AI skill sets.

There is more out there in NLP space. You can search for "sentiment analysis artificial intelligence" or similar key words. I know a guy who applies natural language processing to tweets and uses such as input to some of his quant models. But he is quite secretive about it hence I don't know much details of how he structures his framework.

Thanks, good idea to search for sentiment analysis and AI, it might get close to how people are using the natural language articles etc to add into their strategies. There was a story a month or two ago that people did sentiment analysis on Trumps tweets, and when they were positive the market went up, and negative it went down.

Probably need something a little more advanced than that.

Thanks again.

 

[HobbyTrading]

I recall a programming series on YouTube by a guy called sentdex. He made a Python program to do sentiment analysis. I'm not sure if that was specific to tweets, or that it can be used on other written material as well.

 

[2weels]

I recall a programming series on YouTube by a guy called sentdex. He made a Python program to do sentiment analysis. I'm not sure if that was specific to tweets, or that it can be used on other written material as well.

Thanks. There is a significant amount out there on sentiment analysis. I'm wondering how it's applied to trading. So, for example, a newsletter has mention of a specific company with positive sentiment, then you bet on that company to go up? It must be something like that. Then you must have feeds for reams of natural language info about companies.

Or, one could just gauge the sentiment of the trading community by inputing Bloomberg news, Marketwatch, even Cramer. He might blow a tube in the computer though.
So when the community's sentiment is up, the market will be up, and so on.