Machine Learning is the new C++

[Cdntrader]

IMO we are probably 20-30 years behind what we could have been if we had just called these tools "automated statistical and probabilistic methods" or something like that and divorced the language from these science fiction fantasies.
The most pernicious aspect of this is it creates an artificial benchmark so any actual progress is judged to be unworthy compared to the science fiction fantasy. That has to have real world effects on funding, motivation, time spent on problems, etc.
To say machine learning has no use for trading is like saying a tool box is not good at drilling a hole in the wall to hang a picture. It is missing the most basic concept that the tool box is a container for different tools for different jobs.

Good points. I agree.

 

[botpro]

Biological evolution is slow.
Artifical evolution will be superfast.
And that means: robot brains will outperform human brains...
Good morning, mankind!

Humanoid robots will do slave work (mining etc.) on other planets like Mars, the moon etc., and will be used in the military as supernatural killer soldiers.

And C++ is the best language for AI, ie. for bootstrapping the robot brains.

 

[eusdaiki]

Found an interesting article.

It states that there's no such thing as machine learning, it's just learning.
It takes a few premise that basically argue that the distinction between machine and organism is not as clear as one would assume.

quote:

An ecorithm is an algorithm, but its performance is evaluated against input it gets from a rather uncontrolled and unpredictable world. And its goal is to perform well in that same complicated world. You think of an algorithm as something running on your computer, but it could just as easily run on a biological organism. But in either case an ecorithm lives in an external world and interacts with that world.


So the concept of an ecorithm is meant to dislodge this mistaken intuition many of us have that “machine learning” is fundamentally different from “non-machine learning”?

Yes, certainly. Scientifically, the point has been made for more than half a century that if our brains run computations, then if we could identify the algorithms producing those computations, we could simulate them on a machine, and “artificial intelligence” and “intelligence” would become the same. But the practical difficulty has been to determine exactly what these computations running on the brain are. Machine learning is proving to be an effective way of bypassing this difficulty.​

 

[Cdntrader]

FWIW Google did a series of presentations on using ML/DL in the cloud recently. A Lot of concrete examples from business were presented. Very informative.

It's pretty wild how fast this stuff is rolling out to industry. Google used DL to revamp their Google Translation app in 9 months last year. Tensorflow use is exploding and almost every product Google has, now uses ML/DL in some way.

https://www.youtube.com/playlist?list=PLIivdWyY5sqLq-eM4W2bIgbrpAsP5aLtZ

 

[sle]

http://iknowfirst.com/machine-learning-trading-stock-market-and-chaos

now, that looks like a proper scam

My PhD was in an SML-related field and over my years in finance I tried to find a proper use for them. The only real application that I've seen is the use of NLP to process the news and earnings calls, otherwise it's kinda useless. Most effects can be captured using much simpler statistical techniques and second order effects have such complex causality that there usually insufficient input data for a good learning data set. I think it's an exciting field overall, but it has very little application to finance.

 

[Cdntrader]

Yes it does look like a scam.


ML is already being applied to finance. Not by me though.

 

[eusdaiki]

now, that looks like a proper scam

My PhD was in an SML-related field and over my years in finance I tried to find a proper use for them. The only real application that I've seen is the use of NLP to process the news and earnings calls, otherwise it's kinda useless. Most effects can be captured using much simpler statistical techniques and second order effects have such complex causality that there usually insufficient input data for a good learning data set. I think it's an exciting field overall, but it has very little application to finance.

Which statistical techniques have you found to be effective ?

 

[sle]

Which statistical techniques have you found to be effective ?

Depends for what type of analysis. I use robust regression a lot, a fair bit of simple co-integration analysis using regular tool, some noise reduction techniques. No ML or anything fancy.

 

[Cdntrader]

Hedge Funds Are Training Their Computers to Think Like You

https://www.bloomberg.com/news/arti...close-in-on-designing-ultimate-trader-s-brain

 

[Zzzz1]

I would disagree entirely with you on this, if you applied your statement to AI in general (or did you just mean that NLP is not much applicable to finance?).

I personally developed and use a classification algorithm that with positive expectancy predicts regime changes. And that is in the core trading field. I can list tons of different applications for AI to be deployed in non-core trading but still finance affiliated areas. First and foremost in risk management, but also in order management/DMA, regulatory environments, detection of insider trading,....

now, that looks like a proper scam

My PhD was in an SML-related field and over my years in finance I tried to find a proper use for them. The only real application that I've seen is the use of NLP to process the news and earnings calls, otherwise it's kinda useless. Most effects can be captured using much simpler statistical techniques and second order effects have such complex causality that there usually insufficient input data for a good learning data set. I think it's an exciting field overall, but it has very little application to finance.