Hello

I have been taking a course in Machine Learning, hoping to use it in an automated trading system.

I want to ask about your experiences in this. Anybody was able to use ML in an automated system?

Here are some research papers that claim they had positive results:

Modeling and Trading the EUR/USD Exchange Rate Using Machine Learning Techniques
https://etasr.com/index.php/ETASR/article/view/200/135
Results:
Trading strategies produced by the machine learning techniques of Support Vector Machines and Random Forests clearly outperformed all other strategies in terms of annualized return
and sharp ratio. To the best of our knowledge, this is the first application of Random Forests in the problem of trading with the EUR/USD exchange rate providing extremely satisfactory results.

Automated Trading with Machine Learning on Big Data
https://www.researchgate.net/publication/270824159_Automated_Trading_with_Machine_Learning_on_Big_Data/link/555ad73f08aeaaff3bfad2ec/download
Results:
The strategy turned out to be quite successful returning healthy profits at the average sharpe ratio = 2.08 and the average profit per traded unit equal to 3.2 ticks, which on average translates into around $30.

In the forum there is a guy @artemiso who would have if not direct, at least relevant tangential knowledge on this topic.

Off platform look up Earik Beann he developed a platform with significant features...I think it was called bee hive on Wave 59.

I hold both men in high regard and acknowledge that their specific gifts while compelling and interesting are outside of my regular ability/participation.

-Dan

Thank you Dan!

This course may give more insight on this subject:
https://www.coursera.org/learn/guided-tour-machine-learning-finance

I thought about that (again)..
wrote years ago, when i was younger..

[sorry, had to remove links the system wont let me yet... ]

the biggest issue you will have would be getting trading data (unless you can scrape it or pay for it)
I scraped lots of data off of yahoo... but intraday is like hens teeth... (i would be happy with 1 min data)

when i did it machine learning was still in its nascent beginner stage..
today it would be a lot easier (and cheaper), but again, it has to do with the data you have, the learning set and how carefully you pick your training... (i have been programming since thel late 1970s starting at bronx sci)

this was something i was thinking of doing again, but using this:
Movidious image removed due to 5 post limit...

aint that cool? Sure not like the old days!
Intel Movidius Neural Compute Stick 2 USB Stick - $69.99
wish i bought one before losing my long term job

The Intel Movidius Neural Compute Stick 2 is purpose built for computer vision and AI inference applications, making it faster and easier to develop products for network edge devices. This USB stick supports plug-and-play functionality to start prototyping right away.

The future of AI development is being democratized. No longer restricted to specialists, challenges such as computer-based perception and learning can now be tackled by the general developer and maker community. Hardware and software tools have become viable for mainstream use, in terms of both cost and complexity, and the knowledge to guide new participants in this field has become widely available.

[link removed]

Popular Deep Learning Frameworks
[link removed]

See: TensorFlow, PyTorch, PaddlePaddle (open framework) - and more

it also depends on what kind of machine you have too..

I made my own monster box (as i usually do)..
Processor Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz, 3501 Mhz, 6 Core(s), 12 Logical Processor(s)
32 gigs of 4ghz ripjaw memory
Titan X GPU

The above open source can run using the titan X

Deep Learning Frameworks
[link removed]

Deep learning frameworks offer building blocks for designing, training and validating deep neural networks, through a high level programming interface. Widely used deep learning frameworks such as Caffe2, Cognitive toolkit, MXNet, PyTorch, TensorFlow and others rely on GPU-accelerated libraries such as cuDNN and NCCL to deliver high-performance multi-GPU accelerated training.

I am hoping that some on this forum with complimentary skills could help in this area
Right now i have a few things to create for think or swim to help the human neural networks see things clearer

IF i can get an edge so that i can trade enough to live, then i can work on this kind of thing ad nauseum..

Be great to use my unique resources... people like me are very rare...
just getting into Bronx Sci is not easy, and my career is without degrees (thx feminism)
i am self made that way...

hope this helps..

**note i just wrote a very simple monte carlo sim of day AON (all or nothing) day trading patterns.
the most important things were to be big enough, and have enough hit rate...
the less you have to trade, the better your odds have to be
the more you can trade, the easier it is... so undercapitalized really meters outcome...

was going to share, but so far, no one has answered my TOS thread, and i am too new..