New ATH and Invitation to get involved in my Open Source project

Open Source project to make Machine learning development easy and accessible for producing automated trading systems for the stock market.

 

The gap in recourses and possibilities between institutional- and private investors is shrinking all the time. This is true in terms of capital, trading platforms and information access. But it is not yet true in terms of system development cooperation and easy access to the powerful statistical tools of Machine Learning and Big Data.

 

Capital – Before (+20 years) you needed large sums of capital to minimize the trading related cost of commission and spread between the bid and ask price. Due to competition and order leverage those costs are significant lower now. This means that more theoretical trading strategies are possible to try, even for small size investors. This is especially true in the forex market.

 

Platforms – It is not longer necessary to have access to a Bloomberg Terminal ($20K/year) to be a full time professional trader. Most of the retail platforms today are very sophisticated and comes with a relatively low cost, some are even free to use.

 

Information – In our present Internet age, all the basic market information is mostly free and easy accessible. And due to the giant amount of free social media content, all kind of exotic information is also easy accessible.

 

PROBLEM WITH MACHINE LEARNING & BIG DATA

 

The problems today if you want to use machine learning to create a fully automated trading system is that you need to build your own integrated platform. Most of the established trading system development platforms use their own programming language, with so far no equivalent to Pythons famous machine learning library Scikit-learn. Without a pre-coded library, the trader needs to code the whole machine learning algorithmic from the beginning. Some of the latest and most popular trading communities offer the possibilities to use Python and Scikit-learn at their platforms, but due to restrictions in capacity, the full use of big data and CPU power is not satisfying for a machine learning purpose.

 

MY SOLUTION FOR MACHINE LEARNING & BIG DATA

 

I have solved all the above problems and today a have a fully automated trading system running since April 2017. I have built the system myself with around 4000 lines of Python- and MQL4 code. This includes a preprocessor, an optimizer, a predictor and a trading robot (Expert Advisor) for the MetaTrader4 platform.

 

PROBLEM WITH DEVELOPMENT COOPERATON

It has not been established many standards in the field development of trading system, the reason is that many private traders are working from home with various platforms and programming languages. Together with the problem that not so many people want to share their success or failures, due to competiveness, shame or privacy. In the recent years a couple of large communities has appear, but they are in the early stages and in my opinion there are room for a lot more collaborations in the system development area.

 

MY SOLUTION WITH DEVELOPMENT COOPERATON

So far I have built a large online social network in Sweden with around 6000 twitter (mostly trading/investor) followers, 450 members in a Facebook group in Swedish called “Machine Learning and robot trading”, a webpage with around 200 hits a day, YouTube channel with 100 followers and I got in personal contact with around 10 of the 80 Swedish hedge funds. With controlling those information platforms and channels it will not be a hard start for me to get even more people deeper involved in development cooperation. My opinion is that with a fully function development toolset It will be even easier.

 

PROJECT GOAL

 

First I want to attract more skilled and motivated people, both to develop the platform and to develop trading ideas.

Second I want to build standards around the cooperation and development of trading systems with the platform.

Third I want anyone involved get easy access to test and use the best trading systems coming out from the project. Letting other use the trading systems is achieved with either using the code to create fully automated trading, or semi-automated where the predictions are created without automated trades or with having the system developers put their systems on any of the largest social trading platforms to be copy-traded.

 

ADDED VALUE

 

Investors – The investors may only use 10-50 percent of the intended capital directly in to the project/company. The rest of the money is separately used on the best trading strategies coming out from the project. The part used directly to the project could be used to shareholder steering on what asset classes and instruments that the RnD should be targeting. Later it could also be possible to get revenue from:

 

  • Selling the latest Trading Strategies and letting old ones be free
  • Selling latest RnD Techniques from the platform and let old ones be free
  • Making license arrangements with brokerages, hedge funds or platforms provides
  • Getting compensations from different social trading platforms when acting as a portfolio manager

 

Volunteer programmers – Work experience and/or using existing experience help them with private savings. I have meet with several seasonal programmers that feel that they want to contribute in some way and then letting the trading system hopefully create a passive extra income.

 

Students – Getting fast and coached in to real life machine learning problems and solutions. Students also got the possibility to invest in the best strategies.

 

CONTACT INFO: mikael [at] furesjo.se

ZuluTrade Account: https://www.zulutrade.com/trader/354968/trading

Facebook group: https://www.facebook.com/groups/Borsrobotar/

Youtube Channel: https://www.youtube.com/user/FuresjoFinancialTrad/videos

Twitter: https://twitter.com/Mikael_Furesjo

 

 

EXTERNAL INFORMATION

http://socialtradingguru.com/networks/social-trading-networks

https://www.quandl.com/

https://www.quantopian.com/

By | 2017-11-12T21:46:15+00:00 November 12th, 2017|RnD|0 Comments

About the Author: