Search
× Search
Sunday, December 22, 2024

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

Trading Using Machine Learning In Python - SVM (Support Vector Machine)

photo

 Milind Paradkar, Senior Manager - Technical Content at QuantInsti | Algorithmic Trader

 Tuesday, August 8, 2017

https://www.quantinsti.com/blog/trading-using-machine-learning-python-svm-support-vector-machine/


Print

4 comments on article "Trading Using Machine Learning In Python - SVM (Support Vector Machine)"

photo

 Valerii Salov, Director, Quant Risk Management at CME Group

 Sunday, August 13, 2017



Machine learning, ML, roots are the works of Vapnik and Chervonenkis. SVM is one of many. Inherently, linear nature of some of these techniques may not well interact with non-linear, chaotic and/or random prices. There are topological methods linearizing the data manifold and also helping to reduce dimension (the number of factors) of the task. Nevertheless, presented demonstration how SVM can be adopted to prices and trading is valuable because the principals of switching to other methods, possibly more robust, are common. Python is nice. For C++ enthusiasts Shark 3.1 is useful. Again, a danger is that many ML methods can be easily applied under conditions, where they are not supposed to be applied. This may create an illusion of success overridden by failures in real trading. A "scientific dress" can hide an "astrological nature of investigation", if determination that a method corresponds to conditions is forgotten. Certainly, thank you for posting. Best Regards, Valerii


photo

 Valerii Salov, Director, Quant Risk Management at CME Group

 Sunday, August 13, 2017



When one in O'Hare airport inserts a biometric passport into a scanner, [s]he rarely thinks about machine learning, ML. Patterns of typed or handwritten characters vary but have a "static nature" and can be classified using SVM. Application of such methods to prices is massively undertaken by companies and individuals. In contrast with posted material, that research is rarely published. Illustrations of "financial" neural nets, wavelets and principal component analysis preprocessing is found in: Lequeux, Pierre (Editor). "Financial Markets Tick by Tick", New Work: John Wiley & Sons, 1999. A friendly description of ML is Bishop, Christopher, M. "Pattern Recognition and Machine Learning", Cambridge: Springer, 2006. For recent research on properties of futures prices, I suggest mine: "The Wandering of Corn" https://arxiv.org/pdf/1704.01179.pdf 04/03/2017, and "Optimal Trading Strategies ..." https://arxiv.org/pdf/1312.2004.pdf 12/06/2013. Best Regards, Valerii


photo

 Dr. Debashis Dutta, Senior Manager, Financial Service Risk Management, Advisory at EY

 Sunday, August 13, 2017



Very interesting note Valerii!


photo

 Yanis Tazi, Data Scientist - Final Year Student at Mines ParisTech

 Wednesday, August 16, 2017



Great job!

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association