Search
× Search
Friday, December 27, 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.

It is so easy to get high accuracy when you don’t setup your Machine Learning properly.

photo

 Honghai Yu, Scientist at Data Analytics Department, I2R

 Friday, April 22, 2016

It is so easy to get high accuracy when you don't setup your Machine Learning properly.


Print

3 comments on article "It is so easy to get high accuracy when you don't setup your Machine Learning properly. "

photo

 Deepak Sahijwala, Business Editor at Afternoon DC

 Wednesday, April 27, 2016



Superbly explained.


photo

 Stephane Hardy, Computational Finance Quant and Options Trader

 Wednesday, April 27, 2016



In physics, the fact that you observe a particle changes its state. This is more true in financial markets. Except that in financial markets, this makes them more predictable.


photo

 Graeme Smith, Investment Manager at Uncorrelated Alpha Management

 Thursday, April 28, 2016



This is a good article. Financial series are quite different than normal data series. For instance if stocks are randomly assigned to the in or out of sample group (training/testing), and a model is developed on the in sample (training) group, it will perform extraordinarily well on the out of sample (test) group. However, if an arbitrary date, say 1/1/2008 is chosen to separate the in and out of sample, a model based on pre-2008 will perform poorly post 2008. At any point in time stocks tend to be driven by the same factors, even when market movement is taken out as a factor. But across time those factors have "regime changes", meaning that what worked for most stocks at one point in time, will fail to work for any stock at another time

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