Portfolio Selection with Support Vector RegressionHenrique, Pedro Alexandre University of Brasilia, Brazil
• Machine Learning
• SVM & SVR• Stocks selection
WHY SVM?
• Multiple dimensions
Expand the information from the variables The importance of choosing KernelFrom Dr.Sead Sayad web site -Support Vector
Machine - Regression (SVR)
Chih-Chung Chang and Chih-Jen Lin, LIBSVM : a library for support vector machines, 2001.
• SVR – Support Vector Regression
APPLICATION• Test different 15
Kernels for portfolio selection to beat the market
The dual function:
Kernel ( Multi dimensional mapping)
Predict function.
Gaussian Radial Basis Kernel:
• Fundamentalist analysis• Feature Selection
• From 127 down to 24 features
• S&P 100 – from 06/30/2014
• Fundamental data from 06/29/1990 to 06/30/2014.
Training
52,5%
Validation
22,5%
Test
25%
Cross Validation
Random Selection
WORKFLOW
• Forecasting the quarterly return of the stocks for the Portfolio Selections.
• 15 portfolios - weighted by the forecast return
• Benchmark for the portfolios:• Equal weighted portfolios with the
100 stocks.
STRATEGY
• Machine per sector
• Other inputs
• Kernel combination
• SVM with risk management tools
RESEARCHERS IN PROGRESS
THANK YOU!
Packages:RobustbasePerformanceAnalyticsGgplot2robustbaseDplyrScalesKernlab
FselectorMlbenchForeachdoParalleldoSNOWrgl
Fan, A., & Palaniswami, M. (2001). Stock selection using support vector machines. Paper presented at the Neural Networks, 2001. Proceedings. IJCNN'01. International Joint Conference on.
Marcelino, S., Henrique, P. A., & Albuquerque, P. H. M. (2015). Portfolio selection with support vector machines in low economic perspectives in emerging markets. Economic Computation & Economic Cybernetics Studies & Research, 49(4).
Huerta, Ramon, Fernando Corbacho, and Charles Elkan. "Nonlinear support vector machines can systematically identify stocks with high and low future returns." Algorithmic Finance 2.1 (2013): 45-58.
Emir, S., Dinçer, H., & Timor, M. (2012). A Stock Selection Model Based on Fundamental and Technical Analysis Variables by Using Artificial Neural Networks and Support Vector Machines. Review of Economics & Finance, 106-122.
Pedro Alexandre M.B. Henrique.