Title: Asymmetric autoencoders for factor-based covariance matrix estimation
Authors: Kevin Huynh - University of Basel (Switzerland) [presenting]
Gregor Lenhard - University of Basel (Switzerland)
Abstract: Estimating high dimensional covariance matrices for portfolio optimization is challenging because the number of parameters to be estimated grows quadratically in the number of assets. When the matrix dimension exceeds the sample size, the sample covariance matrix becomes singular. A possible solution is to impose a (latent) factor structure for the cross-section of asset returns as in the popular capital asset pricing model. Recent research suggests dimension reduction techniques to estimate the factors in a data-driven fashion. An asymmetric autoencoder neural network-based estimator is presented that incorporates the factor structure in its architecture and jointly estimates the factors and their loadings. The method is tested against well-established alternatives from the literature in an empirical experiment using stock returns of the past five decades. Results show that the proposed estimator is very competitive across different scenarios. The estimated loadings further reveal that the constructed factors are related to the stocks' sector classification.