CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: Testing the residual sparsity of a high-dimensional continuous-time factor model Authors:  Yuta Koike - University of Tokyo (Japan) [presenting]
Abstract: Investigation of the correlation structure of the residual process of a factor model is an important problem to assess systematic risk factors unexplained in the model. Such a problem is also important in high-dimensional covariance estimation because approximate factor models are often employed to reduce the curse of dimensionality, especially in the context of financial applications. A multiple testing procedure is developed to detect correlated pairs in the residual processes of a continuous-time factor model for multiple assets observed at a high-frequency in a high-dimensional setting such that the number of assets is possibly larger than the sample size.