On asset-allocation and high-frequency data: are there financial gains from using different covariance estimators?

Abstract

We analyze the economic benefits of several covariance estimation approaches on a tactical asset-allocation problem in the presence of high-frequency return data. Our analysis confirms that the use of robust-to-noise and asynchronicity estimators not only gives statistically more accurate results, but the statistical efficiency is reflected into a financial benefit in most cases.

Keywords:

• Asset-allocation
• Non-parametric covariance estimators
• High-frequency data
• Financial performance measures