Robust estimation of intraweek periodicity in volatility and jump detection

Opening, lunch and closing of financial markets induce a periodic component in the volatility of high-frequency returns. We propose a non-parametric weighted standard deviation and parametric truncated maximum likelihood estimation procedure for the periodic component in volatility and show that they are robust to price jumps. We also show that robust periodicity estimates can be used to increase the accuracy of jump detection methods. We compare the classical and robust methods for the 5-minute EUR/USD returns. The robust intraweek periodicity estimates are lower than the classical ones on Tuesday-Friday 8:30-8:35 EST and Monday-Friday 10:00-10:05 EST. The higher values for the non-robust estimates are likely to be due to jumps. Accounting for the periodicity in the volatility of high-frequency returns is especially important to detect the relatively small jumps occurring at times for which volatility is periodically low and to reduce the number of spurious jump detections at times of periodically high volatility.status: publishe