Jumps in High-Frequency Data: Spurious Detections, Dynamics, and News

Abstract

Applying tests for jumps to financial data sets can lead to an important number of spurious detections. Bursts of volatility are often incorrectly identified as jumps when the sampling is too sparse. At a higher frequency, methods robust to microstructure noise are required. We argue that whatever the jump detection test and the sampling frequency, a highly relevant number of spurious detections remain because of multiple testing issues. We propose a formal treatment based on an explicit thresholding on available test statistics. We prove that our method eliminates asymptotically all remaining spurious detections. In Dow Jones stocks between 2006 and 2008, spurious detections can represent up to 90% of the jumps detected initially. For the stocks considered, jumps are rare events, they do not cluster in time, and no cojump affects all stocks simultaneously, suggesting jump risk is diversifiable. We relate the remaining jumps to macroeconomic news, prescheduled company-specific announcements, and stories from news agencies which include a variety of unscheduled and uncategorized events. The vast majority of news do not cause jumps but may generate a market reaction of the form of bursts of volatility.

The local fractional bootstrap

“…We believe, however, that it is quite unlikely that jumps are influencing the analyses presented here to any meaningful degree. In fact, Christensen, Oomen, and Podolskij (2014) and Bajgrowicz, Scaillet, and Treccani (2016) have recently presented compelling evidence of the relative infrequency of jumps in asset prices, especially for highly liquid assets similar to the E-mini futures contract studied here.…”

Are Corn Futures Prices Getting “Jumpy”?

“…The performance of the jump test is measured by its ability to identify actual jumps and avoid type I statistical errors (i.e., reject the null when there is no jump). By performing the test repeatedly within each day, the number of jumps spuriously detected converges to the test significance level (Bajgrowicz, Scaillet, and Treccani ). For example, if a jump test is conducted with a 5% significance level over 200 intraday returns, on average, ten jumps will be erroneously identified.…”

“…Bajgrowicz, Scaillet, and Treccani () have characterized Lee and Mykland's approach (2008) as excessively conservative when applied to high frequency data.…”

“…In supplementary online appendix B, we provide the sensitivity of our results for values of W = 150 and 200 (which is approximately equivalent to using the data over the preceding 76 and 103 minutes), with the main results holding for the different window sizes. The robustness check conducted is limited by data availability within a day trading session.2 Bajgrowicz, Scaillet, and Treccani (2016) have characterized Lee and Mykland's approach (2008) as excessively conservative when applied to high frequency data.…”

Cluster Analysis of Jumps on Capital Markets

“…To je pravděpodobně důvod, proč v literatuře nedochází ke shodě ohledně optimality jednotlivých indikátorů (viz Dumitru a Urga, 2012). Na problém řádné identifi kace nedávno upozornilo několik studií, které hodnotí indiká-tory cenových skoků pomocí metody Monte Carlo (Theodosiou a Žikeš, 2011;Bajgrowicz a Scaillet, 2011;Dumitru a Urga, 2012;Hanousek et al, 2012;Vortelinos a Thomakos, 2013). I rozsáhlé studie metody Monte Carlo však stále nedokáží celý problém průkazně 1 Toto je nutné vnímat v kontextu stále se více integrujících trhů.…”